update train

main.py

@@ -45,7 +45,7 @@ def get_args_parser():
         nargs="+",
     )
     parser.add_argument("--lr_linear_proj_mult", default=0.1, type=float)
-    parser.add_argument("--batch_size", default=2, type=int)
+    parser.add_argument("--", default=2, type=int)
     parser.add_argument("--weight_decay", default=1e-4, type=float)
     parser.add_argument("--epochs", default=50, type=int)
     parser.add_argument("--lr_drop", default=40, type=int)

train_multi.sh

@@ -8,9 +8,10 @@ export USE_MIOPEN_BATCHNORM=1
 echo "Training start ..."
 
 # coco_path是训练数据集地址，数据是coco format
-GPUS_PER_NODE=4 ./tools/run_dist_launch.sh 4 <config path> --coco_path <coco path>
+# GPUS_PER_NODE=4 ./tools/run_dist_launch.sh 4 <config path> --coco_path <coco path>
 # example
-# GPUS_PER_NODE=4 ./tools/run_dist_launch.sh 4 \
-#       ./configs/two_stage/deformable-detr-hybrid-branch/12eps/r50_hybrid_branch_lambda1_group6_t1500_dp0_mqs_lft_deformable_detr_plus_iterative_bbox_refinement_plus_plus_two_stage.sh \
-
+GPUS_PER_NODE=8 ./tools/run_dist_launch.sh 8 \
+      ./configs/two_stage/deformable-detr-hybrid-branch/12eps/r50_hybrid_branch_lambda1_group6_t1500_dp0_mqs_lft_deformable_detr_plus_iterative_bbox_refinement_plus_plus_two_stage.sh > log_hdetr_2_8_dcu.20231205
+# 线上
+# GPUS_PER_NODE=4 ./tools/run_dist_slurm.sh wzhdexclu10 hdetr_2_8 8 ./configs/two_stage/deformable-detr-hybrid-branch/12eps/r50_hybrid_branch_lambda1_group6_t1500_dp0_mqs_lft_deformable_detr_plus_iterative_bbox_refinement_plus_plus_two_stage.sh
 echo "Training finished."