stylegan2_mmcv

configs/positional_encoding_in_gans/metafile.yml

+Collections:
+- Metadata:
+    Architecture:
+    - Positional Encoding in GANs
+  Name: Positional Encoding in GANs
+  Paper:
+  - https://openaccess.thecvf.com/content/CVPR2021/html/Xu_Positional_Encoding_As_Spatial_Inductive_Bias_in_GANs_CVPR_2021_paper.html
+  README: configs/positional_encoding_in_gans/README.md
+Models:
+- Config: https://github.com/open-mmlab/mmgeneration/tree/master/configs/positional_encoding_in_gans/stylegan2_c2_ffhq_256_b3x8_1100k.py
+  In Collection: Positional Encoding in GANs
+  Metadata:
+    Training Data: FFHQ
+  Name: stylegan2_c2_ffhq_256_b3x8_1100k
+  Results:
+  - Dataset: FFHQ
+    Metrics:
+      FID50k: 5.56
+      P&R10k: 75.92/51.24
+      Reference in Paper: Tab.5 config-a
+      Scales: 256.0
+    Task: Unconditional GANs
+  Weights: https://download.openmmlab.com/mmgen/pe_in_gans/stylegan2_c2_config-a_ffhq_256x256_b3x8_1100k_20210406_145127-71d9634b.pth
+- Config: https://github.com/open-mmlab/mmgeneration/tree/master/configs/positional_encoding_in_gans/stylegan2_c2_ffhq_512_b3x8_1100k.py
+  In Collection: Positional Encoding in GANs
+  Metadata:
+    Training Data: FFHQ
+  Name: stylegan2_c2_ffhq_512_b3x8_1100k
+  Results:
+  - Dataset: FFHQ
+    Metrics:
+      FID50k: 4.91
+      P&R10k: 75.65/54.58
+      Reference in Paper: Tab.5 config-b
+      Scales: 512.0
+    Task: Unconditional GANs
+  Weights: https://download.openmmlab.com/mmgen/pe_in_gans/stylegan2_c2_config-b_ffhq_512x512_b3x8_1100k_20210406_145142-e85e5cf4.pth
+- Config: https://github.com/open-mmlab/mmgeneration/tree/master/configs/positional_encoding_in_gans/mspie-stylegan2_c2_config-c_ffhq_256-512_b3x8_1100k.py
+  In Collection: Positional Encoding in GANs
+  Metadata:
+    Training Data: FFHQ
+  Name: mspie-stylegan2_c2_config-c_ffhq_256-512_b3x8_1100k
+  Results:
+  - Dataset: FFHQ
+    Metrics:
+      FID50k: 3.35
+      P&R10k: 73.84/55.77
+      Reference in Paper: Tab.5 config-c
+      Scales: 256, 384, 512
+    Task: Unconditional GANs
+  Weights: https://download.openmmlab.com/mmgen/pe_in_gans/mspie-stylegan2_c2_config-c_ffhq_256-512_b3x8_1100k_20210406_144824-9f43b07d.pth
+- Config: https://github.com/open-mmlab/mmgeneration/tree/master/configs/positional_encoding_in_gans/mspie-stylegan2_c2_config-d_ffhq_256-512_b3x8_1100k.py
+  In Collection: Positional Encoding in GANs
+  Metadata:
+    Training Data: FFHQ
+  Name: mspie-stylegan2_c2_config-d_ffhq_256-512_b3x8_1100k
+  Results:
+  - Dataset: FFHQ
+    Metrics:
+      FID50k: 3.5
+      P&R10k: 73.28/56.16
+      Reference in Paper: Tab.5 config-d
+      Scales: 256, 384, 512
+    Task: Unconditional GANs
+  Weights: https://download.openmmlab.com/mmgen/pe_in_gans/mspie-stylegan2_c2_config-d_ffhq_256-512_b3x8_1100k_20210406_144840-dbefacf6.pth
+- Config: https://github.com/open-mmlab/mmgeneration/tree/master/configs/positional_encoding_in_gans/mspie-stylegan2_c2_config-e_ffhq_256-512_b3x8_1100k.py
+  In Collection: Positional Encoding in GANs
+  Metadata:
+    Training Data: FFHQ
+  Name: mspie-stylegan2_c2_config-e_ffhq_256-512_b3x8_1100k
+  Results:
+  - Dataset: FFHQ
+    Metrics:
+      FID50k: 3.15
+      P&R10k: 74.13/56.88
+      Reference in Paper: Tab.5 config-e
+      Scales: 256, 384, 512
+    Task: Unconditional GANs
+  Weights: https://download.openmmlab.com/mmgen/pe_in_gans/mspie-stylegan2_c2_config-e_ffhq_256-512_b3x8_1100k_20210406_144906-98d5a42a.pth
+- Config: https://github.com/open-mmlab/mmgeneration/tree/master/configs/positional_encoding_in_gans/mspie-stylegan2_c2_config-f_ffhq_256-512_b3x8_1100k.py
+  In Collection: Positional Encoding in GANs
+  Metadata:
+    Training Data: FFHQ
+  Name: mspie-stylegan2_c2_config-f_ffhq_256-512_b3x8_1100k
+  Results:
+  - Dataset: FFHQ
+    Metrics:
+      FID50k: 2.93
+      P&R10k: 73.51/57.32
+      Reference in Paper: Tab.5 config-f
+      Scales: 256, 384, 512
+    Task: Unconditional GANs
+  Weights: https://download.openmmlab.com/mmgen/pe_in_gans/mspie-stylegan2_c2_config-f_ffhq_256-512_b3x8_1100k_20210406_144927-4f4d5391.pth
+- Config: https://github.com/open-mmlab/mmgeneration/tree/master/configs/positional_encoding_in_gans/mspie-stylegan2_c1_config-g_ffhq_256-512_b3x8_1100k.py
+  In Collection: Positional Encoding in GANs
+  Metadata:
+    Training Data: FFHQ
+  Name: mspie-stylegan2_c1_config-g_ffhq_256-512_b3x8_1100k
+  Results:
+  - Dataset: FFHQ
+    Metrics:
+      FID50k: 3.4
+      P&R10k: 73.05/56.45
+      Reference in Paper: Tab.5 config-g
+      Scales: 256, 384, 512
+    Task: Unconditional GANs
+  Weights: https://download.openmmlab.com/mmgen/pe_in_gans/mspie-stylegan2_c1_config-g_ffhq_256-512_b3x8_1100k_20210406_144758-2df61752.pth
+- Config: https://github.com/open-mmlab/mmgeneration/tree/master/configs/positional_encoding_in_gans/mspie-stylegan2_c2_config-h_ffhq_256-512_b3x8_1100k.py
+  In Collection: Positional Encoding in GANs
+  Metadata:
+    Training Data: FFHQ
+  Name: mspie-stylegan2_c2_config-h_ffhq_256-512_b3x8_1100k
+  Results:
+  - Dataset: FFHQ
+    Metrics:
+      FID50k: 4.01
+      P&R10k: 72.81/54.35
+      Reference in Paper: Tab.5 config-h
+      Scales: 256, 384, 512
+    Task: Unconditional GANs
+  Weights: https://download.openmmlab.com/mmgen/pe_in_gans/mspie-stylegan2_c2_config-h_ffhq_256-512_b3x8_1100k_20210406_145006-84cf3f48.pth
+- Config: https://github.com/open-mmlab/mmgeneration/tree/master/configs/positional_encoding_in_gans/mspie-stylegan2_c2_config-i_ffhq_256-512_b3x8_1100k.py
+  In Collection: Positional Encoding in GANs
+  Metadata:
+    Training Data: FFHQ
+  Name: mspie-stylegan2_c2_config-i_ffhq_256-512_b3x8_1100k
+  Results:
+  - Dataset: FFHQ
+    Metrics:
+      FID50k: 3.76
+      P&R10k: 73.26/54.71
+      Reference in Paper: Tab.5 config-i
+      Scales: 256, 384, 512
+    Task: Unconditional GANs
+  Weights: https://download.openmmlab.com/mmgen/pe_in_gans/mspie-stylegan2_c2_config-i_ffhq_256-512_b3x8_1100k_20210406_145023-c2b0accf.pth
+- Config: https://github.com/open-mmlab/mmgeneration/tree/master/configs/positional_encoding_in_gans/mspie-stylegan2_c2_config-j_ffhq_256-512_b3x8_1100k.py
+  In Collection: Positional Encoding in GANs
+  Metadata:
+    Training Data: FFHQ
+  Name: mspie-stylegan2_c2_config-j_ffhq_256-512_b3x8_1100k
+  Results:
+  - Dataset: FFHQ
+    Metrics:
+      FID50k: 4.23
+      P&R10k: 73.11/54.63
+      Reference in Paper: Tab.5 config-j
+      Scales: 256, 384, 512
+    Task: Unconditional GANs
+  Weights: https://download.openmmlab.com/mmgen/pe_in_gans/mspie-stylegan2_c2_config-j_ffhq_256-512_b3x8_1100k_20210406_145044-c407481b.pth
+- Config: https://github.com/open-mmlab/mmgeneration/tree/master/configs/positional_encoding_in_gans/mspie-stylegan2_c2_config-k_ffhq_256-512_b3x8_1100k.py
+  In Collection: Positional Encoding in GANs
+  Metadata:
+    Training Data: FFHQ
+  Name: mspie-stylegan2_c2_config-k_ffhq_256-512_b3x8_1100k
+  Results:
+  - Dataset: FFHQ
+    Metrics:
+      FID50k: 4.17
+      P&R10k: 73.05/51.07
+      Reference in Paper: Tab.5 config-k
+      Scales: 256, 384, 512
+    Task: Unconditional GANs
+  Weights: https://download.openmmlab.com/mmgen/pe_in_gans/mspie-stylegan2_c2_config-k_ffhq_256-512_b3x8_1100k_20210406_145105-6d8cc39f.pth
+- Config: https://github.com/open-mmlab/mmgeneration/tree/master/configs/positional_encoding_in_gans/mspie-stylegan2_c2_config-f_ffhq_256-896_b3x8_1100k.py
+  In Collection: Positional Encoding in GANs
+  Metadata:
+    Training Data: FFHQ
+  Name: mspie-stylegan2_c2_config-f_ffhq_256-896_b3x8_1100k
+  Results:
+  - Dataset: FFHQ
+    Metrics:
+      FID50k: 4.1
+      P&R10k: 72.21/50.29
+      Reference in Paper: higher-resolution
+      Scales: 256, 512, 896
+    Task: Unconditional GANs
+  Weights: https://download.openmmlab.com/mmgen/pe_in_gans/mspie-stylegan2_c2_config-f_ffhq_256-896_b3x8_1100k_20210406_144943-6c18ad5d.pth
+- Config: https://github.com/open-mmlab/mmgeneration/tree/master/configs/positional_encoding_in_gans/mspie-stylegan2_c1_config-f_ffhq_256-1024_b2x8_1600k.py
+  In Collection: Positional Encoding in GANs
+  Metadata:
+    Training Data: FFHQ
+  Name: mspie-stylegan2_c1_config-f_ffhq_256-1024_b2x8_1600k
+  Results:
+  - Dataset: FFHQ
+    Metrics:
+      FID50k: 6.24
+      P&R10k: 71.79/49.92
+      Reference in Paper: higher-resolution
+      Scales: 256, 512, 1024
+    Task: Unconditional GANs
+  Weights: https://download.openmmlab.com/mmgen/pe_in_gans/mspie-stylegan2_c1_config-f_ffhq_256-1024_b2x8_1600k_20210406_144716-81cbdc96.pth
+- Config: https://github.com/open-mmlab/mmgeneration/tree/master/configs/positional_encoding_in_gans/singan_interp-pad_balloons.py
+  In Collection: Positional Encoding in GANs
+  Metadata:
+    Training Data: Others
+  Name: singan_interp-pad_balloons
+  Results:
+  - Dataset: Others
+    Metrics:
+      Num Scales: 8.0
+    Task: Unconditional GANs
+  Weights: https://download.openmmlab.com/mmgen/pe_in_gans/singan_interp-pad_balloons_20210406_180014-96f51555.pth
+- Config: https://github.com/open-mmlab/mmgeneration/tree/master/configs/positional_encoding_in_gans/singan_interp-pad_disc-nobn_balloons.py
+  In Collection: Positional Encoding in GANs
+  Metadata:
+    Training Data: Others
+  Name: singan_interp-pad_disc-nobn_balloons
+  Results:
+  - Dataset: Others
+    Metrics:
+      Num Scales: 8.0
+    Task: Unconditional GANs
+  Weights: https://download.openmmlab.com/mmgen/pe_in_gans/singan_interp-pad_disc-nobn_balloons_20210406_180059-7d63e65d.pth
+- Config: https://github.com/open-mmlab/mmgeneration/tree/master/configs/positional_encoding_in_gans/singan_interp-pad_disc-nobn_fish.py
+  In Collection: Positional Encoding in GANs
+  Metadata:
+    Training Data: Others
+  Name: singan_interp-pad_disc-nobn_fish
+  Results:
+  - Dataset: Others
+    Metrics:
+      Num Scales: 10.0
+    Task: Unconditional GANs
+  Weights: https://download.openmmlab.com/mmgen/pe_in_gans/singan_interp-pad_disc-nobn_fis_20210406_175720-9428517a.pth
+- Config: https://github.com/open-mmlab/mmgeneration/tree/master/configs/positional_encoding_in_gans/singan_csg_fish.py
+  In Collection: Positional Encoding in GANs
+  Metadata:
+    Training Data: Others
+  Name: singan_csg_fish
+  Results:
+  - Dataset: Others
+    Metrics:
+      Num Scales: 10.0
+    Task: Unconditional GANs
+  Weights: https://download.openmmlab.com/mmgen/pe_in_gans/singan_csg_fis_20210406_175532-f0ec7b61.pth
+- Config: https://github.com/open-mmlab/mmgeneration/tree/master/configs/positional_encoding_in_gans/singan_csg_bohemian.py
+  In Collection: Positional Encoding in GANs
+  Metadata:
+    Training Data: Others
+  Name: singan_csg_bohemian
+  Results:
+  - Dataset: Others
+    Metrics:
+      Num Scales: 10.0
+    Task: Unconditional GANs
+  Weights: https://download.openmmlab.com/mmgen/pe_in_gans/singan_csg_bohemian_20210407_195455-5ed56db2.pth
+- Config: https://github.com/open-mmlab/mmgeneration/tree/master/configs/positional_encoding_in_gans/singan_spe-dim4_fish.py
+  In Collection: Positional Encoding in GANs
+  Metadata:
+    Training Data: Others
+  Name: singan_spe-dim4_fish
+  Results:
+  - Dataset: Others
+    Metrics:
+      Num Scales: 10.0
+    Task: Unconditional GANs
+  Weights: https://download.openmmlab.com/mmgen/pe_in_gans/singan_spe-dim4_fish_20210406_175933-f483a7e3.pth
+- Config: https://github.com/open-mmlab/mmgeneration/tree/master/configs/positional_encoding_in_gans/singan_spe-dim4_bohemian.py
+  In Collection: Positional Encoding in GANs
+  Metadata:
+    Training Data: Others
+  Name: singan_spe-dim4_bohemian
+  Results:
+  - Dataset: Others
+    Metrics:
+      Num Scales: 10.0
+    Task: Unconditional GANs
+  Weights: https://download.openmmlab.com/mmgen/pe_in_gans/singan_spe-dim4_bohemian_20210406_175820-6e484a35.pth
+- Config: https://github.com/open-mmlab/mmgeneration/tree/master/configs/positional_encoding_in_gans/singan_spe-dim8_bohemian.py
+  In Collection: Positional Encoding in GANs
+  Metadata:
+    Training Data: Others
+  Name: singan_spe-dim8_bohemian
+  Results:
+  - Dataset: Others
+    Metrics:
+      Num Scales: 10.0
+    Task: Unconditional GANs
+  Weights: https://download.openmmlab.com/mmgen/pe_in_gans/singan_spe-dim8_bohemian_20210406_175858-7faa50f3.pth

configs/positional_encoding_in_gans/mspie-stylegan2_c1_config-f_ffhq_256-1024_b2x8_1600k.py

+_base_ = [
+    '../_base_/datasets/ffhq_flip.py',
+    '../_base_/models/stylegan/stylegan2_base.py',
+    '../_base_/default_runtime.py'
+]
+
+model = dict(
+    type='MSPIEStyleGAN2',
+    generator=dict(
+        type='MSStyleGANv2Generator',
+        head_pos_encoding=dict(
+            type='SPE',
+            embedding_dim=256,
+            padding_idx=0,
+            init_size=256,
+            center_shift=100),
+        deconv2conv=True,
+        up_after_conv=True,
+        up_config=dict(scale_factor=2, mode='bilinear', align_corners=True),
+        out_size=256,
+        channel_multiplier=1),
+    discriminator=dict(
+        type='MSStyleGAN2Discriminator', in_size=256, with_adaptive_pool=True))
+
+train_cfg = dict(
+    num_upblocks=6,
+    multi_input_scales=[0, 4, 12],
+    multi_scale_probability=[0.5, 0.25, 0.25])
+
+data = dict(
+    samples_per_gpu=2,
+    train=dict(dataset=dict(
+        imgs_root='./data/ffhq/images')))  # path for 1024 scales
+
+ema_half_life = 10.
+custom_hooks = [
+    dict(
+        type='VisualizeUnconditionalSamples',
+        output_dir='training_samples',
+        interval=5000),
+    dict(
+        type='ExponentialMovingAverageHook',
+        module_keys=('generator_ema', ),
+        interval=1,
+        interp_cfg=dict(momentum=0.5**(32. / (ema_half_life * 1000.))),
+        priority='VERY_HIGH')
+]
+
+checkpoint_config = dict(interval=10000, by_epoch=False, max_keep_ckpts=40)
+lr_config = None
+
+log_config = dict(
+    interval=100,
+    hooks=[
+        dict(type='TextLoggerHook'),
+        # dict(type='TensorboardLoggerHook'),
+    ])
+
+cudnn_benchmark = False
+total_iters = 1500002
+
+metrics = dict(
+    fid50k=dict(
+        type='FID',
+        num_images=50000,
+        inception_pkl='work_dirs/inception_pkl/ffhq-256-50k-rgb.pkl',
+        bgr2rgb=True),
+    pr10k3=dict(type='PR', num_images=10000, k=3))

configs/positional_encoding_in_gans/mspie-stylegan2_c1_config-g_ffhq_256-512_b3x8_1100k.py

+_base_ = [
+    '../_base_/datasets/ffhq_flip.py',
+    '../_base_/models/stylegan/stylegan2_base.py',
+    '../_base_/default_runtime.py'
+]
+
+model = dict(
+    type='MSPIEStyleGAN2',
+    generator=dict(
+        type='MSStyleGANv2Generator',
+        head_pos_encoding=dict(
+            type='SPE',
+            embedding_dim=256,
+            padding_idx=0,
+            init_size=256,
+            center_shift=100),
+        deconv2conv=True,
+        up_after_conv=True,
+        up_config=dict(scale_factor=2, mode='bilinear', align_corners=True),
+        out_size=256,
+        channel_multiplier=1),
+    discriminator=dict(
+        type='MSStyleGAN2Discriminator',
+        in_size=256,
+        with_adaptive_pool=True,
+        channel_multiplier=1))
+
+train_cfg = dict(
+    num_upblocks=6,
+    multi_input_scales=[0, 2, 4],
+    multi_scale_probability=[0.5, 0.25, 0.25])
+
+data = dict(
+    samples_per_gpu=3,
+    train=dict(dataset=dict(imgs_root='./data/ffhq/ffhq_imgs/ffhq_512')))
+
+ema_half_life = 10.
+custom_hooks = [
+    dict(
+        type='VisualizeUnconditionalSamples',
+        output_dir='training_samples',
+        interval=5000),
+    dict(
+        type='ExponentialMovingAverageHook',
+        module_keys=('generator_ema', ),
+        interval=1,
+        interp_cfg=dict(momentum=0.5**(32. / (ema_half_life * 1000.))),
+        priority='VERY_HIGH')
+]
+
+checkpoint_config = dict(interval=10000, by_epoch=False, max_keep_ckpts=40)
+lr_config = None
+
+log_config = dict(
+    interval=100,
+    hooks=[
+        dict(type='TextLoggerHook'),
+        # dict(type='TensorboardLoggerHook'),
+    ])
+
+cudnn_benchmark = False
+total_iters = 1100002
+
+metrics = dict(
+    fid50k=dict(
+        type='FID',
+        num_images=50000,
+        inception_pkl='work_dirs/inception_pkl/ffhq-256-50k-rgb.pkl',
+        bgr2rgb=True),
+    pr10k3=dict(type='PR', num_images=10000, k=3))

configs/positional_encoding_in_gans/mspie-stylegan2_c2_config-c_ffhq_256-512_b3x8_1100k.py

+_base_ = [
+    '../_base_/datasets/ffhq_flip.py',
+    '../_base_/models/stylegan/stylegan2_base.py',
+    '../_base_/default_runtime.py'
+]
+
+model = dict(
+    type='MSPIEStyleGAN2',
+    generator=dict(
+        type='MSStyleGANv2Generator',
+        head_pos_encoding=None,
+        deconv2conv=True,
+        up_after_conv=True,
+        head_pos_size=(4, 4),
+        interp_head=True,
+        up_config=dict(scale_factor=2, mode='bilinear', align_corners=True),
+        out_size=256),
+    discriminator=dict(
+        type='MSStyleGAN2Discriminator', in_size=256, with_adaptive_pool=True))
+
+train_cfg = dict(
+    num_upblocks=6,
+    multi_input_scales=[0, 2, 4],
+    multi_scale_probability=[0.5, 0.25, 0.25])
+
+data = dict(
+    samples_per_gpu=3,
+    train=dict(dataset=dict(imgs_root='./data/ffhq/ffhq_imgs/ffhq_512')))
+
+ema_half_life = 10.
+custom_hooks = [
+    dict(
+        type='VisualizeUnconditionalSamples',
+        output_dir='training_samples',
+        interval=5000),
+    dict(
+        type='ExponentialMovingAverageHook',
+        module_keys=('generator_ema', ),
+        interval=1,
+        interp_cfg=dict(momentum=0.5**(32. / (ema_half_life * 1000.))),
+        priority='VERY_HIGH')
+]
+
+checkpoint_config = dict(interval=10000, by_epoch=False, max_keep_ckpts=40)
+lr_config = None
+
+log_config = dict(
+    interval=100,
+    hooks=[
+        dict(type='TextLoggerHook'),
+        # dict(type='TensorboardLoggerHook'),
+    ])
+
+cudnn_benchmark = False
+total_iters = 1100002
+
+metrics = dict(
+    fid50k=dict(
+        type='FID',
+        num_images=50000,
+        inception_pkl='work_dirs/inception_pkl/ffhq-256-50k-rgb.pkl',
+        bgr2rgb=True),
+    pr10k3=dict(type='PR', num_images=10000, k=3))

configs/positional_encoding_in_gans/mspie-stylegan2_c2_config-d_ffhq_256-512_b3x8_1100k.py

+_base_ = [
+    '../_base_/datasets/ffhq_flip.py',
+    '../_base_/models/stylegan/stylegan2_base.py',
+    '../_base_/default_runtime.py'
+]
+
+model = dict(
+    type='MSPIEStyleGAN2',
+    generator=dict(
+        type='MSStyleGANv2Generator',
+        head_pos_encoding=dict(type='CSG'),
+        deconv2conv=True,
+        up_after_conv=True,
+        head_pos_size=(4, 4),
+        up_config=dict(scale_factor=2, mode='bilinear', align_corners=True),
+        out_size=256),
+    discriminator=dict(
+        type='MSStyleGAN2Discriminator', in_size=256, with_adaptive_pool=True))
+
+train_cfg = dict(
+    num_upblocks=6,
+    multi_input_scales=[0, 2, 4],
+    multi_scale_probability=[0.5, 0.25, 0.25])
+
+data = dict(
+    samples_per_gpu=3,
+    train=dict(dataset=dict(imgs_root='./data/ffhq/ffhq_imgs/ffhq_512')))
+
+ema_half_life = 10.
+custom_hooks = [
+    dict(
+        type='VisualizeUnconditionalSamples',
+        output_dir='training_samples',
+        interval=5000),
+    dict(
+        type='ExponentialMovingAverageHook',
+        module_keys=('generator_ema', ),
+        interval=1,
+        interp_cfg=dict(momentum=0.5**(32. / (ema_half_life * 1000.))),
+        priority='VERY_HIGH')
+]
+
+checkpoint_config = dict(interval=10000, by_epoch=False, max_keep_ckpts=40)
+lr_config = None
+
+log_config = dict(
+    interval=100,
+    hooks=[
+        dict(type='TextLoggerHook'),
+        # dict(type='TensorboardLoggerHook'),
+    ])
+
+cudnn_benchmark = False
+total_iters = 1100002
+
+metrics = dict(
+    fid50k=dict(
+        type='FID',
+        num_images=50000,
+        inception_pkl='work_dirs/inception_pkl/ffhq-256-50k-rgb.pkl',
+        bgr2rgb=True),
+    pr10k3=dict(type='PR', num_images=10000, k=3))

configs/positional_encoding_in_gans/mspie-stylegan2_c2_config-e_ffhq_256-512_b3x8_1100k.py

+_base_ = [
+    '../_base_/datasets/ffhq_flip.py',
+    '../_base_/models/stylegan/stylegan2_base.py',
+    '../_base_/default_runtime.py'
+]
+
+model = dict(
+    type='MSPIEStyleGAN2',
+    generator=dict(
+        type='MSStyleGANv2Generator',
+        head_pos_encoding=dict(
+            type='SPE',
+            embedding_dim=256,
+            padding_idx=0,
+            init_size=256,
+            center_shift=100),
+        deconv2conv=True,
+        up_after_conv=True,
+        head_pos_size=(4, 4),
+        interp_head=True,
+        up_config=dict(scale_factor=2, mode='bilinear', align_corners=True),
+        out_size=256),
+    discriminator=dict(
+        type='MSStyleGAN2Discriminator', in_size=256, with_adaptive_pool=True))
+
+train_cfg = dict(
+    num_upblocks=6,
+    multi_input_scales=[0, 2, 4],
+    multi_scale_probability=[0.5, 0.25, 0.25])
+
+data = dict(
+    samples_per_gpu=3,
+    train=dict(dataset=dict(imgs_root='./data/ffhq/ffhq_imgs/ffhq_512')))
+
+ema_half_life = 10.
+custom_hooks = [
+    dict(
+        type='VisualizeUnconditionalSamples',
+        output_dir='training_samples',
+        interval=5000),
+    dict(
+        type='ExponentialMovingAverageHook',
+        module_keys=('generator_ema', ),
+        interval=1,
+        interp_cfg=dict(momentum=0.5**(32. / (ema_half_life * 1000.))),
+        priority='VERY_HIGH')
+]
+
+checkpoint_config = dict(interval=10000, by_epoch=False, max_keep_ckpts=40)
+lr_config = None
+
+log_config = dict(
+    interval=100,
+    hooks=[
+        dict(type='TextLoggerHook'),
+        # dict(type='TensorboardLoggerHook'),
+    ])
+
+cudnn_benchmark = False
+total_iters = 1100002
+
+metrics = dict(
+    fid50k=dict(
+        type='FID',
+        num_images=50000,
+        inception_pkl='work_dirs/inception_pkl/ffhq-256-50k-rgb.pkl',
+        bgr2rgb=True),
+    pr10k3=dict(type='PR', num_images=10000, k=3))

configs/positional_encoding_in_gans/mspie-stylegan2_c2_config-f_ffhq_256-512_b3x8_1100k.py

+_base_ = [
+    '../_base_/datasets/ffhq_flip.py',
+    '../_base_/models/stylegan/stylegan2_base.py',
+    '../_base_/default_runtime.py'
+]
+
+model = dict(
+    type='MSPIEStyleGAN2',
+    generator=dict(
+        type='MSStyleGANv2Generator',
+        head_pos_encoding=dict(
+            type='SPE',
+            embedding_dim=256,
+            padding_idx=0,
+            init_size=256,
+            center_shift=100),
+        deconv2conv=True,
+        up_after_conv=True,
+        up_config=dict(scale_factor=2, mode='bilinear', align_corners=True),
+        out_size=256),
+    discriminator=dict(
+        type='MSStyleGAN2Discriminator', in_size=256, with_adaptive_pool=True))
+
+train_cfg = dict(
+    num_upblocks=6,
+    multi_input_scales=[0, 2, 4],
+    multi_scale_probability=[0.5, 0.25, 0.25])
+
+data = dict(
+    samples_per_gpu=3,
+    train=dict(dataset=dict(imgs_root='./data/ffhq/ffhq_imgs/ffhq_512')))
+
+ema_half_life = 10.
+custom_hooks = [
+    dict(
+        type='VisualizeUnconditionalSamples',
+        output_dir='training_samples',
+        interval=5000),
+    dict(
+        type='ExponentialMovingAverageHook',
+        module_keys=('generator_ema', ),
+        interval=1,
+        interp_cfg=dict(momentum=0.5**(32. / (ema_half_life * 1000.))),
+        priority='VERY_HIGH')
+]
+
+checkpoint_config = dict(interval=10000, by_epoch=False, max_keep_ckpts=40)
+lr_config = None
+
+log_config = dict(
+    interval=100,
+    hooks=[
+        dict(type='TextLoggerHook'),
+        # dict(type='TensorboardLoggerHook'),
+    ])
+
+cudnn_benchmark = False
+total_iters = 1100002
+
+metrics = dict(
+    fid50k=dict(
+        type='FID',
+        num_images=50000,
+        inception_pkl='work_dirs/inception_pkl/ffhq-256-50k-rgb.pkl',
+        bgr2rgb=True),
+    pr10k3=dict(type='PR', num_images=10000, k=3))

configs/positional_encoding_in_gans/mspie-stylegan2_c2_config-f_ffhq_256-896_b3x8_1100k.py

+_base_ = [
+    '../_base_/datasets/ffhq_flip.py',
+    '../_base_/models/stylegan/stylegan2_base.py',
+    '../_base_/default_runtime.py'
+]
+
+model = dict(
+    type='MSPIEStyleGAN2',
+    generator=dict(
+        type='MSStyleGANv2Generator',
+        head_pos_encoding=dict(
+            type='SPE',
+            embedding_dim=256,
+            padding_idx=0,
+            init_size=256,
+            center_shift=100),
+        deconv2conv=True,
+        up_after_conv=True,
+        up_config=dict(scale_factor=2, mode='bilinear', align_corners=True),
+        out_size=256),
+    discriminator=dict(
+        type='MSStyleGAN2Discriminator', in_size=256, with_adaptive_pool=True))
+
+train_cfg = dict(
+    num_upblocks=6,
+    multi_input_scales=[0, 4, 10],
+    multi_scale_probability=[0.5, 0.25, 0.25])
+
+data = dict(
+    samples_per_gpu=3,
+    train=dict(dataset=dict(
+        imgs_root='./data/ffhq/images')))  # path for 1024 scales
+
+ema_half_life = 10.
+custom_hooks = [
+    dict(
+        type='VisualizeUnconditionalSamples',
+        output_dir='training_samples',
+        interval=5000),
+    dict(
+        type='ExponentialMovingAverageHook',
+        module_keys=('generator_ema', ),
+        interval=1,
+        interp_cfg=dict(momentum=0.5**(32. / (ema_half_life * 1000.))),
+        priority='VERY_HIGH')
+]
+
+checkpoint_config = dict(interval=10000, by_epoch=False, max_keep_ckpts=40)
+lr_config = None
+
+log_config = dict(
+    interval=100,
+    hooks=[
+        dict(type='TextLoggerHook'),
+        # dict(type='TensorboardLoggerHook'),
+    ])
+
+cudnn_benchmark = False
+total_iters = 1100002
+
+metrics = dict(
+    fid50k=dict(
+        type='FID',
+        num_images=50000,
+        inception_pkl='work_dirs/inception_pkl/ffhq-256-50k-rgb.pkl',
+        bgr2rgb=True),
+    pr10k3=dict(type='PR', num_images=10000, k=3))

configs/positional_encoding_in_gans/mspie-stylegan2_c2_config-h_ffhq_256-512_b3x8_1100k.py

+_base_ = [
+    '../_base_/datasets/ffhq_flip.py',
+    '../_base_/models/stylegan/stylegan2_base.py',
+    '../_base_/default_runtime.py'
+]
+
+model = dict(
+    type='MSPIEStyleGAN2',
+    generator=dict(
+        type='MSStyleGANv2Generator',
+        head_pos_encoding=None,
+        deconv2conv=True,
+        up_after_conv=False,
+        interp_pad=4,
+        no_pad=True,
+        head_pos_size=(6, 6),
+        interp_head=True,
+        up_config=dict(scale_factor=2, mode='bilinear', align_corners=True),
+        out_size=256),
+    discriminator=dict(
+        type='MSStyleGAN2Discriminator', in_size=256, with_adaptive_pool=True))
+
+train_cfg = dict(
+    num_upblocks=6,
+    multi_input_scales=[0, 2, 4],
+    multi_scale_probability=[0.5, 0.25, 0.25])
+
+data = dict(
+    samples_per_gpu=3,
+    train=dict(dataset=dict(imgs_root='./data/ffhq/ffhq_imgs/ffhq_512')))
+
+ema_half_life = 10.
+custom_hooks = [
+    dict(
+        type='VisualizeUnconditionalSamples',
+        output_dir='training_samples',
+        interval=5000),
+    dict(
+        type='ExponentialMovingAverageHook',
+        module_keys=('generator_ema', ),
+        interval=1,
+        interp_cfg=dict(momentum=0.5**(32. / (ema_half_life * 1000.))),
+        priority='VERY_HIGH')
+]
+
+checkpoint_config = dict(interval=10000, by_epoch=False, max_keep_ckpts=40)
+lr_config = None
+
+log_config = dict(
+    interval=100,
+    hooks=[
+        dict(type='TextLoggerHook'),
+        # dict(type='TensorboardLoggerHook'),
+    ])
+
+cudnn_benchmark = False
+total_iters = 1100002
+
+metrics = dict(
+    fid50k=dict(
+        type='FID',
+        num_images=50000,
+        inception_pkl='work_dirs/inception_pkl/ffhq-256-50k-rgb.pkl',
+        bgr2rgb=True),
+    pr10k3=dict(type='PR', num_images=10000, k=3))

configs/positional_encoding_in_gans/mspie-stylegan2_c2_config-i_ffhq_256-512_b3x8_1100k.py

+_base_ = [
+    '../_base_/datasets/ffhq_flip.py',
+    '../_base_/models/stylegan/stylegan2_base.py',
+    '../_base_/default_runtime.py'
+]
+
+model = dict(
+    type='MSPIEStyleGAN2',
+    generator=dict(
+        type='MSStyleGANv2Generator',
+        head_pos_encoding=dict(type='CSG'),
+        deconv2conv=True,
+        up_after_conv=False,
+        interp_pad=4,
+        no_pad=True,
+        head_pos_size=(6, 6),
+        up_config=dict(scale_factor=2, mode='bilinear', align_corners=True),
+        out_size=256),
+    discriminator=dict(
+        type='MSStyleGAN2Discriminator', in_size=256, with_adaptive_pool=True))
+
+train_cfg = dict(
+    num_upblocks=6,
+    multi_input_scales=[0, 2, 4],
+    multi_scale_probability=[0.5, 0.25, 0.25])
+
+data = dict(
+    samples_per_gpu=3,
+    train=dict(dataset=dict(imgs_root='./data/ffhq/ffhq_imgs/ffhq_512')))
+
+ema_half_life = 10.
+custom_hooks = [
+    dict(
+        type='VisualizeUnconditionalSamples',
+        output_dir='training_samples',
+        interval=5000),
+    dict(
+        type='ExponentialMovingAverageHook',
+        module_keys=('generator_ema', ),
+        interval=1,
+        interp_cfg=dict(momentum=0.5**(32. / (ema_half_life * 1000.))),
+        priority='VERY_HIGH')
+]
+
+checkpoint_config = dict(interval=10000, by_epoch=False, max_keep_ckpts=40)
+lr_config = None
+
+log_config = dict(
+    interval=100,
+    hooks=[
+        dict(type='TextLoggerHook'),
+        # dict(type='TensorboardLoggerHook'),
+    ])
+
+cudnn_benchmark = False
+total_iters = 1100002
+
+metrics = dict(
+    fid50k=dict(
+        type='FID',
+        num_images=50000,
+        inception_pkl='work_dirs/inception_pkl/ffhq-256-50k-rgb.pkl',
+        bgr2rgb=True),
+    pr10k3=dict(type='PR', num_images=10000, k=3))

configs/positional_encoding_in_gans/mspie-stylegan2_c2_config-j_ffhq_256-512_b3x8_1100k.py

+_base_ = [
+    '../_base_/datasets/ffhq_flip.py',
+    '../_base_/models/stylegan/stylegan2_base.py',
+    '../_base_/default_runtime.py'
+]
+
+model = dict(
+    type='MSPIEStyleGAN2',
+    generator=dict(
+        type='MSStyleGANv2Generator',
+        head_pos_encoding=dict(
+            type='SPE',
+            embedding_dim=256,
+            padding_idx=0,
+            init_size=256,
+            center_shift=100),
+        deconv2conv=True,
+        up_after_conv=False,
+        interp_pad=4,
+        no_pad=True,
+        head_pos_size=(6, 6),
+        interp_head=True,
+        up_config=dict(scale_factor=2, mode='bilinear', align_corners=True),
+        out_size=256),
+    discriminator=dict(
+        type='MSStyleGAN2Discriminator', in_size=256, with_adaptive_pool=True))
+
+train_cfg = dict(
+    num_upblocks=6,
+    multi_input_scales=[0, 2, 4],
+    multi_scale_probability=[0.5, 0.25, 0.25])
+
+data = dict(
+    samples_per_gpu=3,
+    train=dict(dataset=dict(imgs_root='./data/ffhq/ffhq_imgs/ffhq_512')))
+
+ema_half_life = 10.
+custom_hooks = [
+    dict(
+        type='VisualizeUnconditionalSamples',
+        output_dir='training_samples',
+        interval=5000),
+    dict(
+        type='ExponentialMovingAverageHook',
+        module_keys=('generator_ema', ),
+        interval=1,
+        interp_cfg=dict(momentum=0.5**(32. / (ema_half_life * 1000.))),
+        priority='VERY_HIGH')
+]
+
+checkpoint_config = dict(interval=10000, by_epoch=False, max_keep_ckpts=40)
+lr_config = None
+
+log_config = dict(
+    interval=100,
+    hooks=[
+        dict(type='TextLoggerHook'),
+        # dict(type='TensorboardLoggerHook'),
+    ])
+
+cudnn_benchmark = False
+total_iters = 1100002
+
+metrics = dict(
+    fid50k=dict(
+        type='FID',
+        num_images=50000,
+        inception_pkl='work_dirs/inception_pkl/ffhq-256-50k-rgb.pkl',
+        bgr2rgb=True),
+    pr10k3=dict(type='PR', num_images=10000, k=3))

configs/positional_encoding_in_gans/mspie-stylegan2_c2_config-k_ffhq_256-512_b3x8_1100k.py

+_base_ = [
+    '../_base_/datasets/ffhq_flip.py',
+    '../_base_/models/stylegan/stylegan2_base.py',
+    '../_base_/default_runtime.py'
+]
+
+model = dict(
+    type='MSPIEStyleGAN2',
+    generator=dict(
+        type='MSStyleGANv2Generator',
+        head_pos_encoding=dict(
+            type='SPE',
+            embedding_dim=256,
+            padding_idx=0,
+            init_size=256,
+            center_shift=100),
+        deconv2conv=True,
+        up_after_conv=False,
+        interp_pad=4,
+        no_pad=True,
+        head_pos_size=(6, 6),
+        up_config=dict(scale_factor=2, mode='bilinear', align_corners=True),
+        out_size=256),
+    discriminator=dict(
+        type='MSStyleGAN2Discriminator', in_size=256, with_adaptive_pool=True))
+
+train_cfg = dict(
+    num_upblocks=6,
+    multi_input_scales=[0, 2, 4],
+    multi_scale_probability=[0.5, 0.25, 0.25])
+
+data = dict(
+    samples_per_gpu=3,
+    train=dict(dataset=dict(imgs_root='./data/ffhq/ffhq_imgs/ffhq_512')))
+
+ema_half_life = 10.
+custom_hooks = [
+    dict(
+        type='VisualizeUnconditionalSamples',
+        output_dir='training_samples',
+        interval=5000),
+    dict(
+        type='ExponentialMovingAverageHook',
+        module_keys=('generator_ema', ),
+        interval=1,
+        interp_cfg=dict(momentum=0.5**(32. / (ema_half_life * 1000.))),
+        priority='VERY_HIGH')
+]
+
+checkpoint_config = dict(interval=10000, by_epoch=False, max_keep_ckpts=40)
+lr_config = None
+
+log_config = dict(
+    interval=100,
+    hooks=[
+        dict(type='TextLoggerHook'),
+        # dict(type='TensorboardLoggerHook'),
+    ])
+
+cudnn_benchmark = False
+total_iters = 1100002
+
+metrics = dict(
+    fid50k=dict(
+        type='FID',
+        num_images=50000,
+        inception_pkl='work_dirs/inception_pkl/ffhq-256-50k-rgb.pkl',
+        bgr2rgb=True),
+    pr10k3=dict(type='PR', num_images=10000, k=3))

configs/positional_encoding_in_gans/singan_csg_bohemian.py

+_base_ = ['../singan/singan_bohemian.py']
+
+num_scales = 10  # start from zero
+model = dict(
+    type='PESinGAN',
+    generator=dict(
+        type='SinGANMSGeneratorPE',
+        num_scales=num_scales,
+        padding=1,
+        pad_at_head=False,
+        first_stage_in_channels=2,
+        positional_encoding=dict(type='CSG')),
+    discriminator=dict(num_scales=num_scales))
+
+train_cfg = dict(first_fixed_noises_ch=2)
+
+data = dict(
+    train=dict(
+        img_path='./data/singan/bohemian.png',
+        min_size=25,
+        max_size=500,
+    ))
+
+dist_params = dict(backend='nccl')
+total_iters = 22000

configs/positional_encoding_in_gans/singan_csg_fish.py

+_base_ = ['../singan/singan_fish.py']
+
+num_scales = 10  # start from zero
+model = dict(
+    type='PESinGAN',
+    generator=dict(
+        type='SinGANMSGeneratorPE',
+        num_scales=num_scales,
+        padding=1,
+        pad_at_head=False,
+        first_stage_in_channels=2,
+        positional_encoding=dict(type='CSG')),
+    discriminator=dict(num_scales=num_scales))
+
+train_cfg = dict(first_fixed_noises_ch=2)
+
+data = dict(
+    train=dict(
+        img_path='./data/singan/fish-crop.jpg',
+        min_size=25,
+        max_size=300,
+    ))
+
+dist_params = dict(backend='nccl')
+total_iters = 22000

configs/positional_encoding_in_gans/singan_interp-pad_balloons.py

+_base_ = ['../singan/singan_balloons.py']
+
+model = dict(
+    type='PESinGAN',
+    generator=dict(
+        type='SinGANMSGeneratorPE', interp_pad=True, noise_with_pad=True))
+
+train_cfg = dict(fixed_noise_with_pad=True)
+
+dist_params = dict(backend='nccl')

configs/positional_encoding_in_gans/singan_interp-pad_disc-nobn_balloons.py

+_base_ = ['../singan/singan_balloons.py']
+
+model = dict(
+    type='PESinGAN',
+    generator=dict(
+        type='SinGANMSGeneratorPE', interp_pad=True, noise_with_pad=True),
+    discriminator=dict(norm_cfg=None))
+
+train_cfg = dict(fixed_noise_with_pad=True)
+
+dist_params = dict(backend='nccl')

configs/positional_encoding_in_gans/singan_interp-pad_disc-nobn_fish.py

+_base_ = ['../singan/singan_fish.py']
+
+model = dict(
+    type='PESinGAN',
+    generator=dict(
+        type='SinGANMSGeneratorPE', interp_pad=True, noise_with_pad=True),
+    discriminator=dict(norm_cfg=None))
+
+train_cfg = dict(fixed_noise_with_pad=True)
+
+data = dict(
+    train=dict(
+        img_path='./data/singan/fish-crop.jpg',
+        min_size=25,
+        max_size=300,
+    ))
+
+dist_params = dict(backend='nccl')

configs/positional_encoding_in_gans/singan_spe-dim4_bohemian.py

+_base_ = ['../singan/singan_bohemian.py']
+
+embedding_dim = 4
+num_scales = 10  # start from zero
+model = dict(
+    type='PESinGAN',
+    generator=dict(
+        type='SinGANMSGeneratorPE',
+        num_scales=num_scales,
+        padding=1,
+        pad_at_head=False,
+        first_stage_in_channels=embedding_dim * 2,
+        positional_encoding=dict(
+            type='SPE',
+            embedding_dim=embedding_dim,
+            padding_idx=0,
+            init_size=512,
+            div_half_dim=False,
+            center_shift=200)),
+    discriminator=dict(num_scales=num_scales))
+
+train_cfg = dict(first_fixed_noises_ch=embedding_dim * 2)
+
+data = dict(
+    train=dict(
+        img_path='./data/singan/bohemian.png',
+        min_size=25,
+        max_size=500,
+    ))
+
+dist_params = dict(backend='nccl')
+total_iters = 22000

configs/positional_encoding_in_gans/singan_spe-dim4_fish.py

+_base_ = ['../singan/singan_fish.py']
+
+embedding_dim = 4
+num_scales = 10  # start from zero
+model = dict(
+    type='PESinGAN',
+    generator=dict(
+        type='SinGANMSGeneratorPE',
+        num_scales=num_scales,
+        padding=1,
+        pad_at_head=False,
+        first_stage_in_channels=embedding_dim * 2,
+        positional_encoding=dict(
+            type='SPE',
+            embedding_dim=embedding_dim,
+            padding_idx=0,
+            init_size=512,
+            div_half_dim=False,
+            center_shift=200)),
+    discriminator=dict(num_scales=num_scales))
+
+data = dict(
+    train=dict(
+        img_path='./data/singan/fish-crop.jpg',
+        min_size=25,
+        max_size=300,
+    ))
+
+dist_params = dict(backend='nccl')
+total_iters = 22000

configs/positional_encoding_in_gans/singan_spe-dim8_bohemian.py

+_base_ = ['../singan/singan_bohemian.py']
+
+embedding_dim = 4
+num_scales = 10  # start from zero
+model = dict(
+    type='PESinGAN',
+    generator=dict(
+        type='SinGANMSGeneratorPE',
+        num_scales=num_scales,
+        padding=1,
+        pad_at_head=False,
+        first_stage_in_channels=embedding_dim * 2,
+        positional_encoding=dict(
+            type='SPE',
+            embedding_dim=embedding_dim,
+            padding_idx=0,
+            init_size=512,
+            div_half_dim=False,
+            center_shift=200)),
+    discriminator=dict(num_scales=num_scales))
+
+train_cfg = dict(first_fixed_noises_ch=embedding_dim * 2)
+
+data = dict(
+    train=dict(
+        img_path='./data/singan/bohemian.png',
+        min_size=25,
+        max_size=500,
+    ))
+
+dist_params = dict(backend='nccl')
+total_iters = 22000