Refactor default hooks configs

configs/_base_/default_runtime.py

@@ -1,16 +1,37 @@
-# checkpoint saving
-checkpoint_config = dict(interval=1)
-# yapf:disable
-log_config = dict(
-    interval=100,
-    hooks=[
-        dict(type='TextLoggerHook'),
-        # dict(type='TensorboardLoggerHook')
-    ])
-# yapf:enable
+default_scope = 'mmcls'
 
-dist_params = dict(backend='nccl')
+default_hooks = dict(
+    # optimizer configure
+    optimizer=dict(type='OptimizerHook', grad_clip=None),
+
+    # record the time to load data and the time it takes to iterate once
+    timer=dict(type='IterTimerHook'),
+
+    # logger configure
+    logger=dict(type='LoggerHook', interval=50),
+
+    # Parameter Scheduler
+    param_scheduler=dict(type='ParamSchedulerHook'),
+
+    # checkpoint saving
+    checkpoint=dict(type='CheckpointHook', interval=1),
+
+    # Sampler for distributed training
+    sampler_seed=dict(type='DistSamplerSeedHook'),
+)
+
+# Environment configure
+env_cfg = dict(
+    cudnn_benchmark=False,
+    mp_cfg=dict(mp_start_method='fork', opencv_num_threads=0),
+    dist_cfg=dict(backend='nccl'),
+)
+
+# Log level configuration
 log_level = 'INFO'
+
+# Load from weight
 load_from = None
-resume_from = None
-workflow = [('train', 1)]
+
+# resume training
+resume = False

configs/_base_/schedules/cifar10_bs128.py

@@ -1,6 +1,5 @@
 # optimizer
 optimizer = dict(type='SGD', lr=0.1, momentum=0.9, weight_decay=0.0001)
-optimizer_config = dict(grad_clip=None)
 # learning policy
 lr_config = dict(policy='step', step=[100, 150])
 runner = dict(type='EpochBasedRunner', max_epochs=200)

configs/_base_/schedules/cub_bs64.py

@@ -1,7 +1,6 @@
 # optimizer
 optimizer = dict(
     type='SGD', lr=0.01, momentum=0.9, weight_decay=0.0005, nesterov=True)
-optimizer_config = dict(grad_clip=None)
 # learning policy
 lr_config = dict(
     policy='CosineAnnealing',
@@ -10,4 +9,5 @@ lr_config = dict(
     warmup_iters=5,
     warmup_ratio=0.01,
     warmup_by_epoch=True)
+
 runner = dict(type='EpochBasedRunner', max_epochs=100)

configs/_base_/schedules/imagenet_bs1024_adamw_conformer.py

@@ -14,7 +14,6 @@ optimizer = dict(
     eps=1e-8,
     betas=(0.9, 0.999),
     paramwise_cfg=paramwise_cfg)
-optimizer_config = dict(grad_clip=None)
 
 # learning policy
 lr_config = dict(

configs/_base_/schedules/imagenet_bs1024_adamw_swin.py

@@ -15,7 +15,6 @@ optimizer = dict(
     eps=1e-8,
     betas=(0.9, 0.999),
     paramwise_cfg=paramwise_cfg)
-optimizer_config = dict(grad_clip=dict(max_norm=5.0))
 
 # learning policy
 lr_config = dict(

configs/_base_/schedules/imagenet_bs1024_coslr.py

@@ -1,6 +1,5 @@
 # optimizer
 optimizer = dict(type='SGD', lr=0.8, momentum=0.9, weight_decay=5e-5)
-optimizer_config = dict(grad_clip=None)
 # learning policy
 lr_config = dict(
     policy='CosineAnnealing',

configs/_base_/schedules/imagenet_bs1024_linearlr_bn_nowd.py

@@ -5,7 +5,6 @@ optimizer = dict(
     momentum=0.9,
     weight_decay=0.00004,
     paramwise_cfg=dict(norm_decay_mult=0))
-optimizer_config = dict(grad_clip=None)
 # learning policy
 lr_config = dict(
     policy='poly',

configs/_base_/schedules/imagenet_bs2048.py

@@ -1,7 +1,7 @@
 # optimizer
 optimizer = dict(
     type='SGD', lr=0.8, momentum=0.9, weight_decay=0.0001, nesterov=True)
-optimizer_config = dict(grad_clip=None)
+
 # learning policy
 lr_config = dict(
     policy='step',

configs/_base_/schedules/imagenet_bs2048_AdamW.py

@@ -2,7 +2,6 @@
 # In ClassyVision, the lr is set to 0.003 for bs4096.
 # In this implementation(bs2048), lr = 0.003 / 4096 * (32bs * 64gpus) = 0.0015
 optimizer = dict(type='AdamW', lr=0.0015, weight_decay=0.3)
-optimizer_config = dict(grad_clip=dict(max_norm=1.0))
 
 # specific to vit pretrain
 paramwise_cfg = dict(

configs/_base_/schedules/imagenet_bs2048_coslr.py

@@ -1,7 +1,6 @@
 # optimizer
 optimizer = dict(
     type='SGD', lr=0.8, momentum=0.9, weight_decay=0.0001, nesterov=True)
-optimizer_config = dict(grad_clip=None)
 # learning policy
 lr_config = dict(
     policy='CosineAnnealing',

configs/_base_/schedules/imagenet_bs2048_rsb.py

@@ -1,6 +1,6 @@
 # optimizer
 optimizer = dict(type='Lamb', lr=0.005, weight_decay=0.02)
-optimizer_config = dict(grad_clip=None)
+
 # learning policy
 lr_config = dict(
     policy='CosineAnnealing',

configs/_base_/schedules/imagenet_bs256.py

@@ -1,6 +1,5 @@
 # optimizer
 optimizer = dict(type='SGD', lr=0.1, momentum=0.9, weight_decay=0.0001)
-optimizer_config = dict(grad_clip=None)
 # learning policy
 lr_config = dict(policy='step', step=[30, 60, 90])
 runner = dict(type='EpochBasedRunner', max_epochs=100)

configs/_base_/schedules/imagenet_bs256_140e.py

@@ -1,6 +1,6 @@
 # optimizer
 optimizer = dict(type='SGD', lr=0.1, momentum=0.9, weight_decay=0.0001)
-optimizer_config = dict(grad_clip=None)
 # learning policy
 lr_config = dict(policy='step', step=[40, 80, 120])
+
 runner = dict(type='EpochBasedRunner', max_epochs=140)

configs/_base_/schedules/imagenet_bs256_200e_coslr_warmup.py

@@ -1,6 +1,5 @@
 # optimizer
 optimizer = dict(type='SGD', lr=0.1, momentum=0.9, weight_decay=0.0001)
-optimizer_config = dict(grad_clip=None)
 # learning policy
 lr_config = dict(
     policy='CosineAnnealing',

configs/_base_/schedules/imagenet_bs256_coslr.py

@@ -1,6 +1,6 @@
 # optimizer
 optimizer = dict(type='SGD', lr=0.1, momentum=0.9, weight_decay=0.0001)
-optimizer_config = dict(grad_clip=None)
 # learning policy
 lr_config = dict(policy='CosineAnnealing', min_lr=0)
+
 runner = dict(type='EpochBasedRunner', max_epochs=100)

configs/_base_/schedules/imagenet_bs256_epochstep.py

@@ -1,6 +1,5 @@
 # optimizer
 optimizer = dict(type='SGD', lr=0.045, momentum=0.9, weight_decay=0.00004)
-optimizer_config = dict(grad_clip=None)
 # learning policy
 lr_config = dict(policy='step', gamma=0.98, step=1)
 runner = dict(type='EpochBasedRunner', max_epochs=300)

configs/_base_/schedules/imagenet_bs4096_AdamW.py

@@ -11,7 +11,6 @@ optimizer = dict(
     weight_decay=0.3,
     paramwise_cfg=paramwise_cfg,
 )
-optimizer_config = dict(grad_clip=dict(max_norm=1.0))
 
 # learning policy
 lr_config = dict(

configs/convmixer/convmixer-1024-20_10xb64_in1k.py

@@ -5,6 +5,8 @@ _base_ = [
     '../_base_/default_runtime.py',
 ]
 
+default_hooks = dict(optimizer=dict(grad_clip=dict(max_norm=5.0)))
+
 optimizer = dict(lr=0.01)
 
 runner = dict(type='EpochBasedRunner', max_epochs=150)

configs/convmixer/convmixer-1536-20_10xb64_in1k.py

@@ -5,6 +5,8 @@ _base_ = [
     '../_base_/default_runtime.py',
 ]
 
+default_hooks = dict(optimizer=dict(grad_clip=dict(max_norm=5.0)))
+
 optimizer = dict(lr=0.01)
 
 runner = dict(type='EpochBasedRunner', max_epochs=150)

configs/convmixer/convmixer-768-32_10xb64_in1k.py

@@ -5,6 +5,8 @@ _base_ = [
     '../_base_/default_runtime.py',
 ]
 
+default_hooks = dict(optimizer=dict(grad_clip=dict(max_norm=5.0)))
+
 optimizer = dict(lr=0.01)
 
 runner = dict(type='EpochBasedRunner', max_epochs=300)

configs/convnext/convnext-base_32xb128_in1k.py

@@ -5,6 +5,8 @@ _base_ = [
     '../_base_/default_runtime.py',
 ]
 
+default_hooks = dict(optimizer=dict(grad_clip=dict(max_norm=5.0)))
+
 data = dict(samples_per_gpu=128)
 
 optimizer = dict(lr=4e-3)

configs/convnext/convnext-large_64xb64_in1k.py

@@ -5,6 +5,8 @@ _base_ = [
     '../_base_/default_runtime.py',
 ]
 
+default_hooks = dict(optimizer=dict(grad_clip=dict(max_norm=5.0)))
+
 data = dict(samples_per_gpu=64)
 
 optimizer = dict(lr=4e-3)

configs/convnext/convnext-small_32xb128_in1k.py

@@ -5,6 +5,8 @@ _base_ = [
     '../_base_/default_runtime.py',
 ]
 
+default_hooks = dict(optimizer=dict(grad_clip=dict(max_norm=5.0)))
+
 data = dict(samples_per_gpu=128)
 
 optimizer = dict(lr=4e-3)

configs/convnext/convnext-tiny_32xb128_in1k.py

@@ -5,6 +5,8 @@ _base_ = [
     '../_base_/default_runtime.py',
 ]
 
+default_hooks = dict(optimizer=dict(grad_clip=dict(max_norm=5.0)))
+
 data = dict(samples_per_gpu=128)
 
 optimizer = dict(lr=4e-3)

configs/convnext/convnext-xlarge_64xb64_in1k.py

@@ -5,6 +5,8 @@ _base_ = [
     '../_base_/default_runtime.py',
 ]
 
+default_hooks = dict(optimizer=dict(grad_clip=dict(max_norm=5.0)))
+
 data = dict(samples_per_gpu=64)
 
 optimizer = dict(lr=4e-3)

configs/deit/deit-base_ft-16xb32_in1k-384px.py

@@ -4,6 +4,8 @@ _base_ = [
     '../_base_/default_runtime.py'
 ]
 
+default_hooks = dict(optimizer=dict(grad_clip=dict(max_norm=1.0)))
+
 # model settings
 model = dict(
     type='ImageClassifier',

configs/deit/deit-small_pt-4xb256_in1k.py

@@ -6,6 +6,8 @@ _base_ = [
     '../_base_/default_runtime.py'
 ]
 
+default_hooks = dict(optimizer=dict(grad_clip=dict(max_norm=5.0)))
+
 # model settings
 model = dict(
     type='ImageClassifier',

configs/fp16/resnet50_b32x8_fp16_dynamic_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = '../resnet/resnet50_8xb32-fp16-dynamic_in1k.py'
-
-_deprecation_ = dict(
-    expected='../resnet/resnet50_8xb32-fp16-dynamic_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/mlp_mixer/mlp-mixer-base-p16_64xb64_in1k.py

@@ -4,3 +4,5 @@ _base_ = [
     '../_base_/schedules/imagenet_bs4096_AdamW.py',
     '../_base_/default_runtime.py',
 ]
+
+default_hooks = dict(optimizer=dict(grad_clip=dict(max_norm=1.0)))

configs/mlp_mixer/mlp-mixer-large-p16_64xb64_in1k.py

@@ -4,3 +4,5 @@ _base_ = [
     '../_base_/schedules/imagenet_bs4096_AdamW.py',
     '../_base_/default_runtime.py',
 ]
+
+default_hooks = dict(optimizer=dict(grad_clip=dict(max_norm=1.0)))

configs/mobilenet_v2/mobilenet_v2_b32x8_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'mobilenet-v2_8xb32_in1k.py'
-
-_deprecation_ = dict(
-    expected='mobilenet-v2_8xb32_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/mobilenet_v3/mobilenet_v3_large_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'mobilenet-v3-large_8xb32_in1k.py'
-
-_deprecation_ = dict(
-    expected='mobilenet-v3-large_8xb32_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/mobilenet_v3/mobilenet_v3_small_cifar.py

@@ -1,6 +0,0 @@
-_base_ = 'mobilenet-v3-small_8xb16_cifar10.py'
-
-_deprecation_ = dict(
-    expected='mobilenet-v3-small_8xb16_cifar10.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/mobilenet_v3/mobilenet_v3_small_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'mobilenet-v3-small_8xb32_in1k.py'
-
-_deprecation_ = dict(
-    expected='mobilenet-v3-small_8xb32_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/poolformer/poolformer-m36_32xb128_in1k.py

@@ -5,4 +5,6 @@ _base_ = [
     '../_base_/default_runtime.py',
 ]
 
+default_hooks = dict(optimizer=dict(grad_clip=dict(max_norm=5.0)))
+
 optimizer = dict(lr=4e-3)

configs/poolformer/poolformer-m48_32xb128_in1k.py

@@ -5,4 +5,6 @@ _base_ = [
     '../_base_/default_runtime.py',
 ]
 
+default_hooks = dict(optimizer=dict(grad_clip=dict(max_norm=5.0)))
+
 optimizer = dict(lr=4e-3)

configs/poolformer/poolformer-s12_32xb128_in1k.py

@@ -5,4 +5,6 @@ _base_ = [
     '../_base_/default_runtime.py',
 ]
 
+default_hooks = dict(optimizer=dict(grad_clip=dict(max_norm=5.0)))
+
 optimizer = dict(lr=4e-3)

configs/poolformer/poolformer-s24_32xb128_in1k.py

@@ -5,4 +5,6 @@ _base_ = [
     '../_base_/default_runtime.py',
 ]
 
+default_hooks = dict(optimizer=dict(grad_clip=dict(max_norm=5.0)))
+
 optimizer = dict(lr=4e-3)

configs/poolformer/poolformer-s36_32xb128_in1k.py

@@ -5,4 +5,6 @@ _base_ = [
     '../_base_/default_runtime.py',
 ]
 
+default_hooks = dict(optimizer=dict(grad_clip=dict(max_norm=5.0)))
+
 optimizer = dict(lr=4e-3)

configs/repmlp/repmlp-base_8xb64_in1k-256px.py

@@ -5,6 +5,8 @@ _base_ = [
     '../_base_/default_runtime.py'
 ]
 
+default_hooks = dict(optimizer=dict(grad_clip=dict(max_norm=1.0)))
+
 model = dict(backbone=dict(img_size=256))
 
 img_norm_cfg = dict(

configs/repmlp/repmlp-base_8xb64_in1k.py

@@ -5,6 +5,8 @@ _base_ = [
     '../_base_/default_runtime.py'
 ]
 
+default_hooks = dict(optimizer=dict(grad_clip=dict(max_norm=5.0)))
+
 img_norm_cfg = dict(
     mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
 test_pipeline = [

configs/resnest/resnest101_b64x32_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'resnest101_32xb64_in1k.py'
-
-_deprecation_ = dict(
-    expected='resnest101_32xb64_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/resnest/resnest200_b32x64_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'resnest200_64xb32_in1k.py'
-
-_deprecation_ = dict(
-    expected='resnest200_64xb32_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/resnest/resnest269_b32x64_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'resnest269_64xb32_in1k.py'
-
-_deprecation_ = dict(
-    expected='resnest269_64xb32_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/resnest/resnest50_b64x32_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'resnest50_32xb64_in1k.py'
-
-_deprecation_ = dict(
-    expected='resnest50_32xb64_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/resnet/resnet101_b16x8_cifar10.py

@@ -1,6 +0,0 @@
-_base_ = 'resnet101_8xb16_cifar10.py'
-
-_deprecation_ = dict(
-    expected='resnet101_8xb16_cifar10.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/resnet/resnet101_b32x8_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'resnet101_8xb32_in1k.py'
-
-_deprecation_ = dict(
-    expected='resnet101_8xb32_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/resnet/resnet152_b16x8_cifar10.py

@@ -1,6 +0,0 @@
-_base_ = 'resnet152_8xb16_cifar10.py'
-
-_deprecation_ = dict(
-    expected='resnet152_8xb16_cifar10.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/resnet/resnet152_b32x8_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'resnet152_8xb32_in1k.py'
-
-_deprecation_ = dict(
-    expected='resnet152_8xb32_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/resnet/resnet18_b16x8_cifar10.py

@@ -1,6 +0,0 @@
-_base_ = 'resnet18_8xb16_cifar10.py'
-
-_deprecation_ = dict(
-    expected='resnet18_8xb16_cifar10.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/resnet/resnet18_b32x8_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'resnet18_8xb32_in1k.py'
-
-_deprecation_ = dict(
-    expected='resnet18_8xb32_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/resnet/resnet34_b32x8_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'resnet34_8xb32_in1k.py'
-
-_deprecation_ = dict(
-    expected='resnet34_8xb32_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/resnet/resnet50_b16x8_cifar100.py

@@ -1,6 +0,0 @@
-_base_ = 'resnet50_8xb16_cifar100.py'
-
-_deprecation_ = dict(
-    expected='resnet50_8xb16_cifar100.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/resnet/resnet50_b16x8_cifar10_mixup.py

@@ -1,6 +0,0 @@
-_base_ = 'resnet50_8xb16-mixup_cifar10.py'
-
-_deprecation_ = dict(
-    expected='resnet50_8xb16-mixup_cifar10.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/resnet/resnet50_b32x8_coslr_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'resnet50_8xb32-coslr_in1k.py'
-
-_deprecation_ = dict(
-    expected='resnet50_8xb32-coslr_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/resnet/resnet50_b32x8_cutmix_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'resnet50_8xb32-cutmix_in1k.py'
-
-_deprecation_ = dict(
-    expected='resnet50_8xb32-cutmix_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/resnet/resnet50_b32x8_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'resnet50_8xb32_in1k.py'
-
-_deprecation_ = dict(
-    expected='resnet50_8xb32_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/resnet/resnet50_b32x8_label_smooth_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'resnet50_8xb32-lbs_in1k.py'
-
-_deprecation_ = dict(
-    expected='resnet50_8xb32-lbs_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/resnet/resnet50_b32x8_mixup_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'resnet50_8xb32-mixup_in1k.py'
-
-_deprecation_ = dict(
-    expected='resnet50_8xb32-mixup_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/resnet/resnet50_b64x32_warmup_coslr_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'resnet50_32xb64-warmup-coslr_in1k.py'
-
-_deprecation_ = dict(
-    expected='resnet50_32xb64-warmup-coslr_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/resnet/resnet50_b64x32_warmup_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'resnet50_32xb64-warmup_in1k.py'
-
-_deprecation_ = dict(
-    expected='resnet50_32xb64-warmup_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/resnet/resnet50_b64x32_warmup_label_smooth_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'resnet50_32xb64-warmup-lbs_in1k.py'
-
-_deprecation_ = dict(
-    expected='resnet50_32xb64-warmup-lbs_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/resnet/resnetv1d101_b32x8_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'resnetv1d101_8xb32_in1k.py'
-
-_deprecation_ = dict(
-    expected='resnetv1d101_8xb32_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/resnet/resnetv1d152_b32x8_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'resnetv1d152_8xb32_in1k.py'
-
-_deprecation_ = dict(
-    expected='resnetv1d152_8xb32_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/resnet/resnetv1d50_b32x8_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'resnetv1d50_8xb32_in1k.py'
-
-_deprecation_ = dict(
-    expected='resnetv1d50_8xb32_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/resnext/resnext101_32x4d_b32x8_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'resnext101-32x4d_8xb32_in1k.py'
-
-_deprecation_ = dict(
-    expected='resnext101-32x4d_8xb32_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/resnext/resnext101_32x8d_b32x8_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'resnext101-32x8d_8xb32_in1k.py'
-
-_deprecation_ = dict(
-    expected='resnext101-32x8d_8xb32_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/resnext/resnext152_32x4d_b32x8_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'resnext152-32x4d_8xb32_in1k.py'
-
-_deprecation_ = dict(
-    expected='resnext152-32x4d_8xb32_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/resnext/resnext50_32x4d_b32x8_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'resnext50-32x4d_8xb32_in1k.py'
-
-_deprecation_ = dict(
-    expected='resnext50-32x4d_8xb32_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/seresnet/seresnet101_b32x8_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'seresnet101_8xb32_in1k.py'
-
-_deprecation_ = dict(
-    expected='seresnet101_8xb32_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/seresnet/seresnet50_b32x8_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'seresnet50_8xb32_in1k.py'
-
-_deprecation_ = dict(
-    expected='seresnet50_8xb32_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/seresnet/seresnext101_32x4d_b32x8_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'seresnext101-32x4d_8xb32_in1k.py'
-
-_deprecation_ = dict(
-    expected='seresnext101-32x4d_8xb32_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/seresnet/seresnext50_32x4d_b32x8_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'seresnext50-32x4d_8xb32_in1k.py'
-
-_deprecation_ = dict(
-    expected='seresnext50-32x4d_8xb32_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/shufflenet_v1/shufflenet_v1_1x_b64x16_linearlr_bn_nowd_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'shufflenet-v1-1x_16xb64_in1k.py'
-
-_deprecation_ = dict(
-    expected='shufflenet-v1-1x_16xb64_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/swin_transformer/swin-base_16xb64_in1k-384px.py

@@ -5,3 +5,5 @@ _base_ = [
     '../_base_/schedules/imagenet_bs1024_adamw_swin.py',
     '../_base_/default_runtime.py'
 ]
+
+default_hooks = dict(optimizer=dict(grad_clip=dict(max_norm=5.0)))

configs/swin_transformer/swin-base_16xb64_in1k.py

@@ -4,3 +4,5 @@ _base_ = [
     '../_base_/schedules/imagenet_bs1024_adamw_swin.py',
     '../_base_/default_runtime.py'
 ]
+
+default_hooks = dict(optimizer=dict(grad_clip=dict(max_norm=5.0)))

configs/swin_transformer/swin-large_16xb64_in1k-384px.py

@@ -5,3 +5,5 @@ _base_ = [
     '../_base_/schedules/imagenet_bs1024_adamw_swin.py',
     '../_base_/default_runtime.py'
 ]
+
+default_hooks = dict(optimizer=dict(grad_clip=dict(max_norm=5.0)))

configs/swin_transformer/swin-large_16xb64_in1k.py

@@ -5,3 +5,5 @@ _base_ = [
     '../_base_/schedules/imagenet_bs1024_adamw_swin.py',
     '../_base_/default_runtime.py'
 ]
+
+default_hooks = dict(optimizer=dict(grad_clip=dict(max_norm=5.0)))

configs/swin_transformer/swin-small_16xb64_in1k.py

@@ -4,3 +4,5 @@ _base_ = [
     '../_base_/schedules/imagenet_bs1024_adamw_swin.py',
     '../_base_/default_runtime.py'
 ]
+
+default_hooks = dict(optimizer=dict(grad_clip=dict(max_norm=5.0)))

configs/swin_transformer/swin-tiny_16xb64_in1k.py

@@ -4,3 +4,5 @@ _base_ = [
     '../_base_/schedules/imagenet_bs1024_adamw_swin.py',
     '../_base_/default_runtime.py'
 ]
+
+default_hooks = dict(optimizer=dict(grad_clip=dict(max_norm=5.0)))

configs/swin_transformer/swin_base_224_b16x64_300e_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'swin-base_16xb64_in1k.py'
-
-_deprecation_ = dict(
-    expected='swin-base_16xb64_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/swin_transformer/swin_base_384_evalonly_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'swin-base_16xb64_in1k-384px.py'
-
-_deprecation_ = dict(
-    expected='swin-base_16xb64_in1k-384px.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/swin_transformer/swin_large_224_evalonly_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'swin-large_16xb64_in1k.py'
-
-_deprecation_ = dict(
-    expected='swin-large_16xb64_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/swin_transformer/swin_large_384_evalonly_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'swin-large_16xb64_in1k-384px.py'
-
-_deprecation_ = dict(
-    expected='swin-large_16xb64_in1k-384px.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/swin_transformer/swin_small_224_b16x64_300e_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'swin-small_16xb64_in1k.py'
-
-_deprecation_ = dict(
-    expected='swin-small_16xb64_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/swin_transformer/swin_tiny_224_b16x64_300e_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'swin-tiny_16xb64_in1k.py'
-
-_deprecation_ = dict(
-    expected='swin-tiny_16xb64_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/tnt/tnt_s_patch16_224_evalonly_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'tnt-s-p16_16xb64_in1k.py'
-
-_deprecation_ = dict(
-    expected='tnt-s-p16_16xb64_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/twins/twins-pcpvt-base_8xb128_in1k.py

@@ -5,6 +5,9 @@ _base_ = [
     '../_base_/default_runtime.py'
 ]
 
+default_hooks = dict(
+    optimizer=dict(_delete_=True, grad_clip=dict(max_norm=5.0)))
+
 data = dict(samples_per_gpu=128)
 
 paramwise_cfg = dict(_delete=True, norm_decay_mult=0.0, bias_decay_mult=0.0)
@@ -18,7 +21,6 @@ optimizer = dict(
     eps=1e-8,
     betas=(0.9, 0.999),
     paramwise_cfg=paramwise_cfg)
-optimizer_config = dict(_delete_=True, grad_clip=dict(max_norm=5.0))
 
 # learning policy
 lr_config = dict(

configs/twins/twins-svt-base_8xb128_in1k.py

@@ -5,6 +5,9 @@ _base_ = [
     '../_base_/default_runtime.py'
 ]
 
+default_hooks = dict(
+    optimizer=dict(_delete_=True, grad_clip=dict(max_norm=5.0)))
+
 data = dict(samples_per_gpu=128)
 
 paramwise_cfg = dict(_delete=True, norm_decay_mult=0.0, bias_decay_mult=0.0)
@@ -18,9 +21,7 @@ optimizer = dict(
     eps=1e-8,
     betas=(0.9, 0.999),
     paramwise_cfg=paramwise_cfg)
-optimizer_config = dict(_delete_=True, grad_clip=dict(max_norm=5.0))
 
-# learning policy
 lr_config = dict(
     policy='CosineAnnealing',
     by_epoch=True,

configs/van/van-base_8xb128_in1k.py

@@ -5,6 +5,8 @@ _base_ = [
     '../_base_/default_runtime.py'
 ]
 
+default_hooks = dict(optimizer=dict(grad_clip=dict(max_norm=5.0)))
+
 # Note that the mean and variance used here are different from other configs
 img_norm_cfg = dict(
     mean=[127.5, 127.5, 127.5], std=[127.5, 127.5, 127.5], to_rgb=True)

configs/van/van-large_8xb128_in1k.py

@@ -5,6 +5,8 @@ _base_ = [
     '../_base_/default_runtime.py'
 ]
 
+default_hooks = dict(optimizer=dict(grad_clip=dict(max_norm=5.0)))
+
 # Note that the mean and variance used here are different from other configs
 img_norm_cfg = dict(
     mean=[127.5, 127.5, 127.5], std=[127.5, 127.5, 127.5], to_rgb=True)

configs/van/van-small_8xb128_in1k.py

@@ -5,6 +5,8 @@ _base_ = [
     '../_base_/default_runtime.py'
 ]
 
+default_hooks = dict(optimizer=dict(grad_clip=dict(max_norm=5.0)))
+
 # Note that the mean and variance used here are different from other configs
 img_norm_cfg = dict(
     mean=[127.5, 127.5, 127.5], std=[127.5, 127.5, 127.5], to_rgb=True)

configs/van/van-tiny_8xb128_in1k.py

@@ -5,6 +5,8 @@ _base_ = [
     '../_base_/default_runtime.py'
 ]
 
+default_hooks = dict(optimizer=dict(grad_clip=dict(max_norm=5.0)))
+
 # Note that the mean and variance used here are different from other configs
 img_norm_cfg = dict(
     mean=[127.5, 127.5, 127.5], std=[127.5, 127.5, 127.5], to_rgb=True)

configs/vgg/vgg11_b32x8_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'vgg11_8xb32_in1k.py'
-
-_deprecation_ = dict(
-    expected='vgg11_8xb32_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/vgg/vgg11bn_b32x8_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'vgg11bn_8xb32_in1k.py'
-
-_deprecation_ = dict(
-    expected='vgg11bn_8xb32_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/vgg/vgg13_b32x8_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'vgg13_8xb32_in1k.py'
-
-_deprecation_ = dict(
-    expected='vgg13_8xb32_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/vgg/vgg13bn_b32x8_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'vgg13bn_8xb32_in1k.py'
-
-_deprecation_ = dict(
-    expected='vgg13bn_8xb32_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/vgg/vgg16_8xb16_voc.py

@@ -22,4 +22,5 @@ optimizer = dict(
 optimizer_config = dict(grad_clip=None)
 # learning policy
 lr_config = dict(policy='step', step=20, gamma=0.1)
+
 runner = dict(type='EpochBasedRunner', max_epochs=40)

configs/vgg/vgg16_b32x8_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'vgg16_8xb32_in1k.py'
-
-_deprecation_ = dict(
-    expected='vgg16_8xb32_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)

configs/vgg/vgg16bn_b32x8_imagenet.py

@@ -1,6 +0,0 @@
-_base_ = 'vgg16bn_8xb32_in1k.py'
-
-_deprecation_ = dict(
-    expected='vgg16bn_8xb32_in1k.py',
-    reference='https://github.com/open-mmlab/mmclassification/pull/508',
-)