fix cifar_c and configs

configs/classification/_base_/datasets/cifar100/sz224_corruption_bs100.py

@@ -0,0 +1,45 @@
+# dataset settings
+data_source_cfg = dict(type='CIFAR_C', root='data/cifar100_c/')
+
+dataset_type = 'ClassificationDataset'
+img_norm_cfg = dict(mean=[0.4914, 0.4822, 0.4465], std=[0.2023, 0.1994, 0.201])
+train_pipeline = [
+    dict(type='RandomResizedCrop', size=224, scale=[0.8, 1], interpolation=3),  # bicubic
+    dict(type='RandomHorizontalFlip'),
+]
+test_pipeline = [
+    dict(type='Resize', size=224, interpolation=3),
+]
+# prefetch
+prefetch = False
+if not prefetch:
+    train_pipeline.extend([dict(type='ToTensor'), dict(type='Normalize', **img_norm_cfg)])
+test_pipeline.extend([dict(type='ToTensor'), dict(type='Normalize', **img_norm_cfg)])
+
+data = dict(
+    imgs_per_gpu=100,  # 100 x 1gpu = 100
+    workers_per_gpu=4,
+    train=dict(
+        type=dataset_type,
+        data_source=dict(split='train', **data_source_cfg),
+        pipeline=train_pipeline,
+        prefetch=prefetch,
+    ),
+    val=dict(
+        type=dataset_type,
+        data_source=dict(split='test', **data_source_cfg),
+        pipeline=test_pipeline,
+        prefetch=False),
+)
+
+# validation hook
+evaluation = dict(
+    initial=False,
+    interval=1,
+    imgs_per_gpu=100,
+    workers_per_gpu=4,
+    eval_param=dict(topk=(1, 5)),
+    save_best='auto')
+
+# checkpoint
+checkpoint_config = dict(interval=10, max_keep_ckpts=1)

configs/classification/aircrafts/automix/r18_l2_a2_near_lam_cat_mb_mlr5e_4_1xb16_ep200.py

@@ -58,7 +58,7 @@
         end_momentum=0.99999,
         adjust_scope=[0.1, 1.0],
         warming_up="constant",
-        interval=1)
+        update_interval=1)
 ]
 # optimizer
 optimizer = dict(type='SGD', lr=0.001, momentum=0.9, weight_decay=0.0005,

configs/classification/aircrafts/automix/rx50_l2_a2_near_lam_cat_mb_mlr5e_4_1xb16_ep200.py

@@ -59,7 +59,7 @@
         end_momentum=0.99999,
         adjust_scope=[0.1, 1.0],
         warming_up="constant",
-        interval=1)
+        update_interval=1)
 ]
 # optimizer
 optimizer = dict(type='SGD', lr=0.001, momentum=0.9, weight_decay=0.0005,

configs/classification/aircrafts/samix/r18_l2_a2_bili_val_dp01_mb_mlr5e_4_1xb16_ep200.py

@@ -66,7 +66,7 @@
         end_momentum=0.99996,
         adjust_scope=[0.1, 1.0],
         warming_up="constant",
-        interval=1)
+        update_interval=1)
 ]
 # optimizer
 optimizer = dict(type='SGD', lr=0.001, momentum=0.9, weight_decay=0.0005,

configs/classification/aircrafts/samix/rx50_l2_a2_bili_val_dp01_mb_mlr5e_4_1xb16_ep200.py

@@ -66,7 +66,7 @@
         end_momentum=0.99996,
         adjust_scope=[0.1, 1.0],
         warming_up="constant",
-        interval=1)
+        update_interval=1)
 ]
 # optimizer
 optimizer = dict(type='SGD', lr=0.001, momentum=0.9, weight_decay=0.0005,

configs/classification/cifar100/automix/basic/r18/unsampling_modenearest/r18_l2_a2_near_L1_01_mlr5e_2_ep800.py

@@ -50,7 +50,7 @@
         end_momentum=0.999999,
         adjust_scope=[0.1, 1.0],
         warming_up="constant",
-        interval=1),
+        update_interval=1),
     dict(type='SAVEHook',
         iter_per_epoch=500,
         save_interval=12500,  # plot every 500 x 25 ep

configs/classification/cifar100/automix/basic/r18_l2_a2_near_L1_01_mlr5e_2.py

@@ -50,7 +50,7 @@
         end_momentum=0.999999,
         adjust_scope=[0.1, 1.0],
         warming_up="constant",
-        interval=1),
+        update_interval=1),
     dict(type='SAVEHook',
         iter_per_epoch=500,
         save_interval=12500,  # plot every 500 x 25 ep

configs/classification/cifar100/automix/basic/rx50_l2_a2_near_L1_01_mlr5e_2.py

@@ -51,7 +51,7 @@
         end_momentum=0.999999,
         adjust_scope=[0.1, 1.0],
         warming_up="constant",
-        interval=1),
+        update_interval=1),
     dict(type='SAVEHook',
         iter_per_epoch=500,
         save_interval=12500,  # plot every 500 x 25 ep

configs/classification/cifar100/automix/basic/wrn28_8_l1_a2_near_L1_01_mlr1e_3.py

@@ -52,7 +52,7 @@
         end_momentum=0.999999,
         adjust_scope=[0.1, 1.0],
         warming_up="constant",
-        interval=1),
+        update_interval=1),
     dict(type='SAVEHook',
         iter_per_epoch=500,
         save_interval=12500,  # plot every 500 x 25 ep

configs/classification/cifar100/automix/vits/convnext_t_l2_a2_near_val_dp01_lr2e_3_mlr1e_4_bs100_ep200.py

@@ -72,7 +72,7 @@
         adjust_scope=[0.25, 1.0],
         warming_up="constant",
         update_interval=update_interval,
-        interval=1)
+    )
 ]
 
 # optimizer

configs/classification/cifar100/automix/vits/convnext_t_l2_a2_near_val_dp01_lr2e_3_mlr1e_4_bs100_ep600.py

@@ -72,7 +72,7 @@
         adjust_scope=[0.25, 1.0],
         warming_up="constant",
         update_interval=update_interval,
-        interval=1)
+    )
 ]
 
 # optimizer

configs/classification/cifar100/automix/vits/deit_s_l6_a2_near_L1_01_sz224_mlr1e_4_bs100_ep200.py

@@ -69,7 +69,7 @@
         adjust_scope=[0.25, 1.0],
         warming_up="constant",
         update_interval=update_interval,
-        interval=1)
+    )
 ]
 
 # optimizer

configs/classification/cifar100/automix/vits/deit_s_l6_a2_near_L1_01_sz224_mlr1e_4_bs100_ep600.py

@@ -69,7 +69,7 @@
         adjust_scope=[0.25, 1.0],
         warming_up="constant",
         update_interval=update_interval,
-        interval=1)
+    )
 ]
 
 # optimizer

configs/classification/cifar100/automix/vits/swin_t_l3_a2_near_L1_01_sz224_mlr1e_4_bs100_ep200.py

@@ -71,7 +71,7 @@
         adjust_scope=[0.25, 1.0],
         warming_up="constant",
         update_interval=update_interval,
-        interval=1)
+    )
 ]
 
 # optimizer

configs/classification/cifar100/automix/vits/swin_t_l3_a2_near_bili_L1_01_sz224_mlr2e_4_bs100_ep600.py

@@ -71,7 +71,7 @@
         adjust_scope=[0.25, 1.0],
         warming_up="constant",
         update_interval=update_interval,
-        interval=1)
+    )
 ]
 
 # optimizer

configs/classification/cifar100/mixups/vits/deit_s_cifar100_c_eval_sz224_bs100.py

@@ -0,0 +1,76 @@
+_base_ = [
+    '../../../_base_/datasets/cifar100/sz224_corruption_bs100.py',
+    '../../../_base_/default_runtime.py',
+]
+
+# model settings
+model = dict(
+    type='MixUpClassification',
+    pretrained=None,
+    alpha=[1, 0.8],
+    mix_mode=['cutmix', 'mixup'],
+    mix_args=dict(
+        alignmix=dict(eps=0.1, max_iter=100),
+        attentivemix=dict(grid_size=32, top_k=None, beta=8),  # AttentiveMix+ in this repo (use pre-trained)
+        automix=dict(mask_adjust=0, lam_margin=0),  # require pre-trained mixblock
+        fmix=dict(decay_power=3, size=(224,224), max_soft=0., reformulate=False),
+        gridmix=dict(n_holes=(2, 6), hole_aspect_ratio=1.,
+            cut_area_ratio=(0.5, 1), cut_aspect_ratio=(0.5, 2)),
+        manifoldmix=dict(layer=(0, 3)),
+        puzzlemix=dict(transport=True, t_batch_size=32, t_size=-1,  # adjust t_batch_size if CUDA out of memory
+            mp=None, block_num=4,  # block_num<=4 and mp=2/4 for fast training
+            beta=1.2, gamma=0.5, eta=0.2, neigh_size=4, n_labels=3, t_eps=0.8),
+        resizemix=dict(scope=(0.1, 0.8), use_alpha=True, interpolate_mode="bilinear"),
+        samix=dict(mask_adjust=0, lam_margin=0.08),  # require pre-trained mixblock
+        transmix=dict(mix_mode="cutmix"),
+    ),
+    backbone=dict(
+        type='VisionTransformer',
+        arch='deit-small',
+        img_size=224, patch_size=16,
+        drop_path=0.1,
+    ),
+    head=dict(
+        type='VisionTransformerClsHead',  # mixup CE + label smooth
+        loss=dict(type='LabelSmoothLoss',
+            label_smooth_val=0.1, num_classes=100, mode='original', loss_weight=1.0),
+        in_channels=384, num_classes=100),
+    init_cfg=[
+        dict(type='TruncNormal', layer='Linear', std=0.02, bias=0.),
+        dict(type='Constant', layer=['LayerNorm', 'BatchNorm'], val=1., bias=0.)
+    ],
+)
+
+# optimizer
+optimizer = dict(
+    type='AdamW',
+    lr=1e-3,
+    weight_decay=0.05, eps=1e-8, betas=(0.9, 0.999),
+    paramwise_options={
+        '(bn|ln|gn)(\d+)?.(weight|bias)': dict(weight_decay=0.),
+        'norm': dict(weight_decay=0.),
+        'bias': dict(weight_decay=0.),
+        'cls_token': dict(weight_decay=0.),
+        'pos_embed': dict(weight_decay=0.),
+    })
+
+# interval for accumulate gradient
+update_interval = 1  # total: 1 x bs100 x 1 accumulates = bs100
+
+# fp16
+use_fp16 = False
+fp16 = dict(type='mmcv', loss_scale='dynamic')
+optimizer_config = dict(
+    grad_clip=dict(max_norm=5.0), update_interval=update_interval)
+
+# learning policy
+lr_config = dict(
+    policy='CosineAnnealing',
+    by_epoch=False, min_lr=1e-6,
+    warmup='linear',
+    warmup_iters=20, warmup_by_epoch=True,
+    warmup_ratio=1e-5,
+)
+
+# runtime settings
+runner = dict(type='EpochBasedRunner', max_epochs=200)

configs/classification/cifar100/mixups/vits/swin_t_cifar100_c_eval_sz224_bs100.py

@@ -0,0 +1,78 @@
+_base_ = [
+    '../../../_base_/datasets/cifar100/sz224_corruption_bs100.py',
+    '../../../_base_/default_runtime.py',
+]
+
+# model settings
+model = dict(
+    type='MixUpClassification',
+    pretrained=None,
+    alpha=[1, 0.8],
+    mix_mode=['cutmix', 'mixup'],
+    mix_args=dict(
+        alignmix=dict(eps=0.1, max_iter=100),
+        attentivemix=dict(grid_size=32, top_k=None, beta=8),  # AttentiveMix+ in this repo (use pre-trained)
+        automix=dict(mask_adjust=0, lam_margin=0),  # require pre-trained mixblock
+        fmix=dict(decay_power=3, size=(224,224), max_soft=0., reformulate=False),
+        gridmix=dict(n_holes=(2, 6), hole_aspect_ratio=1.,
+            cut_area_ratio=(0.5, 1), cut_aspect_ratio=(0.5, 2)),
+        manifoldmix=dict(layer=(0, 3)),
+        puzzlemix=dict(transport=True, t_batch_size=32, t_size=-1,  # adjust t_batch_size if CUDA out of memory
+            mp=None, block_num=4,  # block_num<=4 and mp=2/4 for fast training
+            beta=1.2, gamma=0.5, eta=0.2, neigh_size=4, n_labels=3, t_eps=0.8),
+        resizemix=dict(scope=(0.1, 0.8), use_alpha=True, interpolate_mode="bilinear"),
+        samix=dict(mask_adjust=0, lam_margin=0.08),  # require pre-trained mixblock
+        transmix=dict(mix_mode="cutmix"),
+    ),
+    backbone=dict(
+        type='SwinTransformer',
+        arch='tiny',
+        img_size=224,
+        drop_path_rate=0.2,
+        out_indices=(3,),  # x-1: stage-x
+    ),
+    head=dict(
+        type='ClsMixupHead',  # mixup CE + label smooth
+        loss=dict(type='LabelSmoothLoss',
+            label_smooth_val=0.1, num_classes=100, mode='original', loss_weight=1.0),
+        with_avg_pool=True,
+        in_channels=768, num_classes=100),
+    init_cfg=[
+        dict(type='TruncNormal', layer='Linear', std=0.02, bias=0.),
+        dict(type='Constant', layer=['LayerNorm', 'BatchNorm'], val=1., bias=0.)
+    ],
+)
+
+# optimizer
+optimizer = dict(
+    type='AdamW',
+    lr=5e-4,
+    weight_decay=0.05, eps=1e-8, betas=(0.9, 0.999),
+    paramwise_options={
+        '(bn|ln|gn)(\d+)?.(weight|bias)': dict(weight_decay=0.),
+        'norm': dict(weight_decay=0.),
+        'bias': dict(weight_decay=0.),
+        'absolute_pos_embed': dict(weight_decay=0.),
+        'relative_position_bias_table': dict(weight_decay=0.),
+    })
+
+# interval for accumulate gradient
+update_interval = 1  # total: 1 x bs100 x 1 accumulates = bs100
+
+# fp16
+use_fp16 = False
+fp16 = dict(type='mmcv', loss_scale='dynamic')
+optimizer_config = dict(
+    grad_clip=dict(max_norm=5.0), update_interval=update_interval)
+
+# learning policy
+lr_config = dict(
+    policy='CosineAnnealing',
+    by_epoch=False, min_lr=1e-6,
+    warmup='linear',
+    warmup_iters=20, warmup_by_epoch=True,
+    warmup_ratio=1e-5,
+)
+
+# runtime settings
+runner = dict(type='EpochBasedRunner', max_epochs=200)

configs/classification/cifar100/samix/basic/r18/unsampling_modebilinear/r18_l2_a2_bili_val_dp0_mul_x_cat_L1_var_01_mlr5e_2_lam_mul_k0_25_mask_adjust0_25_ep800.py

@@ -64,7 +64,7 @@
         end_momentum=0.999999,
         adjust_scope=[0.1, 1.0],
         warming_up="constant",
-        interval=1),
+        update_interval=1),
     dict(type='SAVEHook',
         iter_per_epoch=500,
         save_interval=12500,  # plot every 500 x 25 ep

configs/classification/cifar100/samix/basic/r18_l2_a2_bili_val_dp0_mul_x_cat_L1_var_01_mlr5e_2.py

@@ -64,7 +64,7 @@
         end_momentum=0.999999,
         adjust_scope=[0.1, 1.0],
         warming_up="constant",
-        interval=1),
+        update_interval=1),
     dict(type='SAVEHook',
         iter_per_epoch=500,
         save_interval=12500,  # plot every 500 x 25 ep

configs/classification/cifar100/samix/basic/rx50_l2_a2_bili_val_dp0_mul_x_cat_L1_var_01_mlr5e_2.py

@@ -65,7 +65,7 @@
         end_momentum=0.999999,
         adjust_scope=[0.1, 1.0],
         warming_up="constant",
-        interval=1),
+        update_interval=1),
     dict(type='SAVEHook',
         iter_per_epoch=500,
         save_interval=12500,  # plot every 500 x 25 ep

configs/classification/cifar100/samix/basic/wrn28_8_l1_a2_bili_val_dp0_mul_x_cat_L1_var_01_mlr1e_3.py

@@ -66,7 +66,7 @@
         end_momentum=0.999999,
         adjust_scope=[0.1, 1.0],
         warming_up="constant",
-        interval=1),
+        update_interval=1),
     dict(type='SAVEHook',
         iter_per_epoch=500,
         save_interval=12500,  # plot every 500 x 25 ep

configs/classification/cifar100/samix/vits/convnext_t_l2_a2_bili_val_dp01_lr2e_3_mlr1e_4_bs100_ep200.py

@@ -82,7 +82,7 @@
         adjust_scope=[0.25, 1.0],
         warming_up="constant",
         update_interval=update_interval,
-        interval=1)
+    )
 ]
 
 # optimizer

configs/classification/cifar100/samix/vits/convnext_t_l2_a2_bili_val_dp01_lr2e_3_mlr1e_4_bs100_ep600.py

@@ -82,7 +82,7 @@
         adjust_scope=[0.25, 1.0],
         warming_up="constant",
         update_interval=update_interval,
-        interval=1)
+    )
 ]
 
 # optimizer

configs/classification/cifar100/samix/vits/deit_s_l6_a2_near_val_dp01_sz224_mlr1e_4_bs100_ep200.py

@@ -79,7 +79,7 @@
         adjust_scope=[0.25, 1.0],
         warming_up="constant",
         update_interval=update_interval,
-        interval=1)
+    )
 ]
 
 # optimizer

configs/classification/cifar100/samix/vits/deit_s_l6_a2_near_val_dp01_sz224_mlr1e_4_bs100_ep600.py

@@ -79,7 +79,7 @@
         adjust_scope=[0.25, 1.0],
         warming_up="constant",
         update_interval=update_interval,
-        interval=1)
+    )
 ]
 
 # optimizer