mmpretrain/configs/swin_transformer/metafile.yml

Collections:
  - Name: Swin-Transformer
    Metadata:
      Training Data: ImageNet-1k
      Training Techniques:
        - AdamW
        - Weight Decay
      Training Resources: 16x V100 GPUs
      Epochs: 300
      Batch Size: 1024
      Architecture:
        - Shift Window Multihead Self Attention
    Paper:
      URL: https://arxiv.org/abs/2103.14030
      Title: "Swin Transformer: Hierarchical Vision Transformer using Shifted Windows"
    README: configs/swin_transformer/README.md
    Code:
      URL: https://github.com/open-mmlab/mmpretrain/blob/v0.15.0/mmcls/models/backbones/swin_transformer.py#L176
      Version: v0.15.0

Models:
  - Name: swin-tiny_16xb64_in1k
    Metadata:
      FLOPs: 4360000000
      Parameters: 28290000
    In Collection: Swin-Transformer
    Results:
      - Dataset: ImageNet-1k
        Metrics:
          Top 1 Accuracy: 81.18
          Top 5 Accuracy: 95.61
        Task: Image Classification
    Weights: https://download.openmmlab.com/mmclassification/v0/swin-transformer/swin_tiny_224_b16x64_300e_imagenet_20210616_090925-66df6be6.pth
    Config: configs/swin_transformer/swin-tiny_16xb64_in1k.py
  - Name: swin-small_16xb64_in1k
    Metadata:
      FLOPs: 8520000000
      Parameters: 49610000
    In Collection: Swin-Transformer
    Results:
      - Dataset: ImageNet-1k
        Metrics:
          Top 1 Accuracy: 83.02
          Top 5 Accuracy: 96.29
        Task: Image Classification
    Weights: https://download.openmmlab.com/mmclassification/v0/swin-transformer/swin_small_224_b16x64_300e_imagenet_20210615_110219-7f9d988b.pth
    Config: configs/swin_transformer/swin-small_16xb64_in1k.py
  - Name: swin-base_16xb64_in1k
    Metadata:
      FLOPs: 15140000000
      Parameters: 87770000
    In Collection: Swin-Transformer
    Results:
      - Dataset: ImageNet-1k
        Metrics:
          Top 1 Accuracy: 83.36
          Top 5 Accuracy: 96.44
        Task: Image Classification
    Weights: https://download.openmmlab.com/mmclassification/v0/swin-transformer/swin_base_224_b16x64_300e_imagenet_20210616_190742-93230b0d.pth
    Config: configs/swin_transformer/swin-base_16xb64_in1k.py
  - Name: swin-tiny_3rdparty_in1k
    Metadata:
      FLOPs: 4360000000
      Parameters: 28290000
    In Collection: Swin-Transformer
    Results:
      - Dataset: ImageNet-1k
        Metrics:
          Top 1 Accuracy: 81.18
          Top 5 Accuracy: 95.52
        Task: Image Classification
    Weights: https://download.openmmlab.com/mmclassification/v0/swin-transformer/convert/swin_tiny_patch4_window7_224-160bb0a5.pth
    Converted From:
      Weights: https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_tiny_patch4_window7_224.pth
      Code: https://github.com/microsoft/Swin-Transformer/blob/777f6c66604bb5579086c4447efe3620344d95a9/models/swin_transformer.py#L458
    Config: configs/swin_transformer/swin-tiny_16xb64_in1k.py
  - Name: swin-small_3rdparty_in1k
    Metadata:
      FLOPs: 8520000000
      Parameters: 49610000
    In Collection: Swin-Transformer
    Results:
      - Dataset: ImageNet-1k
        Metrics:
          Top 1 Accuracy: 83.21
          Top 5 Accuracy: 96.25
        Task: Image Classification
    Weights: https://download.openmmlab.com/mmclassification/v0/swin-transformer/convert/swin_small_patch4_window7_224-cc7a01c9.pth
    Converted From:
      Weights: https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_small_patch4_window7_224.pth
      Code: https://github.com/microsoft/Swin-Transformer/blob/777f6c66604bb5579086c4447efe3620344d95a9/models/swin_transformer.py#L458
    Config: configs/swin_transformer/swin-small_16xb64_in1k.py
  - Name: swin-base_3rdparty_in1k
    Metadata:
      FLOPs: 15140000000
      Parameters: 87770000
    In Collection: Swin-Transformer
    Results:
      - Dataset: ImageNet-1k
        Metrics:
          Top 1 Accuracy: 83.42
          Top 5 Accuracy: 96.44
        Task: Image Classification
    Weights: https://download.openmmlab.com/mmclassification/v0/swin-transformer/convert/swin_base_patch4_window7_224-4670dd19.pth
    Converted From:
      Weights: https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_base_patch4_window7_224.pth
      Code: https://github.com/microsoft/Swin-Transformer/blob/777f6c66604bb5579086c4447efe3620344d95a9/models/swin_transformer.py#L458
    Config: configs/swin_transformer/swin-base_16xb64_in1k.py
  - Name: swin-base_3rdparty_in1k-384
    Metadata:
      FLOPs: 44490000000
      Parameters: 87900000
    In Collection: Swin-Transformer
    Results:
      - Dataset: ImageNet-1k
        Metrics:
          Top 1 Accuracy: 84.49
          Top 5 Accuracy: 96.95
        Task: Image Classification
    Weights: https://download.openmmlab.com/mmclassification/v0/swin-transformer/convert/swin_base_patch4_window12_384-02c598a4.pth
    Converted From:
      Weights: https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_base_patch4_window12_384.pth
      Code: https://github.com/microsoft/Swin-Transformer/blob/777f6c66604bb5579086c4447efe3620344d95a9/models/swin_transformer.py#L458
    Config: configs/swin_transformer/swin-base_16xb64_in1k-384px.py
  - Name: swin-base_in21k-pre-3rdparty_in1k
    Metadata:
      FLOPs: 15140000000
      Parameters: 87770000
    In Collection: Swin-Transformer
    Results:
      - Dataset: ImageNet-1k
        Metrics:
          Top 1 Accuracy: 85.16
          Top 5 Accuracy: 97.50
        Task: Image Classification
    Weights: https://download.openmmlab.com/mmclassification/v0/swin-transformer/convert/swin_base_patch4_window7_224_22kto1k-f967f799.pth
    Converted From:
      Weights: https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_base_patch4_window7_224_22kto1k.pth
      Code: https://github.com/microsoft/Swin-Transformer/blob/777f6c66604bb5579086c4447efe3620344d95a9/models/swin_transformer.py#L458
    Config: configs/swin_transformer/swin-base_16xb64_in1k.py
  - Name: swin-base_in21k-pre-3rdparty_in1k-384
    Metadata:
      FLOPs: 44490000000
      Parameters: 87900000
    In Collection: Swin-Transformer
    Results:
      - Dataset: ImageNet-1k
        Metrics:
          Top 1 Accuracy: 86.44
          Top 5 Accuracy: 98.05
        Task: Image Classification
    Weights: https://download.openmmlab.com/mmclassification/v0/swin-transformer/convert/swin_base_patch4_window12_384_22kto1k-d59b0d1d.pth
    Converted From:
      Weights: https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_base_patch4_window12_384_22kto1k.pth
      Code: https://github.com/microsoft/Swin-Transformer/blob/777f6c66604bb5579086c4447efe3620344d95a9/models/swin_transformer.py#L458
    Config: configs/swin_transformer/swin-base_16xb64_in1k-384px.py
  - Name: swin-large_in21k-pre-3rdparty_in1k
    Metadata:
      FLOPs: 34040000000
      Parameters: 196530000
    In Collection: Swin-Transformer
    Results:
      - Dataset: ImageNet-1k
        Metrics:
          Top 1 Accuracy: 86.24
          Top 5 Accuracy: 97.88
        Task: Image Classification
    Weights: https://download.openmmlab.com/mmclassification/v0/swin-transformer/convert/swin_large_patch4_window7_224_22kto1k-5f0996db.pth
    Converted From:
      Weights: https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_large_patch4_window7_224_22kto1k.pth
      Code: https://github.com/microsoft/Swin-Transformer/blob/777f6c66604bb5579086c4447efe3620344d95a9/models/swin_transformer.py#L458
    Config: configs/swin_transformer/swin-large_16xb64_in1k.py
  - Name: swin-large_in21k-pre-3rdparty_in1k-384
    Metadata:
      FLOPs: 100040000000
      Parameters: 196740000
    In Collection: Swin-Transformer
    Results:
      - Dataset: ImageNet-1k
        Metrics:
          Top 1 Accuracy: 87.25
          Top 5 Accuracy: 98.25
        Task: Image Classification
    Weights: https://download.openmmlab.com/mmclassification/v0/swin-transformer/convert/swin_large_patch4_window12_384_22kto1k-0a40944b.pth
    Converted From:
      Weights: https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_large_patch4_window12_384_22kto1k.pth
      Code: https://github.com/microsoft/Swin-Transformer/blob/777f6c66604bb5579086c4447efe3620344d95a9/models/swin_transformer.py#L458
    Config: configs/swin_transformer/swin-large_16xb64_in1k-384px.py
  - Name: swin-large_8xb8_cub-384px
    Metadata:
      FLOPs: 100040000000
      Parameters: 195510000
    In Collection: Swin-Transformer
    Results:
      - Dataset: CUB-200-2011
        Metrics:
          Top 1 Accuracy: 91.87
        Task: Image Classification
    Pretrain: https://download.openmmlab.com/mmclassification/v0/swin-transformer/convert/swin-large_3rdparty_in21k-384px.pth
    Weights: https://download.openmmlab.com/mmclassification/v0/swin-transformer/swin-large_8xb8_cub_384px_20220307-1bbaee6a.pth
    Config: configs/swin_transformer/swin-large_8xb8_cub-384px.py