add part code

configs/cae/metafile.yml

+Collections:
+  - Name: CAE
+    Metadata:
+      Training Data: ImageNet-1k
+      Training Techniques:
+        - AdamW
+      Training Resources: 8x A100-80G GPUs
+      Architecture:
+        - ViT
+    Paper:
+      Title: Context Autoencoder for Self-Supervised Representation Learning
+      URL: https://arxiv.org/abs/2202.03026
+    README: configs/cae/README.md
+
+Models:
+  - Name: cae_beit-base-p16_8xb256-amp-coslr-300e_in1k
+    Metadata:
+      Epochs: 300
+      Batch Size: 2048
+      FLOPs: 17581976064
+      Parameters: 288429952
+      Training Data: ImageNet-1k
+    In Collection: CAE
+    Results: null
+    Weights: https://download.openmmlab.com/mmselfsup/1.x/cae/cae_vit-base-p16_8xb256-amp-coslr-300e_in1k/cae_vit-base-p16_8xb256-amp-coslr-300e_in1k_20221230-808170f3.pth
+    Config: configs/cae/cae_beit-base-p16_8xb256-amp-coslr-300e_in1k.py
+    Downstream:
+      - beit-base-p16_cae-pre_8xb128-coslr-100e_in1k
+  - Name: beit-base-p16_cae-pre_8xb128-coslr-100e_in1k
+    Metadata:
+      Epochs: 100
+      Batch Size: 1024
+      FLOPs: 17581219584
+      Parameters: 86682280
+      Training Data: ImageNet-1k
+    In Collection: CAE
+    Results:
+      - Task: Image Classification
+        Dataset: ImageNet-1k
+        Metrics:
+          Top 1 Accuracy: 83.2
+    Weights: https://download.openmmlab.com/mmselfsup/1.x/cae/cae_vit-base-p16_16xb128-fp16-coslr-300e_in1k/vit-base-p16_ft-8xb128-coslr-100e-rpe_in1k/vit-base-p16_ft-8xb128-coslr-100e-rpe_in1k_20220825-f3d234cd.pth
+    Config: configs/cae/benchmarks/beit-base-p16_8xb128-coslr-100e_in1k.py

configs/chinese_clip/README.md

+# ChineseCLIP
+
+> [Chinese CLIP: Contrastive Vision-Language Pretraining in Chinese](https://arxiv.org/abs/2211.01335)
+
+<!-- [ALGORITHM] -->
+
+## Abstract
+
+The tremendous success of CLIP (Radford et al., 2021) has promoted the research and application of contrastive learning for vision-language pretraining. In this work, we construct a large-scale dataset of image-text pairs in Chinese, where most data are retrieved from publicly available datasets, and we pretrain Chinese CLIP models on the new dataset. We develop 5 Chinese CLIP models of multiple sizes, spanning from 77 to 958 million parameters. Furthermore, we propose a two-stage pretraining method, where the model is first trained with the image encoder frozen and then trained with all parameters being optimized, to achieve enhanced model performance. Our comprehensive experiments demonstrate that Chinese CLIP can achieve the state-of-the-art performance on MUGE, Flickr30K-CN, and COCO-CN in the setups of zero-shot learning and finetuning, and it is able to achieve competitive performance in zero-shot image classification based on the evaluation on the ELEVATER benchmark (Li et al., 2022). We have released our codes, models, and demos in https://github.com/OFA-Sys/Chinese-CLIP
+
+<div align=center>
+<img src="https://github.com/open-mmlab/mmpretrain/assets/36138628/4d05e51f-d834-4ef5-bbf0-0e2f80fea461" width="80%"/>
+</div>
+
+## How to use it?
+
+<!-- [TABS-BEGIN] -->
+
+**Use the model for zero-shot classification**
+
+```python
+from mmpretrain import ImageClassificationInferencer
+
+inferencer = ImageClassificationInferencer(
+    'cn-clip_resnet50_zeroshot-cls_cifar100',
+    pretrained=True,
+    classes=['鸟', '狗', '猫', '蛇'],
+    text_prototype=['鸟', '狗', '猫', '蛇'],
+)
+
+prediction = inferencer('./demo/bird.JPEG')[0]
+print('Results:', prediction['pred_class'])
+```
+
+**Train/Test Command**
+
+Prepare your dataset according to the [docs](https://mmpretrain.readthedocs.io/en/latest/user_guides/dataset_prepare.html#prepare-dataset).
+
+Test:
+
+```shell
+python tools/test.py configs/chinese_clip/cn-clip_resnet50_zeroshot-cls_cifar100.py https://download.openmmlab.com/mmpretrain/v1.0/chinese_clip/cn-clip_resnet50_3rdparty_20230519-6a2b3eb2.pth
+```
+
+<!-- [TABS-END] -->
+
+## Models and results
+
+### Image Classification on CIFAR100
+
+| Model                                           | Params (M) | Top-1 (%) |                          Config                          |                                    Download                                    |
+| :---------------------------------------------- | :--------: | :-------: | :------------------------------------------------------: | :----------------------------------------------------------------------------: |
+| `cn-clip_resnet50_zeroshot-cls_cifar100`\*      |   77.00    |   40.70   |   [config](cn-clip_resnet50_zeroshot-cls_cifar100.py)    | [model](https://download.openmmlab.com/mmpretrain/v1.0/chinese_clip/cn-clip_resnet50_3rdparty_20230519-6a2b3eb2.pth) |
+| `cn-clip_vit-base-p16_zeroshot-cls_cifar100`\*  |   188.00   |   64.50   | [config](cn-clip_vit-base-p16_zeroshot-cls_cifar100.py)  | [model](https://download.openmmlab.com/mmpretrain/v1.0/chinese_clip/cn-clip_vit-base-p16_3rdparty_20230519-37fbc59e.pth) |
+| `cn-clip_vit-large-p14_zeroshot-cls_cifar100`\* |   406.00   |   74.80   | [config](cn-clip_vit-large-p14_zeroshot-cls_cifar100.py) | [model](https://download.openmmlab.com/mmpretrain/v1.0/chinese_clip/cn-clip_vit-large-p14_3rdparty_20230519-3f844503.pth) |
+| `cn-clip_vit-huge-p14_zeroshot-cls_cifar100`\*  |   958.00   |   79.10   | [config](cn-clip_vit-huge-p14_zeroshot-cls_cifar100.py)  | [model](https://download.openmmlab.com/mmpretrain/v1.0/chinese_clip/cn-clip_vit-huge-p14_3rdparty_20230519-e4f49b00.pth) |
+
+*Models with * are converted from the [official repo](https://github.com/OFA-Sys/Chinese-CLIP). The config files of these models are only for inference. We haven't reproduce the training results.*
+
+## Citation
+
+```bibtex
+@article{chinese-clip,
+  title={Chinese CLIP: Contrastive Vision-Language Pretraining in Chinese},
+  author={Yang, An and Pan, Junshu and Lin, Junyang and Men, Rui and Zhang, Yichang and Zhou, Jingren and Zhou, Chang},
+  journal={arXiv preprint arXiv:2211.01335},
+  year={2022}
+}
+```

configs/chinese_clip/cn-clip_resnet50_zeroshot-cls_cifar100.py

+_base_ = '../_base_/default_runtime.py'
+
+# data settings
+data_preprocessor = dict(
+    type='MultiModalDataPreprocessor',
+    mean=[0.48145466 * 255, 0.4578275 * 255, 0.40821073 * 255],
+    std=[0.26862954 * 255, 0.26130258 * 255, 0.27577711 * 255],
+    to_rgb=False,
+)
+
+test_pipeline = [
+    dict(type='Resize', scale=(224, 224), interpolation='bicubic'),
+    dict(
+        type='PackInputs',
+        meta_keys=['image_id', 'scale_factor'],
+    ),
+]
+
+train_dataloader = None
+test_dataloader = dict(
+    batch_size=32,
+    num_workers=8,
+    dataset=dict(
+        type='CIFAR100',
+        data_root='data/cifar100',
+        split='test',
+        pipeline=test_pipeline),
+    sampler=dict(type='DefaultSampler', shuffle=False),
+)
+test_evaluator = dict(type='Accuracy', topk=(1, ))
+
+# schedule settings
+train_cfg = None
+val_cfg = None
+test_cfg = dict()
+
+# model settings
+model = dict(
+    type='ChineseCLIP',
+    vision_backbone=dict(
+        type='ModifiedResNet',
+        depth=50,
+        base_channels=64,
+        input_size=224,
+        num_attn_heads=32,
+        output_dim=1024,
+    ),
+    text_backbone=dict(
+        type='BertModelCN',
+        config=dict(
+            vocab_size=21128,
+            pad_token_id=0,
+            add_type_embeddings=True,
+            attention_probs_dropout_prob=0.1,
+            hidden_act='gelu',
+            hidden_dropout_prob=0.1,
+            hidden_size=768,
+            initializer_range=0.02,
+            intermediate_size=3072,
+            max_position_embeddings=512,
+            num_attention_heads=12,
+            num_hidden_layers=3,
+            type_vocab_size=2,
+            layer_norm_eps=1e-12)),
+    tokenizer=dict(
+        type='FullTokenizer',
+        vocab_file=  # noqa
+        'https://download.openmmlab.com/mmpretrain/v1.0/chinese_clip/vocab.txt'
+    ),
+    proj_dim=1024,
+    text_prototype='cifar100',
+)

configs/chinese_clip/cn-clip_vit-base-p16_zeroshot-cls_cifar100.py

+_base_ = '../_base_/default_runtime.py'
+
+# data settings
+data_preprocessor = dict(
+    type='MultiModalDataPreprocessor',
+    mean=[0.48145466 * 255, 0.4578275 * 255, 0.40821073 * 255],
+    std=[0.26862954 * 255, 0.26130258 * 255, 0.27577711 * 255],
+    to_rgb=False,
+)
+
+test_pipeline = [
+    dict(type='Resize', scale=(224, 224), interpolation='bicubic'),
+    dict(
+        type='PackInputs',
+        algorithm_keys=['text'],
+        meta_keys=['image_id', 'scale_factor'],
+    ),
+]
+
+train_dataloader = None
+test_dataloader = dict(
+    batch_size=32,
+    num_workers=8,
+    dataset=dict(
+        type='CIFAR100',
+        data_root='data/cifar100',
+        split='test',
+        pipeline=test_pipeline),
+    sampler=dict(type='DefaultSampler', shuffle=False),
+)
+test_evaluator = dict(type='Accuracy', topk=(1, ))
+
+# schedule settings
+train_cfg = None
+val_cfg = None
+test_cfg = dict()
+
+# model settings
+model = dict(
+    type='ChineseCLIP',
+    vision_backbone=dict(
+        type='VisionTransformer',
+        arch='base',
+        img_size=224,
+        patch_size=16,
+        norm_cfg=dict(type='LN', eps=1e-5),
+        final_norm=True,
+        layer_cfgs=dict(act_cfg=dict(type='QuickGELU')),
+        pre_norm=True,
+        out_type='cls_token',
+    ),
+    text_backbone=dict(
+        type='BertModelCN',
+        config=dict(
+            vocab_size=21128,
+            pad_token_id=0,
+            add_type_embeddings=True,
+            attention_probs_dropout_prob=0.1,
+            hidden_act='gelu',
+            hidden_dropout_prob=0.1,
+            hidden_size=768,
+            initializer_range=0.02,
+            intermediate_size=3072,
+            max_position_embeddings=512,
+            num_attention_heads=12,
+            num_hidden_layers=12,
+            type_vocab_size=2,
+            layer_norm_eps=1e-12)),
+    tokenizer=dict(
+        type='FullTokenizer',
+        vocab_file=  # noqa
+        'https://download.openmmlab.com/mmpretrain/v1.0/chinese_clip/vocab.txt'
+    ),
+    proj_dim=512,
+    text_prototype='cifar100',
+)

configs/chinese_clip/cn-clip_vit-huge-p14_zeroshot-cls_cifar100.py

+_base_ = '../_base_/default_runtime.py'
+
+# data settings
+data_preprocessor = dict(
+    type='MultiModalDataPreprocessor',
+    mean=[0.48145466 * 255, 0.4578275 * 255, 0.40821073 * 255],
+    std=[0.26862954 * 255, 0.26130258 * 255, 0.27577711 * 255],
+    to_rgb=False,
+)
+
+test_pipeline = [
+    dict(type='Resize', scale=(224, 224), interpolation='bicubic'),
+    dict(
+        type='PackInputs',
+        meta_keys=['image_id', 'scale_factor'],
+    ),
+]
+
+train_dataloader = None
+test_dataloader = dict(
+    batch_size=32,
+    num_workers=8,
+    dataset=dict(
+        type='CIFAR100',
+        data_root='data/cifar100',
+        split='test',
+        pipeline=test_pipeline),
+    sampler=dict(type='DefaultSampler', shuffle=False),
+)
+test_evaluator = dict(type='Accuracy', topk=(1, ))
+
+# schedule settings
+train_cfg = None
+val_cfg = None
+test_cfg = dict()
+
+# model settings
+model = dict(
+    type='ChineseCLIP',
+    vision_backbone=dict(
+        type='VisionTransformer',
+        arch='huge',
+        img_size=224,
+        patch_size=14,
+        norm_cfg=dict(type='LN', eps=1e-5),
+        final_norm=True,
+        layer_cfgs=dict(act_cfg=dict(type='QuickGELU')),
+        pre_norm=True,
+        out_type='cls_token',
+    ),
+    text_backbone=dict(
+        type='BertModelCN',
+        config=dict(
+            vocab_size=21128,
+            pad_token_id=0,
+            add_type_embeddings=True,
+            attention_probs_dropout_prob=0.1,
+            hidden_act='gelu',
+            hidden_dropout_prob=0.1,
+            hidden_size=1024,
+            initializer_range=0.02,
+            intermediate_size=4096,
+            max_position_embeddings=512,
+            num_attention_heads=16,
+            num_hidden_layers=24,
+            type_vocab_size=2,
+            layer_norm_eps=1e-12)),
+    tokenizer=dict(
+        type='FullTokenizer',
+        vocab_file=  # noqa
+        'https://download.openmmlab.com/mmpretrain/v1.0/chinese_clip/vocab.txt'
+    ),
+    proj_dim=1024,
+    text_prototype='cifar100',
+)

configs/chinese_clip/cn-clip_vit-large-p14_zeroshot-cls_cifar100.py

+_base_ = '../_base_/default_runtime.py'
+
+# data settings
+data_preprocessor = dict(
+    type='MultiModalDataPreprocessor',
+    mean=[0.48145466 * 255, 0.4578275 * 255, 0.40821073 * 255],
+    std=[0.26862954 * 255, 0.26130258 * 255, 0.27577711 * 255],
+    to_rgb=False,
+)
+
+test_pipeline = [
+    dict(type='Resize', scale=(224, 224), interpolation='bicubic'),
+    dict(
+        type='PackInputs',
+        meta_keys=['image_id', 'scale_factor'],
+    ),
+]
+
+train_dataloader = None
+test_dataloader = dict(
+    batch_size=32,
+    num_workers=8,
+    dataset=dict(
+        type='CIFAR100',
+        data_root='data/cifar100',
+        split='test',
+        pipeline=test_pipeline),
+    sampler=dict(type='DefaultSampler', shuffle=False),
+)
+test_evaluator = dict(type='Accuracy', topk=(1, ))
+
+# schedule settings
+train_cfg = None
+val_cfg = None
+test_cfg = dict()
+
+# model settings
+model = dict(
+    type='ChineseCLIP',
+    vision_backbone=dict(
+        type='VisionTransformer',
+        arch='large',
+        img_size=224,
+        patch_size=14,
+        norm_cfg=dict(type='LN', eps=1e-5),
+        final_norm=True,
+        layer_cfgs=dict(act_cfg=dict(type='QuickGELU')),
+        pre_norm=True,
+        out_type='cls_token',
+    ),
+    text_backbone=dict(
+        type='BertModelCN',
+        config=dict(
+            vocab_size=21128,
+            pad_token_id=0,
+            add_type_embeddings=True,
+            attention_probs_dropout_prob=0.1,
+            hidden_act='gelu',
+            hidden_dropout_prob=0.1,
+            hidden_size=768,
+            initializer_range=0.02,
+            intermediate_size=3072,
+            max_position_embeddings=512,
+            num_attention_heads=12,
+            num_hidden_layers=12,
+            type_vocab_size=2,
+            layer_norm_eps=1e-12)),
+    tokenizer=dict(
+        type='FullTokenizer',
+        vocab_file=  # noqa
+        'https://download.openmmlab.com/mmpretrain/v1.0/chinese_clip/vocab.txt'
+    ),
+    proj_dim=768,
+    text_prototype='cifar100',
+)

configs/chinese_clip/metafile.yml

+Collections:
+  - Name: ChineseCLIP
+    Metadata:
+      Training Data:
+        - LAION-5B
+        - WuKong
+        - VisualGenome
+        - MSCOCO
+      Architecture:
+        - Transformer
+    Paper:
+      Title: 'Chinese CLIP: Contrastive Vision-Language Pretraining in Chinese'
+      URL: https://arxiv.org/abs/2211.01335
+    README: configs/chinese_clip/README.md
+
+Models:
+  - Name: cn-clip_resnet50_zeroshot-cls_cifar100
+    Metadata:
+      FLOPs: null
+      Parameters: 77000000
+    In Collection: ChineseCLIP
+    Results:
+      - Task: Image Classification
+        Dataset: CIFAR100
+        Metrics:
+          Top 1 Accuracy: 40.7
+    Weights: https://download.openmmlab.com/mmpretrain/v1.0/chinese_clip/cn-clip_resnet50_3rdparty_20230519-6a2b3eb2.pth
+    Config: configs/chinese_clip/cn-clip_resnet50_zeroshot-cls_cifar100.py
+    Converted From:
+      Weights: https://clip-cn-beijing.oss-cn-beijing.aliyuncs.com/checkpoints/clip_cn_rn50.pt
+      Code: https://github.com/OFA-Sys/Chinese-CLIP
+
+  - Name: cn-clip_vit-base-p16_zeroshot-cls_cifar100
+    Metadata:
+      FLOPs: null
+      Parameters: 188000000
+    In Collection: ChineseCLIP
+    Results:
+      - Task: Image Classification
+        Dataset: CIFAR100
+        Metrics:
+          Top 1 Accuracy: 64.5
+    Weights: https://download.openmmlab.com/mmpretrain/v1.0/chinese_clip/cn-clip_vit-base-p16_3rdparty_20230519-37fbc59e.pth
+    Config: configs/chinese_clip/cn-clip_vit-base-p16_zeroshot-cls_cifar100.py
+    Converted From:
+      Weights: https://clip-cn-beijing.oss-cn-beijing.aliyuncs.com/checkpoints/clip_cn_vit-b-16.pt
+      Code: https://github.com/OFA-Sys/Chinese-CLIP
+
+  - Name: cn-clip_vit-large-p14_zeroshot-cls_cifar100
+    Metadata:
+      FLOPs: null
+      Parameters: 406000000
+    In Collection: ChineseCLIP
+    Results:
+      - Task: Image Classification
+        Dataset: CIFAR100
+        Metrics:
+          Top 1 Accuracy: 74.8
+    Weights: https://download.openmmlab.com/mmpretrain/v1.0/chinese_clip/cn-clip_vit-large-p14_3rdparty_20230519-3f844503.pth
+    Config: configs/chinese_clip/cn-clip_vit-large-p14_zeroshot-cls_cifar100.py
+    Converted From:
+      Weights: https://clip-cn-beijing.oss-cn-beijing.aliyuncs.com/checkpoints/clip_cn_vit-l-14.pt
+      Code: https://github.com/OFA-Sys/Chinese-CLIP
+
+  - Name: cn-clip_vit-huge-p14_zeroshot-cls_cifar100
+    Metadata:
+      FLOPs: null
+      Parameters: 958000000
+    In Collection: ChineseCLIP
+    Results:
+      - Task: Image Classification
+        Dataset: CIFAR100
+        Metrics:
+          Top 1 Accuracy: 79.1
+    Weights: https://download.openmmlab.com/mmpretrain/v1.0/chinese_clip/cn-clip_vit-huge-p14_3rdparty_20230519-e4f49b00.pth
+    Config: configs/chinese_clip/cn-clip_vit-huge-p14_zeroshot-cls_cifar100.py
+    Converted From:
+      Weights: https://clip-cn-beijing.oss-cn-beijing.aliyuncs.com/checkpoints/clip_cn_vit-h-14.pt
+      Code: https://github.com/OFA-Sys/Chinese-CLIP

configs/clip/README.md

+# CLIP
+
+> [Learning Transferable Visual Models From Natural Language Supervision](https://arxiv.org/abs/2103.00020)
+
+<!-- [ALGORITHM] -->
+
+## Abstract
+
+State-of-the-art computer vision systems are trained to predict a fixed set of predetermined object categories. This restricted form of supervision limits their generality and usability since additional labeled data is needed to specify any other visual concept. Learning directly from raw text about images is a promising alternative which leverages a much broader source of supervision. We demonstrate that the simple pre-training task of predicting which caption goes with which image is an efficient and scalable way to learn SOTA image representations from scratch on a dataset of 400 million (image, text) pairs collected from the internet. After pre-training, natural language is used to reference learned visual concepts (or describe new ones) enabling zero-shot transfer of the model to downstream tasks. We study the performance of this approach by benchmarking on over 30 different existing computer vision datasets, spanning tasks such as OCR, action recognition in videos, geo-localization, and many types of fine-grained object classification. The model transfers non-trivially to most tasks and is often competitive with a fully supervised baseline without the need for any dataset specific training. For instance, we match the accuracy of the original ResNet-50 on ImageNet zero-shot without needing to use any of the 1.28 million training examples it was trained on. We release our code and pre-trained model weights at this https URL.
+
+<div align=center>
+<img src="https://raw.githubusercontent.com/Scarecrow0/figures_cache/main/clip_main_fig.png" width="100%"/>
+</div>
+
+## How to use it?
+
+<!-- [TABS-BEGIN] -->
+
+**Predict image**
+
+```python
+from mmpretrain import inference_model
+
+predict = inference_model('vit-base-p32_clip-laion2b-in12k-pre_3rdparty_in1k', 'demo/bird.JPEG')
+print(predict['pred_class'])
+print(predict['pred_score'])
+```
+
+**Use the model**
+
+```python
+import torch
+from mmpretrain import get_model
+
+model = get_model('vit-base-p32_clip-laion2b-in12k-pre_3rdparty_in1k', pretrained=True)
+inputs = torch.rand(1, 3, 224, 224)
+out = model(inputs)
+print(type(out))
+# To extract features.
+feats = model.extract_feat(inputs)
+print(type(feats))
+```
+
+**Test Command**
+
+Prepare your dataset according to the [docs](https://mmpretrain.readthedocs.io/en/latest/user_guides/dataset_prepare.html#prepare-dataset).
+
+Test:
+
+```shell
+python tools/test.py configs/clip/vit-base-p32_pt-64xb64_in1k.py https://download.openmmlab.com/mmclassification/v0/clip/clip-vit-base-p32_laion2b-in12k-pre_3rdparty_in1k_20221220-b384e830.pth
+```
+
+<!-- [TABS-END] -->
+
+## Models and results
+
+### Image Classification on ImageNet-1k
+
+| Model                                        |         Pretrain          | Params (M) | Flops (G) | Top-1 (%) | Top-5 (%) |                     Config                     |                     Download                     |
+| :------------------------------------------- | :-----------------------: | :--------: | :-------: | :-------: | :-------: | :--------------------------------------------: | :----------------------------------------------: |
+| `vit-base-p32_clip-laion2b-in12k-pre_3rdparty_in1k`\* | CLIP LAION2B ImageNet-12k |   88.22    |   4.36    |   83.06   |   96.49   |    [config](vit-base-p32_pt-64xb64_in1k.py)    | [model](https://download.openmmlab.com/mmclassification/v0/clip/clip-vit-base-p32_laion2b-in12k-pre_3rdparty_in1k_20221220-b384e830.pth) |
+| `vit-base-p32_clip-laion2b-pre_3rdparty_in1k`\* |       CLIP LAION2B        |   88.22    |   4.36    |   82.46   |   96.12   |    [config](vit-base-p32_pt-64xb64_in1k.py)    | [model](https://download.openmmlab.com/mmclassification/v0/clip/clip-vit-base-p32_laion2b-pre_3rdparty_in1k_20221220-194df57f.pth) |
+| `vit-base-p32_clip-openai-pre_3rdparty_in1k`\* |        CLIP OPENAI        |   88.22    |   4.36    |   81.77   |   95.89   |    [config](vit-base-p32_pt-64xb64_in1k.py)    | [model](https://download.openmmlab.com/mmclassification/v0/clip/clip-vit-base-p32_openai-pre_3rdparty_in1k_20221220-a0182ba9.pth) |
+| `vit-base-p32_clip-laion2b-in12k-pre_3rdparty_in1k-384px`\* | CLIP LAION2B ImageNet-12k |   88.22    |   12.66   |   85.39   |   97.67   | [config](vit-base-p32_pt-64xb64_in1k-384px.py) | [model](https://download.openmmlab.com/mmclassification/v0/clip/clip-vit-base-p32_laion2b-in12k-pre_3rdparty_in1k-384px_20221220-c7757552.pth) |
+| `vit-base-p32_clip-openai-in12k-pre_3rdparty_in1k-384px`\* | CLIP OPENAI ImageNet-12k  |   88.22    |   12.66   |   85.13   |   97.42   | [config](vit-base-p32_pt-64xb64_in1k-384px.py) | [model](https://download.openmmlab.com/mmclassification/v0/clip/clip-vit-base-p32_openai-in12k-pre_3rdparty_in1k-384px_20221220-dc2e49ea.pth) |
+| `vit-base-p16_clip-laion2b-in12k-pre_3rdparty_in1k`\* | CLIP LAION2B ImageNet-12k |   86.57    |   16.86   |   86.02   |   97.76   |    [config](vit-base-p16_pt-64xb64_in1k.py)    | [model](https://download.openmmlab.com/mmclassification/v0/clip/clip-vit-base-p16_laion2b-in12k-pre_3rdparty_in1k_20221220-a5e31f8c.pth) |
+| `vit-base-p16_clip-laion2b-pre_3rdparty_in1k`\* |       CLIP LAION2B        |   86.57    |   16.86   |   85.49   |   97.59   |    [config](vit-base-p16_pt-64xb64_in1k.py)    | [model](https://download.openmmlab.com/mmclassification/v0/clip/clip-vit-base-p16_laion2b-pre_3rdparty_in1k_20221220-5e24ff58.pth) |
+| `vit-base-p16_clip-openai-in12k-pre_3rdparty_in1k`\* | CLIP OPENAI ImageNet-12k  |   86.57    |   16.86   |   85.99   |   97.72   |    [config](vit-base-p16_pt-64xb64_in1k.py)    | [model](https://download.openmmlab.com/mmclassification/v0/clip/clip-vit-base-p16_openai-in12k-pre_3rdparty_in1k_20221220-90d930a8.pth) |
+| `vit-base-p16_clip-openai-pre_3rdparty_in1k`\* |        CLIP OPENAI        |   86.57    |   16.86   |   85.30   |   97.50   |    [config](vit-base-p16_pt-64xb64_in1k.py)    | [model](https://download.openmmlab.com/mmclassification/v0/clip/clip-vit-base-p16_openai-pre_3rdparty_in1k_20221220-c7d9c899.pth) |
+| `vit-base-p32_clip-laion2b-in12k-pre_3rdparty_in1k-448px`\* | CLIP LAION2B ImageNet-12k |   88.22    |   17.20   |   85.76   |   97.63   | [config](vit-base-p32_pt-64xb64_in1k-448px.py) | [model](https://download.openmmlab.com/mmclassification/v0/clip/clip-vit-base-p32_laion2b-in12k-pre_3rdparty_in1k-448px_20221220-ca404a7d.pth) |
+| `vit-base-p16_clip-laion2b-in12k-pre_3rdparty_in1k-384px`\* | CLIP LAION2B ImageNet-12k |   86.57    |   49.37   |   87.17   |   98.02   | [config](vit-base-p16_pt-64xb64_in1k-384px.py) | [model](https://download.openmmlab.com/mmclassification/v0/clip/clip-vit-base-p16_laion2b-in12k-pre_3rdparty_in1k-384px_20221220-84ed0cc0.pth) |
+| `vit-base-p16_clip-laion2b-pre_3rdparty_in1k-384px`\* |       CLIP LAION2B        |   86.57    |   49.37   |   86.52   |   97.97   | [config](vit-base-p16_pt-64xb64_in1k-384px.py) | [model](https://download.openmmlab.com/mmclassification/v0/clip/clip-vit-base-p16_laion2b-pre_3rdparty_in1k-384px_20221220-558ed826.pth) |
+| `vit-base-p16_clip-openai-in12k-pre_3rdparty_in1k-384px`\* | CLIP OPENAI ImageNet-12k  |   86.57    |   49.37   |   86.87   |   98.05   | [config](vit-base-p16_pt-64xb64_in1k-384px.py) | [model](https://download.openmmlab.com/mmclassification/v0/clip/clip-vit-base-p16_openai-in12k-pre_3rdparty_in1k-384px_20221220-8df86b74.pth) |
+| `vit-base-p16_clip-openai-pre_3rdparty_in1k-384px`\* |        CLIP OPENAI        |   86.57    |   49.37   |   86.25   |   97.90   | [config](vit-base-p16_pt-64xb64_in1k-384px.py) | [model](https://download.openmmlab.com/mmclassification/v0/clip/clip-vit-base-p16_openai-pre_3rdparty_in1k-384px_20221220-eb012e87.pth) |
+
+*Models with * are converted from the [timm](https://github.com/rwightman/pytorch-image-models). The config files of these models are only for inference. We haven't reproduce the training results.*
+
+## Citation
+
+```bibtex
+@InProceedings{pmlr-v139-radford21a,
+title = {Learning Transferable Visual Models From Natural Language Supervision},
+author = {Radford, Alec and Kim, Jong Wook and Hallacy, Chris and Ramesh, Aditya and Goh, Gabriel and Agarwal, Sandhini and Sastry, Girish and Askell, Amanda and Mishkin, Pamela and Clark, Jack and Krueger, Gretchen and Sutskever, Ilya},
+booktitle = {Proceedings of the 38th International Conference on Machine Learning},
+year = {2021},
+series = {Proceedings of Machine Learning Research},
+publisher = {PMLR},
+}
+```

configs/clip/clip_vit-base-p16_zeroshot-cls_cifar100.py

+_base_ = '../_base_/default_runtime.py'
+
+# data settings
+data_preprocessor = dict(
+    type='MultiModalDataPreprocessor',
+    mean=[0.48145466 * 255, 0.4578275 * 255, 0.40821073 * 255],
+    std=[0.26862954 * 255, 0.26130258 * 255, 0.27577711 * 255],
+    to_rgb=False,
+)
+
+test_pipeline = [
+    dict(type='Resize', scale=(224, 224), interpolation='bicubic'),
+    dict(
+        type='PackInputs',
+        algorithm_keys=['text'],
+        meta_keys=['image_id', 'scale_factor'],
+    ),
+]
+
+train_dataloader = None
+test_dataloader = dict(
+    batch_size=32,
+    num_workers=8,
+    dataset=dict(
+        type='CIFAR100',
+        data_root='data/cifar100',
+        split='test',
+        pipeline=test_pipeline),
+    sampler=dict(type='DefaultSampler', shuffle=False),
+)
+test_evaluator = dict(type='Accuracy', topk=(1, 5))
+
+# schedule settings
+train_cfg = None
+val_cfg = None
+test_cfg = dict()
+
+# model settings
+model = dict(
+    type='CLIPZeroShot',
+    vision_backbone=dict(
+        type='VisionTransformer',
+        arch='base',
+        img_size=224,
+        patch_size=16,
+        drop_rate=0.,
+        layer_cfgs=dict(act_cfg=dict(type='QuickGELU')),
+        pre_norm=True,
+    ),
+    projection=dict(type='CLIPProjection', in_channels=768, out_channels=512),
+    text_backbone=dict(
+        type='CLIPTransformer',
+        width=512,
+        layers=12,
+        heads=8,
+        attn_mask=True,
+    ),
+    tokenizer=dict(
+        type='AutoTokenizer',
+        name_or_path='openai/clip-vit-base-patch16',
+        use_fast=False),
+    vocab_size=49408,
+    transformer_width=512,
+    proj_dim=512,
+    text_prototype='cifar100',
+    text_prompt='openai_cifar100',
+    context_length=77,
+)

configs/clip/clip_vit-base-p16_zeroshot-cls_in1k.py

+_base_ = '../_base_/default_runtime.py'
+
+# data settings
+data_preprocessor = dict(
+    type='MultiModalDataPreprocessor',
+    mean=[0.48145466 * 255, 0.4578275 * 255, 0.40821073 * 255],
+    std=[0.26862954 * 255, 0.26130258 * 255, 0.27577711 * 255],
+    to_rgb=True,
+)
+
+test_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='Resize', scale=(224, 224), interpolation='bicubic'),
+    dict(
+        type='PackInputs',
+        algorithm_keys=['text'],
+        meta_keys=['image_id', 'scale_factor'],
+    ),
+]
+
+train_dataloader = None
+test_dataloader = dict(
+    batch_size=32,
+    num_workers=8,
+    dataset=dict(
+        type='ImageNet',
+        data_root='data/imagenet',
+        split='val',
+        pipeline=test_pipeline),
+    sampler=dict(type='DefaultSampler', shuffle=False),
+)
+test_evaluator = dict(type='Accuracy', topk=(1, 5))
+
+# schedule settings
+train_cfg = None
+val_cfg = None
+test_cfg = dict()
+
+# model settings
+model = dict(
+    type='CLIPZeroShot',
+    vision_backbone=dict(
+        type='VisionTransformer',
+        arch='base',
+        img_size=224,
+        patch_size=16,
+        drop_rate=0.,
+        layer_cfgs=dict(act_cfg=dict(type='QuickGELU')),
+        pre_norm=True,
+    ),
+    projection=dict(type='CLIPProjection', in_channels=768, out_channels=512),
+    text_backbone=dict(
+        type='CLIPTransformer',
+        width=512,
+        layers=12,
+        heads=8,
+        attn_mask=True,
+    ),
+    tokenizer=dict(
+        type='AutoTokenizer',
+        name_or_path='openai/clip-vit-base-patch16',
+        use_fast=False),
+    vocab_size=49408,
+    transformer_width=512,
+    proj_dim=512,
+    text_prototype='imagenet',
+    text_prompt='openai_imagenet_sub',  # openai_imagenet, openai_imagenet_sub
+    context_length=77,
+)

configs/clip/clip_vit-large-p14_zeroshot-cls_cifar100.py

+_base_ = '../_base_/default_runtime.py'
+
+# data settings
+data_preprocessor = dict(
+    type='MultiModalDataPreprocessor',
+    mean=[0.48145466 * 255, 0.4578275 * 255, 0.40821073 * 255],
+    std=[0.26862954 * 255, 0.26130258 * 255, 0.27577711 * 255],
+    to_rgb=False,
+)
+
+test_pipeline = [
+    dict(type='Resize', scale=(224, 224), interpolation='bicubic'),
+    dict(
+        type='PackInputs',
+        algorithm_keys=['text'],
+        meta_keys=['image_id', 'scale_factor'],
+    ),
+]
+
+train_dataloader = None
+test_dataloader = dict(
+    batch_size=32,
+    num_workers=8,
+    dataset=dict(
+        type='CIFAR100',
+        data_root='data/cifar100',
+        split='test',
+        pipeline=test_pipeline),
+    sampler=dict(type='DefaultSampler', shuffle=False),
+)
+test_evaluator = dict(type='Accuracy', topk=(1, 5))
+
+# schedule settings
+train_cfg = None
+val_cfg = None
+test_cfg = dict()
+
+# model settings
+model = dict(
+    type='CLIPZeroShot',
+    vision_backbone=dict(
+        type='VisionTransformer',
+        arch='large',
+        img_size=224,
+        patch_size=14,
+        drop_rate=0.,
+        layer_cfgs=dict(act_cfg=dict(type='QuickGELU')),
+        pre_norm=True,
+    ),
+    projection=dict(type='CLIPProjection', in_channels=1024, out_channels=768),
+    text_backbone=dict(
+        type='CLIPTransformer',
+        width=768,
+        layers=12,
+        heads=12,
+        attn_mask=True,
+    ),
+    tokenizer=dict(
+        type='AutoTokenizer',
+        name_or_path='openai/clip-vit-large-patch14',
+        use_fast=False),
+    vocab_size=49408,
+    transformer_width=768,
+    proj_dim=768,
+    text_prototype='cifar100',
+    text_prompt='openai_cifar100',
+    context_length=77,
+)

configs/clip/clip_vit-large-p14_zeroshot-cls_in1k.py

+_base_ = '../_base_/default_runtime.py'
+
+# data settings
+data_preprocessor = dict(
+    type='MultiModalDataPreprocessor',
+    mean=[0.48145466 * 255, 0.4578275 * 255, 0.40821073 * 255],
+    std=[0.26862954 * 255, 0.26130258 * 255, 0.27577711 * 255],
+    to_rgb=True,
+)
+
+test_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='Resize', scale=(224, 224), interpolation='bicubic'),
+    dict(
+        type='PackInputs',
+        algorithm_keys=['text'],
+        meta_keys=['image_id', 'scale_factor'],
+    ),
+]
+
+train_dataloader = None
+test_dataloader = dict(
+    batch_size=32,
+    num_workers=8,
+    dataset=dict(
+        type='ImageNet',
+        data_root='data/imagenet',
+        split='val',
+        pipeline=test_pipeline),
+    sampler=dict(type='DefaultSampler', shuffle=False),
+)
+test_evaluator = dict(type='Accuracy', topk=(1, 5))
+
+# schedule settings
+train_cfg = None
+val_cfg = None
+test_cfg = dict()
+
+# model settings
+model = dict(
+    type='CLIPZeroShot',
+    vision_backbone=dict(
+        type='VisionTransformer',
+        arch='large',
+        img_size=224,
+        patch_size=14,
+        drop_rate=0.,
+        layer_cfgs=dict(act_cfg=dict(type='QuickGELU')),
+        pre_norm=True,
+    ),
+    projection=dict(type='CLIPProjection', in_channels=1024, out_channels=768),
+    text_backbone=dict(
+        type='CLIPTransformer',
+        width=768,
+        layers=12,
+        heads=12,
+        attn_mask=True,
+    ),
+    tokenizer=dict(
+        type='AutoTokenizer',
+        name_or_path='openai/clip-vit-large-patch14',
+        use_fast=False),
+    vocab_size=49408,
+    transformer_width=768,
+    proj_dim=768,
+    text_prototype='imagenet',
+    text_prompt='openai_imagenet_sub',  # openai_imagenet, openai_imagenet_sub
+    context_length=77,
+)

configs/clip/metafile.yml

+Collections:
+  - Name: CLIP
+    Metadata:
+      Architecture:
+        - Attention Dropout
+        - Convolution
+        - Dense Connections
+        - Dropout
+        - GELU
+        - Layer Normalization
+        - Multi-Head Attention
+        - Scaled Dot-Product Attention
+        - Tanh Activation
+    Paper:
+      Title: Learning Transferable Visual Models From Natural Language Supervision
+      URL: https://arxiv.org/abs/2103.00020
+    README: configs/clip/README.md
+    Code:
+      URL: https://github.com/open-mmlab/mmpretrain/blob/main/mmpretrain/models/backbones/vision_transformer.py
+      Version: v1.0.0
+
+Models:
+  - Name: vit-base-p32_clip-openai-pre_3rdparty_in1k
+    Metadata:
+      FLOPs: 4364335104
+      Parameters: 88225000
+      Training Data:
+        - OpenAI
+        - ImageNet-1k
+    In Collection: CLIP
+    Results:
+      - Dataset: ImageNet-1k
+        Metrics:
+          Top 1 Accuracy: 81.77
+          Top 5 Accuracy: 95.89
+        Task: Image Classification
+    Weights: https://download.openmmlab.com/mmclassification/v0/clip/clip-vit-base-p32_openai-pre_3rdparty_in1k_20221220-a0182ba9.pth
+    Config: configs/clip/vit-base-p32_pt-64xb64_in1k.py
+    Converted From:
+      Code: https://github.com/rwightman/pytorch-image-models
+      Weights: https://huggingface.co/timm/vit_base_patch32_clip_224.openai_ft_in1k
+  - Name: vit-base-p32_clip-laion2b-pre_3rdparty_in1k
+    Metadata:
+      FLOPs: 4364335104
+      Parameters: 88225000
+      Training Data:
+        - LAION-2B
+        - ImageNet-1k
+    In Collection: CLIP
+    Results:
+      - Dataset: ImageNet-1k
+        Metrics:
+          Top 1 Accuracy: 82.46
+          Top 5 Accuracy: 96.12
+        Task: Image Classification
+    Weights: https://download.openmmlab.com/mmclassification/v0/clip/clip-vit-base-p32_laion2b-pre_3rdparty_in1k_20221220-194df57f.pth
+    Config: configs/clip/vit-base-p32_pt-64xb64_in1k.py
+    Converted From:
+      Code: https://github.com/rwightman/pytorch-image-models
+      Weights: https://huggingface.co/timm/vit_base_patch32_clip_224.laion2b_ft_in1k
+  - Name: vit-base-p32_clip-laion2b-in12k-pre_3rdparty_in1k
+    Metadata:
+      FLOPs: 4364335104
+      Parameters: 88225000
+      Training Data:
+        - LAION-2B
+        - ImageNet-12k
+        - ImageNet-1k
+    In Collection: CLIP
+    Results:
+      - Dataset: ImageNet-1k
+        Metrics:
+          Top 1 Accuracy: 83.06
+          Top 5 Accuracy: 96.49
+        Task: Image Classification
+    Weights: https://download.openmmlab.com/mmclassification/v0/clip/clip-vit-base-p32_laion2b-in12k-pre_3rdparty_in1k_20221220-b384e830.pth
+    Config: configs/clip/vit-base-p32_pt-64xb64_in1k.py
+    Converted From:
+      Code: https://github.com/rwightman/pytorch-image-models
+      Weights: https://huggingface.co/timm/vit_base_patch32_clip_224.laion2b_ft_in12k_in1k
+  - Name: vit-base-p32_clip-openai-in12k-pre_3rdparty_in1k-384px
+    Metadata:
+      FLOPs: 12661054464
+      Parameters: 88225000
+      Training Data:
+        - OpenAI
+        - ImageNet-12k
+        - ImageNet-1k
+    In Collection: CLIP
+    Results:
+      - Dataset: ImageNet-1k
+        Metrics:
+          Top 1 Accuracy: 85.13
+          Top 5 Accuracy: 97.42
+        Task: Image Classification
+    Weights: https://download.openmmlab.com/mmclassification/v0/clip/clip-vit-base-p32_openai-in12k-pre_3rdparty_in1k-384px_20221220-dc2e49ea.pth
+    Config: configs/clip/vit-base-p32_pt-64xb64_in1k-384px.py
+    Converted From:
+      Code: https://github.com/rwightman/pytorch-image-models
+      Weights: https://huggingface.co/timm/vit_base_patch32_clip_384.openai_ft_in12k_in1k
+  - Name: vit-base-p32_clip-laion2b-in12k-pre_3rdparty_in1k-384px
+    Metadata:
+      FLOPs: 12661054464
+      Parameters: 88225000
+      Training Data:
+        - LAION-2B
+        - ImageNet-12k
+        - ImageNet-1k
+    In Collection: CLIP
+    Results:
+      - Dataset: ImageNet-1k
+        Metrics:
+          Top 1 Accuracy: 85.39
+          Top 5 Accuracy: 97.67
+        Task: Image Classification
+    Weights: https://download.openmmlab.com/mmclassification/v0/clip/clip-vit-base-p32_laion2b-in12k-pre_3rdparty_in1k-384px_20221220-c7757552.pth
+    Config: configs/clip/vit-base-p32_pt-64xb64_in1k-384px.py
+    Converted From:
+      Code: https://github.com/rwightman/pytorch-image-models
+      Weights: https://huggingface.co/timm/vit_base_patch32_clip_384.laion2b_ft_in12k_in1k
+  - Name: vit-base-p16_clip-openai-pre_3rdparty_in1k
+    Metadata:
+      FLOPs: 16855600128
+      Parameters: 86568424
+      Training Data:
+        - OpenAI
+        - ImageNet-1k
+    In Collection: CLIP
+    Results:
+      - Dataset: ImageNet-1k
+        Metrics:
+          Top 1 Accuracy: 85.3
+          Top 5 Accuracy: 97.5
+        Task: Image Classification
+    Weights: https://download.openmmlab.com/mmclassification/v0/clip/clip-vit-base-p16_openai-pre_3rdparty_in1k_20221220-c7d9c899.pth
+    Config: configs/clip/vit-base-p16_pt-64xb64_in1k.py
+    Converted From:
+      Code: https://github.com/rwightman/pytorch-image-models
+      Weights: https://huggingface.co/timm/vit_base_patch16_clip_224.openai_ft_in1k
+  - Name: vit-base-p16_clip-laion2b-pre_3rdparty_in1k
+    Metadata:
+      FLOPs: 16855600128
+      Parameters: 86568424
+      Training Data:
+        - LAION-2B
+        - ImageNet-1k
+    In Collection: CLIP
+    Results:
+      - Dataset: ImageNet-1k
+        Metrics:
+          Top 1 Accuracy: 85.49
+          Top 5 Accuracy: 97.59
+        Task: Image Classification
+    Weights: https://download.openmmlab.com/mmclassification/v0/clip/clip-vit-base-p16_laion2b-pre_3rdparty_in1k_20221220-5e24ff58.pth
+    Config: configs/clip/vit-base-p16_pt-64xb64_in1k.py
+    Converted From:
+      Code: https://github.com/rwightman/pytorch-image-models
+      Weights: https://huggingface.co/timm/vit_base_patch16_clip_224.laion2b_ft_in1k
+  - Name: vit-base-p16_clip-openai-in12k-pre_3rdparty_in1k
+    Metadata:
+      FLOPs: 16855600128
+      Parameters: 86568424
+      Training Data:
+        - OpenAI
+        - ImageNet-12k
+        - ImageNet-1k
+    In Collection: CLIP
+    Results:
+      - Dataset: ImageNet-1k
+        Metrics:
+          Top 1 Accuracy: 85.99
+          Top 5 Accuracy: 97.72
+        Task: Image Classification
+    Weights: https://download.openmmlab.com/mmclassification/v0/clip/clip-vit-base-p16_openai-in12k-pre_3rdparty_in1k_20221220-90d930a8.pth
+    Config: configs/clip/vit-base-p16_pt-64xb64_in1k.py
+    Converted From:
+      Code: https://github.com/rwightman/pytorch-image-models
+      Weights: https://huggingface.co/timm/vit_base_patch16_clip_224.openai_ft_in12k_in1k
+  - Name: vit-base-p16_clip-laion2b-in12k-pre_3rdparty_in1k
+    Metadata:
+      FLOPs: 16855600128
+      Parameters: 86568424
+      Training Data:
+        - LAION-2B
+        - ImageNet-12k
+        - ImageNet-1k
+    In Collection: CLIP
+    Results:
+      - Dataset: ImageNet-1k
+        Metrics:
+          Top 1 Accuracy: 86.02
+          Top 5 Accuracy: 97.76
+        Task: Image Classification
+    Weights: https://download.openmmlab.com/mmclassification/v0/clip/clip-vit-base-p16_laion2b-in12k-pre_3rdparty_in1k_20221220-a5e31f8c.pth
+    Config: configs/clip/vit-base-p16_pt-64xb64_in1k.py
+    Converted From:
+      Code: https://github.com/rwightman/pytorch-image-models
+      Weights: https://huggingface.co/timm/vit_base_patch16_clip_224.laion2b_ft_in12k_in1k
+  - Name: vit-base-p32_clip-laion2b-in12k-pre_3rdparty_in1k-448px
+    Metadata:
+      FLOPs: 17202416640
+      Parameters: 88225000
+      Training Data:
+        - LAION-2B
+        - ImageNet-12k
+        - ImageNet-1k
+    In Collection: CLIP
+    Results:
+      - Dataset: ImageNet-1k
+        Metrics:
+          Top 1 Accuracy: 85.76
+          Top 5 Accuracy: 97.63
+        Task: Image Classification
+    Weights: https://download.openmmlab.com/mmclassification/v0/clip/clip-vit-base-p32_laion2b-in12k-pre_3rdparty_in1k-448px_20221220-ca404a7d.pth
+    Config: configs/clip/vit-base-p32_pt-64xb64_in1k-448px.py
+    Converted From:
+      Code: https://github.com/rwightman/pytorch-image-models
+      Weights: https://huggingface.co/timm/vit_base_patch32_clip_448.laion2b_ft_in12k_in1k
+  - Name: vit-base-p16_clip-openai-pre_3rdparty_in1k-384px
+    Metadata:
+      FLOPs: 49370078208
+      Parameters: 86568424
+      Training Data:
+        - OpenAI
+        - ImageNet-1k
+    In Collection: CLIP
+    Results:
+      - Dataset: ImageNet-1k
+        Metrics:
+          Top 1 Accuracy: 86.25
+          Top 5 Accuracy: 97.9
+        Task: Image Classification
+    Weights: https://download.openmmlab.com/mmclassification/v0/clip/clip-vit-base-p16_openai-pre_3rdparty_in1k-384px_20221220-eb012e87.pth
+    Config: configs/clip/vit-base-p16_pt-64xb64_in1k-384px.py
+    Converted From:
+      Code: https://github.com/rwightman/pytorch-image-models
+      Weights: https://huggingface.co/timm/vit_base_patch16_clip_384.openai_ft_in1k
+  - Name: vit-base-p16_clip-laion2b-pre_3rdparty_in1k-384px
+    Metadata:
+      FLOPs: 49370078208
+      Parameters: 86568424
+      Training Data:
+        - LAION-2B
+        - ImageNet-1k
+    In Collection: CLIP
+    Results:
+      - Dataset: ImageNet-1k
+        Metrics:
+          Top 1 Accuracy: 86.52
+          Top 5 Accuracy: 97.97
+        Task: Image Classification
+    Weights: https://download.openmmlab.com/mmclassification/v0/clip/clip-vit-base-p16_laion2b-pre_3rdparty_in1k-384px_20221220-558ed826.pth
+    Config: configs/clip/vit-base-p16_pt-64xb64_in1k-384px.py
+    Converted From:
+      Code: https://github.com/rwightman/pytorch-image-models
+      Weights: https://huggingface.co/timm/vit_base_patch16_clip_384.laion2b_ft_in1k
+  - Name: vit-base-p16_clip-openai-in12k-pre_3rdparty_in1k-384px
+    Metadata:
+      FLOPs: 49370078208
+      Parameters: 86568424
+      Training Data:
+        - OpenAI
+        - ImageNet-12k
+        - ImageNet-1k
+    In Collection: CLIP
+    Results:
+      - Dataset: ImageNet-1k
+        Metrics:
+          Top 1 Accuracy: 86.87
+          Top 5 Accuracy: 98.05
+        Task: Image Classification
+    Weights: https://download.openmmlab.com/mmclassification/v0/clip/clip-vit-base-p16_openai-in12k-pre_3rdparty_in1k-384px_20221220-8df86b74.pth
+    Config: configs/clip/vit-base-p16_pt-64xb64_in1k-384px.py
+    Converted From:
+      Code: https://github.com/rwightman/pytorch-image-models
+      Weights: https://huggingface.co/timm/vit_base_patch16_clip_384.openai_ft_in12k_in1k
+  - Name: vit-base-p16_clip-laion2b-in12k-pre_3rdparty_in1k-384px
+    Metadata:
+      FLOPs: 49370078208
+      Parameters: 86568424
+      Training Data:
+        - LAION-2B
+        - ImageNet-12k
+        - ImageNet-1k
+    In Collection: CLIP
+    Results:
+      - Dataset: ImageNet-1k
+        Metrics:
+          Top 1 Accuracy: 87.17
+          Top 5 Accuracy: 98.02
+        Task: Image Classification
+    Weights: https://download.openmmlab.com/mmclassification/v0/clip/clip-vit-base-p16_laion2b-in12k-pre_3rdparty_in1k-384px_20221220-84ed0cc0.pth
+    Config: configs/clip/vit-base-p16_pt-64xb64_in1k-384px.py
+    Converted From:
+      Code: https://github.com/rwightman/pytorch-image-models
+      Weights: https://huggingface.co/timm/vit_base_patch16_clip_384.laion2b_ft_in12k_in1k
+  - Name: vit-large-p14_clip-openai-pre_3rdparty
+    Metadata:
+      FLOPs: 59696580608
+      Parameters: 303302656
+      Training Data:
+        - OpenAI
+    In Collection: CLIP
+    Weights: https://download.openmmlab.com/mmclassification/v0/clip/vit-large-p14_clip-openai-pre_3rdparty_20230517-95e2af0b.pth
+    Config: configs/clip/vit-large-p14_headless.py
+    Converted From:
+      Code: https://github.com/mlfoundations/open_clip
+      Weights: https://openaipublic.azureedge.net/clip/models/b8cca3fd41ae0c99ba7e8951adf17d267cdb84cd88be6f7c2e0eca1737a03836/ViT-L-14.pt

configs/clip/vit-base-p16_pt-64xb64_in1k-384px.py

+_base_ = [
+    '../_base_/models/vit-base-p16.py',
+    '../_base_/datasets/imagenet_bs64_pil_resize.py',
+    '../_base_/schedules/imagenet_bs4096_AdamW.py',
+    '../_base_/default_runtime.py'
+]
+
+# model setting
+model = dict(backbone=dict(pre_norm=True))
+
+# data settings
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='RandomResizedCrop',
+        scale=384,
+        backend='pillow',
+        interpolation='bicubic'),
+    dict(type='RandomFlip', prob=0.5, direction='horizontal'),
+    dict(type='PackInputs'),
+]
+
+test_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='ResizeEdge',
+        scale=384,
+        edge='short',
+        backend='pillow',
+        interpolation='bicubic'),
+    dict(type='CenterCrop', crop_size=384),
+    dict(type='PackInputs'),
+]
+
+train_dataloader = dict(dataset=dict(pipeline=train_pipeline))
+val_dataloader = dict(dataset=dict(pipeline=test_pipeline))
+test_dataloader = dict(dataset=dict(pipeline=test_pipeline))
+
+# schedule setting
+optim_wrapper = dict(clip_grad=dict(max_norm=1.0))

configs/clip/vit-base-p16_pt-64xb64_in1k-448px.py

+_base_ = [
+    '../_base_/models/vit-base-p16.py',
+    '../_base_/datasets/imagenet_bs64_pil_resize.py',
+    '../_base_/schedules/imagenet_bs4096_AdamW.py',
+    '../_base_/default_runtime.py'
+]
+
+# model setting
+model = dict(backbone=dict(pre_norm=True))
+
+# data settings
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='RandomResizedCrop',
+        scale=448,
+        backend='pillow',
+        interpolation='bicubic'),
+    dict(type='RandomFlip', prob=0.5, direction='horizontal'),
+    dict(type='PackInputs'),
+]
+
+test_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='ResizeEdge',
+        scale=448,
+        edge='short',
+        backend='pillow',
+        interpolation='bicubic'),
+    dict(type='CenterCrop', crop_size=448),
+    dict(type='PackInputs'),
+]
+
+train_dataloader = dict(dataset=dict(pipeline=train_pipeline))
+val_dataloader = dict(dataset=dict(pipeline=test_pipeline))
+test_dataloader = dict(dataset=dict(pipeline=test_pipeline))
+
+# schedule setting
+optim_wrapper = dict(clip_grad=dict(max_norm=1.0))

configs/clip/vit-base-p16_pt-64xb64_in1k.py

+_base_ = [
+    '../_base_/models/vit-base-p16.py',
+    '../_base_/datasets/imagenet_bs64_pil_resize.py',
+    '../_base_/schedules/imagenet_bs4096_AdamW.py',
+    '../_base_/default_runtime.py'
+]
+
+# model setting
+model = dict(backbone=dict(pre_norm=True))
+
+# data settings
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='RandomResizedCrop',
+        scale=224,
+        backend='pillow',
+        interpolation='bicubic'),
+    dict(type='RandomFlip', prob=0.5, direction='horizontal'),
+    dict(type='PackInputs'),
+]
+
+test_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='ResizeEdge',
+        scale=224,
+        edge='short',
+        backend='pillow',
+        interpolation='bicubic'),
+    dict(type='CenterCrop', crop_size=224),
+    dict(type='PackInputs'),
+]
+
+train_dataloader = dict(dataset=dict(pipeline=train_pipeline))
+val_dataloader = dict(dataset=dict(pipeline=test_pipeline))
+test_dataloader = dict(dataset=dict(pipeline=test_pipeline))
+
+# schedule setting
+optim_wrapper = dict(clip_grad=dict(max_norm=1.0))

configs/clip/vit-base-p32_pt-64xb64_in1k-384px.py

+_base_ = [
+    '../_base_/models/vit-base-p32.py',
+    '../_base_/datasets/imagenet_bs64_pil_resize.py',
+    '../_base_/schedules/imagenet_bs4096_AdamW.py',
+    '../_base_/default_runtime.py'
+]
+
+# model setting
+model = dict(backbone=dict(pre_norm=True))
+
+# data settings
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='RandomResizedCrop',
+        scale=384,
+        backend='pillow',
+        interpolation='bicubic'),
+    dict(type='RandomFlip', prob=0.5, direction='horizontal'),
+    dict(type='PackInputs'),
+]
+
+test_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='ResizeEdge',
+        scale=384,
+        edge='short',
+        backend='pillow',
+        interpolation='bicubic'),
+    dict(type='CenterCrop', crop_size=384),
+    dict(type='PackInputs'),
+]
+
+train_dataloader = dict(dataset=dict(pipeline=train_pipeline))
+val_dataloader = dict(dataset=dict(pipeline=test_pipeline))
+test_dataloader = dict(dataset=dict(pipeline=test_pipeline))
+
+# schedule setting
+optim_wrapper = dict(clip_grad=dict(max_norm=1.0))

configs/clip/vit-base-p32_pt-64xb64_in1k-448px.py

+_base_ = [
+    '../_base_/models/vit-base-p32.py',
+    '../_base_/datasets/imagenet_bs64_pil_resize.py',
+    '../_base_/schedules/imagenet_bs4096_AdamW.py',
+    '../_base_/default_runtime.py'
+]
+
+# model setting
+model = dict(backbone=dict(pre_norm=True))
+
+# data settings
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='RandomResizedCrop',
+        scale=448,
+        backend='pillow',
+        interpolation='bicubic'),
+    dict(type='RandomFlip', prob=0.5, direction='horizontal'),
+    dict(type='PackInputs'),
+]
+
+test_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='ResizeEdge',
+        scale=448,
+        edge='short',
+        backend='pillow',
+        interpolation='bicubic'),
+    dict(type='CenterCrop', crop_size=448),
+    dict(type='PackInputs'),
+]
+
+train_dataloader = dict(dataset=dict(pipeline=train_pipeline))
+val_dataloader = dict(dataset=dict(pipeline=test_pipeline))
+test_dataloader = dict(dataset=dict(pipeline=test_pipeline))
+
+# schedule setting
+optim_wrapper = dict(clip_grad=dict(max_norm=1.0))

configs/clip/vit-base-p32_pt-64xb64_in1k.py

+_base_ = [
+    '../_base_/models/vit-base-p32.py',
+    '../_base_/datasets/imagenet_bs64_pil_resize.py',
+    '../_base_/schedules/imagenet_bs4096_AdamW.py',
+    '../_base_/default_runtime.py'
+]
+
+# model setting
+model = dict(backbone=dict(pre_norm=True))
+
+# data settings
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='RandomResizedCrop',
+        scale=224,
+        backend='pillow',
+        interpolation='bicubic'),
+    dict(type='RandomFlip', prob=0.5, direction='horizontal'),
+    dict(type='PackInputs'),
+]
+
+test_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='ResizeEdge',
+        scale=224,
+        edge='short',
+        backend='pillow',
+        interpolation='bicubic'),
+    dict(type='CenterCrop', crop_size=224),
+    dict(type='PackInputs'),
+]
+
+train_dataloader = dict(dataset=dict(pipeline=train_pipeline))
+val_dataloader = dict(dataset=dict(pipeline=test_pipeline))
+test_dataloader = dict(dataset=dict(pipeline=test_pipeline))
+
+# schedule setting
+optim_wrapper = dict(clip_grad=dict(max_norm=1.0))

configs/clip/vit-large-p14_headless.py

+_base_ = ['../_base_/default_runtime.py']
+
+# model settings
+model = dict(
+    type='ImageClassifier',
+    backbone=dict(
+        type='VisionTransformer',
+        arch='l',
+        img_size=224,
+        patch_size=16,
+        drop_rate=0.1,
+        pre_norm=True,
+    ),
+)
+
+test_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='ResizeEdge', scale=256, edge='short', backend='pillow'),
+    dict(type='CenterCrop', crop_size=224),
+    dict(type='PackInputs'),
+]
+
+test_dataloader = dict(
+    batch_size=64,
+    num_workers=5,
+    dataset=dict(
+        type='ImageNet',
+        data_root='data/imagenet',
+        ann_file='meta/val.txt',
+        data_prefix='val',
+        pipeline=test_pipeline),
+    sampler=dict(type='DefaultSampler', shuffle=False),
+)
+test_evaluator = None