Add Chinese-CLIP implementation (#20368)

* init chinese-clip model from clip

* init model tests and docs

* implement chinese-clip into hf

* implement chinese-clip into hf

* implement chinese-clip into hf

* implement chinese-clip into hf

* implement chinese-clip into hf

* update usecase example in model implementation

* fix codestyle

* fix model_type typo in readme

* add placeholder in doc

* add placeholder in doc

* update the init script

* update usecase

* fix codestyle

* update testcase

* update testcase

* update testcase

* update testcase

* update testcase

* update testcase

* update testcase

* update testcase

* update testcase

* update testcase

* update testcase

* update testcase

* forward the convert_rgb

* update testcase

* update testcase

* update testcase

* merge the recent update from clip about model_input_name property

* update the doc

* update the doc

* update the doc

* update the doc

* remove unused imports

* reformat code style

* update the doc

* fix isort style

* bypass a weird failed unit test which is unrelated with my PR

* update the doc

* implement independent vision config class

* implement independent vision model class

* fix refactor bug

* fix refactor bug

* fix refactor bug

* make style

* fix refactor bug

* make style

* fix refactor bug

* fix refactor bug

* make style

* fix refactor bug

* fix refactor bug

* doc-build restyle

* implement independent text config class

* implement independent text model class

* implement independent text model class

* make style

* make fix-copies

* fix refactor bug

* fix refactor bug

* fix refactor bug

* fix refactor bug

* fix refactor bug

* fix refactor bug

* fix refactor bug

* fix refactor bug

* fix refactor bug

* fix refactor bug

* make style

* update doc

* black and isort

* update doc

* Update src/transformers/models/chinese_clip/configuration_chinese_clip.py
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Update src/transformers/models/auto/tokenization_auto.py
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* modify the model type from chinese-clip to chinese_clip

* format the example comment of ChineseCLIPVisionConfig

* correct the copyright comment

* fix the tokenizer specification

* add copied from for loss function

* remove unused class

* update CHINESE_CLIP_TEXT_INPUTS_DOCSTRING

* update CHINESE_CLIP_INPUTS_DOCSTRING

* update doc

* update doc

* update code comment in config

* update copied from statement

* make style

* rename the doc file

* add copied statement

* remove unused attention_mask, causal_attention_mask in ChineseCLIPVisionEncoder

* remove ChineseCLIPTextPreTrainedModel

* fix bug

* fix bug

* fix bug

* update doc

* make style

* Update src/transformers/models/chinese_clip/configuration_chinese_clip.py
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Update src/transformers/models/chinese_clip/configuration_chinese_clip.py
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* update ChineseCLIPImageProcessor in image_processing_auto

* fix config_class of chinesecliptextmodel

* fix the test case

* update the docs

* remove the copied from comment for ChineseCLIPTextModel, since it has diverged from BertModel with customed config_class

* update the testcase

* final fix
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>

README.md

@@ -279,6 +279,7 @@ Current number of checkpoints: ![](https://img.shields.io/endpoint?url=https://h
 1. **[ByT5](https://huggingface.co/docs/transformers/model_doc/byt5)** (from Google Research) released with the paper [ByT5: Towards a token-free future with pre-trained byte-to-byte models](https://arxiv.org/abs/2105.13626) by Linting Xue, Aditya Barua, Noah Constant, Rami Al-Rfou, Sharan Narang, Mihir Kale, Adam Roberts, Colin Raffel.
 1. **[CamemBERT](https://huggingface.co/docs/transformers/model_doc/camembert)** (from Inria/Facebook/Sorbonne) released with the paper [CamemBERT: a Tasty French Language Model](https://arxiv.org/abs/1911.03894) by Louis Martin*, Benjamin Muller*, Pedro Javier Ortiz Suárez*, Yoann Dupont, Laurent Romary, Éric Villemonte de la Clergerie, Djamé Seddah and Benoît Sagot.
 1. **[CANINE](https://huggingface.co/docs/transformers/model_doc/canine)** (from Google Research) released with the paper [CANINE: Pre-training an Efficient Tokenization-Free Encoder for Language Representation](https://arxiv.org/abs/2103.06874) by Jonathan H. Clark, Dan Garrette, Iulia Turc, John Wieting.
+1. **[Chinese-CLIP](https://huggingface.co/docs/transformers/main/model_doc/chinese_clip)** (from OFA-Sys) released with the paper [Chinese CLIP: Contrastive Vision-Language Pretraining in Chinese](https://arxiv.org/abs/2211.01335) by An Yang, Junshu Pan, Junyang Lin, Rui Men, Yichang Zhang, Jingren Zhou, Chang Zhou.
 1. **[CLIP](https://huggingface.co/docs/transformers/model_doc/clip)** (from OpenAI) released with the paper [Learning Transferable Visual Models From Natural Language Supervision](https://arxiv.org/abs/2103.00020) by Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, Ilya Sutskever.
 1. **[CLIPSeg](https://huggingface.co/docs/transformers/main/model_doc/clipseg)** (from University of Göttingen) released with the paper [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003) by Timo Lüddecke and Alexander Ecker.
 1. **[CodeGen](https://huggingface.co/docs/transformers/model_doc/codegen)** (from Salesforce) released with the paper [A Conversational Paradigm for Program Synthesis](https://arxiv.org/abs/2203.13474) by Erik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu, Huan Wang, Yingbo Zhou, Silvio Savarese, Caiming Xiong.

README_es.md

@@ -279,6 +279,7 @@ Número actual de puntos de control: ![](https://img.shields.io/endpoint?url=htt
 1. **[ByT5](https://huggingface.co/docs/transformers/model_doc/byt5)** (from Google Research) released with the paper [ByT5: Towards a token-free future with pre-trained byte-to-byte models](https://arxiv.org/abs/2105.13626) by Linting Xue, Aditya Barua, Noah Constant, Rami Al-Rfou, Sharan Narang, Mihir Kale, Adam Roberts, Colin Raffel.
 1. **[CamemBERT](https://huggingface.co/docs/transformers/model_doc/camembert)** (from Inria/Facebook/Sorbonne) released with the paper [CamemBERT: a Tasty French Language Model](https://arxiv.org/abs/1911.03894) by Louis Martin*, Benjamin Muller*, Pedro Javier Ortiz Suárez*, Yoann Dupont, Laurent Romary, Éric Villemonte de la Clergerie, Djamé Seddah and Benoît Sagot.
 1. **[CANINE](https://huggingface.co/docs/transformers/model_doc/canine)** (from Google Research) released with the paper [CANINE: Pre-training an Efficient Tokenization-Free Encoder for Language Representation](https://arxiv.org/abs/2103.06874) by Jonathan H. Clark, Dan Garrette, Iulia Turc, John Wieting.
+1. **[Chinese-CLIP](https://huggingface.co/docs/transformers/main/model_doc/chinese_clip)** (from OFA-Sys) released with the paper [Chinese CLIP: Contrastive Vision-Language Pretraining in Chinese](https://arxiv.org/abs/2211.01335) by An Yang, Junshu Pan, Junyang Lin, Rui Men, Yichang Zhang, Jingren Zhou, Chang Zhou. 
 1. **[CLIP](https://huggingface.co/docs/transformers/model_doc/clip)** (from OpenAI) released with the paper [Learning Transferable Visual Models From Natural Language Supervision](https://arxiv.org/abs/2103.00020) by Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, Ilya Sutskever.
 1. **[CLIPSeg](https://huggingface.co/docs/transformers/main/model_doc/clipseg)** (from University of Göttingen) released with the paper [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003) by Timo Lüddecke and Alexander Ecker.
 1. **[CodeGen](https://huggingface.co/docs/transformers/model_doc/codegen)** (from Salesforce) released with the paper [A Conversational Paradigm for Program Synthesis](https://arxiv.org/abs/2203.13474) by Erik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu, Huan Wang, Yingbo Zhou, Silvio Savarese, Caiming Xiong.

README_ja.md

@@ -314,6 +314,7 @@ Flax、PyTorch、TensorFlowをcondaでインストールする方法は、それ
 1. **[ByT5](https://huggingface.co/docs/transformers/model_doc/byt5)** (from Google Research) released with the paper [ByT5: Towards a token-free future with pre-trained byte-to-byte models](https://arxiv.org/abs/2105.13626) by Linting Xue, Aditya Barua, Noah Constant, Rami Al-Rfou, Sharan Narang, Mihir Kale, Adam Roberts, Colin Raffel.
 1. **[CamemBERT](https://huggingface.co/docs/transformers/model_doc/camembert)** (from Inria/Facebook/Sorbonne) released with the paper [CamemBERT: a Tasty French Language Model](https://arxiv.org/abs/1911.03894) by Louis Martin*, Benjamin Muller*, Pedro Javier Ortiz Suárez*, Yoann Dupont, Laurent Romary, Éric Villemonte de la Clergerie, Djamé Seddah and Benoît Sagot.
 1. **[CANINE](https://huggingface.co/docs/transformers/model_doc/canine)** (from Google Research) released with the paper [CANINE: Pre-training an Efficient Tokenization-Free Encoder for Language Representation](https://arxiv.org/abs/2103.06874) by Jonathan H. Clark, Dan Garrette, Iulia Turc, John Wieting.
+1. **[Chinese-CLIP](https://huggingface.co/docs/transformers/main/model_doc/chinese_clip)** (from OFA-Sys) released with the paper [Chinese CLIP: Contrastive Vision-Language Pretraining in Chinese](https://arxiv.org/abs/2211.01335) by An Yang, Junshu Pan, Junyang Lin, Rui Men, Yichang Zhang, Jingren Zhou, Chang Zhou.
 1. **[CLIP](https://huggingface.co/docs/transformers/model_doc/clip)** (from OpenAI) released with the paper [Learning Transferable Visual Models From Natural Language Supervision](https://arxiv.org/abs/2103.00020) by Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, Ilya Sutskever.
 1. **[CLIPSeg](https://huggingface.co/docs/transformers/main/model_doc/clipseg)** (from University of Göttingen) released with the paper [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003) by Timo Lüddecke and Alexander Ecker.
 1. **[CodeGen](https://huggingface.co/docs/transformers/model_doc/codegen)** (from Salesforce) released with the paper [A Conversational Paradigm for Program Synthesis](https://arxiv.org/abs/2203.13474) by Erik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu, Huan Wang, Yingbo Zhou, Silvio Savarese, Caiming Xiong.

README_ko.md

@@ -229,6 +229,7 @@ Flax, PyTorch, TensorFlow 설치 페이지에서 이들을 conda로 설치하는
 1. **[ByT5](https://huggingface.co/docs/transformers/model_doc/byt5)** (from Google Research) released with the paper [ByT5: Towards a token-free future with pre-trained byte-to-byte models](https://arxiv.org/abs/2105.13626) by Linting Xue, Aditya Barua, Noah Constant, Rami Al-Rfou, Sharan Narang, Mihir Kale, Adam Roberts, Colin Raffel.
 1. **[CamemBERT](https://huggingface.co/docs/transformers/model_doc/camembert)** (from Inria/Facebook/Sorbonne) released with the paper [CamemBERT: a Tasty French Language Model](https://arxiv.org/abs/1911.03894) by Louis Martin*, Benjamin Muller*, Pedro Javier Ortiz Suárez*, Yoann Dupont, Laurent Romary, Éric Villemonte de la Clergerie, Djamé Seddah and Benoît Sagot.
 1. **[CANINE](https://huggingface.co/docs/transformers/model_doc/canine)** (from Google Research) released with the paper [CANINE: Pre-training an Efficient Tokenization-Free Encoder for Language Representation](https://arxiv.org/abs/2103.06874) by Jonathan H. Clark, Dan Garrette, Iulia Turc, John Wieting.
+1. **[Chinese-CLIP](https://huggingface.co/docs/transformers/main/model_doc/chinese_clip)** (from OFA-Sys) released with the paper [Chinese CLIP: Contrastive Vision-Language Pretraining in Chinese](https://arxiv.org/abs/2211.01335) by An Yang, Junshu Pan, Junyang Lin, Rui Men, Yichang Zhang, Jingren Zhou, Chang Zhou.
 1. **[CLIP](https://huggingface.co/docs/transformers/model_doc/clip)** (from OpenAI) released with the paper [Learning Transferable Visual Models From Natural Language Supervision](https://arxiv.org/abs/2103.00020) by Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, Ilya Sutskever.
 1. **[CLIPSeg](https://huggingface.co/docs/transformers/main/model_doc/clipseg)** (from University of Göttingen) released with the paper [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003) by Timo Lüddecke and Alexander Ecker.
 1. **[CodeGen](https://huggingface.co/docs/transformers/model_doc/codegen)** (from Salesforce) released with the paper [A Conversational Paradigm for Program Synthesis](https://arxiv.org/abs/2203.13474) by Erik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu, Huan Wang, Yingbo Zhou, Silvio Savarese, Caiming Xiong.

README_zh-hans.md

@@ -253,6 +253,7 @@ conda install -c huggingface transformers
 1. **[ByT5](https://huggingface.co/docs/transformers/model_doc/byt5)** (来自 Google Research) 伴随论文 [ByT5: Towards a token-free future with pre-trained byte-to-byte models](https://arxiv.org/abs/2105.13626) 由 Linting Xue, Aditya Barua, Noah Constant, Rami Al-Rfou, Sharan Narang, Mihir Kale, Adam Roberts, Colin Raffel 发布。
 1. **[CamemBERT](https://huggingface.co/docs/transformers/model_doc/camembert)** (来自 Inria/Facebook/Sorbonne) 伴随论文 [CamemBERT: a Tasty French Language Model](https://arxiv.org/abs/1911.03894) 由 Louis Martin*, Benjamin Muller*, Pedro Javier Ortiz Suárez*, Yoann Dupont, Laurent Romary, Éric Villemonte de la Clergerie, Djamé Seddah and Benoît Sagot 发布。
 1. **[CANINE](https://huggingface.co/docs/transformers/model_doc/canine)** (来自 Google Research) 伴随论文 [CANINE: Pre-training an Efficient Tokenization-Free Encoder for Language Representation](https://arxiv.org/abs/2103.06874) 由 Jonathan H. Clark, Dan Garrette, Iulia Turc, John Wieting 发布。
+1. **[Chinese-CLIP](https://huggingface.co/docs/transformers/main/model_doc/chinese_clip)** (来自 OFA-Sys) 伴随论文 [Chinese CLIP: Contrastive Vision-Language Pretraining in Chinese](https://arxiv.org/abs/2211.01335) 由 An Yang, Junshu Pan, Junyang Lin, Rui Men, Yichang Zhang, Jingren Zhou, Chang Zhou 发布。
 1. **[CLIP](https://huggingface.co/docs/transformers/model_doc/clip)** (来自 OpenAI) 伴随论文 [Learning Transferable Visual Models From Natural Language Supervision](https://arxiv.org/abs/2103.00020) 由 Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, Ilya Sutskever 发布。
 1. **[CLIPSeg](https://huggingface.co/docs/transformers/main/model_doc/clipseg)** (来自 University of Göttingen) 伴随论文 [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003) 由 Timo Lüddecke and Alexander Ecker 发布。
 1. **[CodeGen](https://huggingface.co/docs/transformers/model_doc/codegen)** (来自 Salesforce) 伴随论文 [A Conversational Paradigm for Program Synthesis](https://arxiv.org/abs/2203.13474) 由 Erik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu, Huan Wang, Yingbo Zhou, Silvio Savarese, Caiming Xiong 发布。

README_zh-hant.md

@@ -265,6 +265,7 @@ conda install -c huggingface transformers
 1. **[ByT5](https://huggingface.co/docs/transformers/model_doc/byt5)** (from Google Research) released with the paper [ByT5: Towards a token-free future with pre-trained byte-to-byte models](https://arxiv.org/abs/2105.13626) by Linting Xue, Aditya Barua, Noah Constant, Rami Al-Rfou, Sharan Narang, Mihir Kale, Adam Roberts, Colin Raffel.
 1. **[CamemBERT](https://huggingface.co/docs/transformers/model_doc/camembert)** (from Inria/Facebook/Sorbonne) released with the paper [CamemBERT: a Tasty French Language Model](https://arxiv.org/abs/1911.03894) by Louis Martin*, Benjamin Muller*, Pedro Javier Ortiz Suárez*, Yoann Dupont, Laurent Romary, Éric Villemonte de la Clergerie, Djamé Seddah and Benoît Sagot.
 1. **[CANINE](https://huggingface.co/docs/transformers/model_doc/canine)** (from Google Research) released with the paper [CANINE: Pre-training an Efficient Tokenization-Free Encoder for Language Representation](https://arxiv.org/abs/2103.06874) by Jonathan H. Clark, Dan Garrette, Iulia Turc, John Wieting.
+1. **[Chinese-CLIP](https://huggingface.co/docs/transformers/main/model_doc/chinese_clip)** (from OFA-Sys) released with the paper [Chinese CLIP: Contrastive Vision-Language Pretraining in Chinese](https://arxiv.org/abs/2211.01335) by An Yang, Junshu Pan, Junyang Lin, Rui Men, Yichang Zhang, Jingren Zhou, Chang Zhou.
 1. **[CLIP](https://huggingface.co/docs/transformers/model_doc/clip)** (from OpenAI) released with the paper [Learning Transferable Visual Models From Natural Language Supervision](https://arxiv.org/abs/2103.00020) by Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, Ilya Sutskever.
 1. **[CLIPSeg](https://huggingface.co/docs/transformers/main/model_doc/clipseg)** (from University of Göttingen) released with the paper [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003) by Timo Lüddecke and Alexander Ecker.
 1. **[CodeGen](https://huggingface.co/docs/transformers/model_doc/codegen)** (from Salesforce) released with the paper [A Conversational Paradigm for Program Synthesis](https://arxiv.org/abs/2203.13474) by Erik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu, Huan Wang, Yingbo Zhou, Silvio Savarese, Caiming Xiong.

docs/source/en/_toctree.yml

@@ -482,6 +482,8 @@
       title: Audio models
     - isExpanded: false
       sections:
+      - local: model_doc/chinese_clip
+        title: Chinese-CLIP
       - local: model_doc/clip
         title: CLIP
       - local: model_doc/clipseg

docs/source/en/index.mdx

@@ -67,6 +67,7 @@ The documentation is organized into five sections:
 1. **[ByT5](model_doc/byt5)** (from Google Research) released with the paper [ByT5: Towards a token-free future with pre-trained byte-to-byte models](https://arxiv.org/abs/2105.13626) by Linting Xue, Aditya Barua, Noah Constant, Rami Al-Rfou, Sharan Narang, Mihir Kale, Adam Roberts, Colin Raffel.
 1. **[CamemBERT](model_doc/camembert)** (from Inria/Facebook/Sorbonne) released with the paper [CamemBERT: a Tasty French Language Model](https://arxiv.org/abs/1911.03894) by Louis Martin*, Benjamin Muller*, Pedro Javier Ortiz Suárez*, Yoann Dupont, Laurent Romary, Éric Villemonte de la Clergerie, Djamé Seddah and Benoît Sagot.
 1. **[CANINE](model_doc/canine)** (from Google Research) released with the paper [CANINE: Pre-training an Efficient Tokenization-Free Encoder for Language Representation](https://arxiv.org/abs/2103.06874) by Jonathan H. Clark, Dan Garrette, Iulia Turc, John Wieting.
+1. **[Chinese-CLIP](model_doc/chinese_clip)** (from OFA-Sys) released with the paper [Chinese CLIP: Contrastive Vision-Language Pretraining in Chinese](https://arxiv.org/abs/2211.01335) by An Yang, Junshu Pan, Junyang Lin, Rui Men, Yichang Zhang, Jingren Zhou, Chang Zhou.
 1. **[CLIP](model_doc/clip)** (from OpenAI) released with the paper [Learning Transferable Visual Models From Natural Language Supervision](https://arxiv.org/abs/2103.00020) by Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, Ilya Sutskever.
 1. **[CLIPSeg](model_doc/clipseg)** (from University of Göttingen) released with the paper [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003) by Timo Lüddecke and Alexander Ecker.
 1. **[CodeGen](model_doc/codegen)** (from Salesforce) released with the paper [A Conversational Paradigm for Program Synthesis](https://arxiv.org/abs/2203.13474) by Erik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu, Huan Wang, Yingbo Zhou, Silvio Savarese, Caiming Xiong.
@@ -232,6 +233,7 @@ Flax), PyTorch, and/or TensorFlow.
 |             BLOOM             |       ❌       |       ✅       |       ✅        |         ❌         |      ❌      |
 |           CamemBERT           |       ✅       |       ✅       |       ✅        |         ✅         |      ❌      |
 |            CANINE             |       ✅       |       ❌       |       ✅        |         ❌         |      ❌      |
+|         Chinese-CLIP          |       ❌       |       ❌       |       ✅        |         ❌         |      ❌      |
 |             CLIP              |       ✅       |       ✅       |       ✅        |         ✅         |      ✅      |
 |            CLIPSeg            |       ❌       |       ❌       |       ✅        |         ❌         |      ❌      |
 |            CodeGen            |       ✅       |       ✅       |       ✅        |         ❌         |      ❌      |

docs/source/en/model_doc/chinese_clip.mdx

+<!--Copyright 2022 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+-->
+
+# Chinese-CLIP
+
+## Overview
+
+The Chinese-CLIP model was proposed in [Chinese CLIP: Contrastive Vision-Language Pretraining in Chinese](https://arxiv.org/abs/2211.01335) by An Yang, Junshu Pan, Junyang Lin, Rui Men, Yichang Zhang, Jingren Zhou, Chang Zhou.
+Chinese-CLIP is an implementation of CLIP (Radford et al., 2021) on a large-scale dataset of Chinese image-text pairs. It is capable of performing cross-modal retrieval and also playing as a vision backbone for vision tasks like zero-shot image classification, open-domain object detection, etc. The original Chinese-CLIP code is released [at this link](https://github.com/OFA-Sys/Chinese-CLIP).
+
+The abstract from the paper is the following:
+
+*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). Our codes, pretrained models, and demos have been released.*
+
+## Usage
+
+The code snippet below shows how to compute image & text features and similarities:
+
+```python
+>>> from PIL import Image
+>>> import requests
+>>> from transformers import ChineseCLIPProcessor, ChineseCLIPModel
+
+>>> model = ChineseCLIPModel.from_pretrained("OFA-Sys/chinese-clip-vit-base-patch16")
+>>> processor = ChineseCLIPProcessor.from_pretrained("OFA-Sys/chinese-clip-vit-base-patch16")
+
+>>> url = "https://clip-cn-beijing.oss-cn-beijing.aliyuncs.com/pokemon.jpeg"
+>>> image = Image.open(requests.get(url, stream=True).raw)
+>>> # Squirtle, Bulbasaur, Charmander, Pikachu in English
+>>> texts = ["杰尼龟", "妙蛙种子", "小火龙", "皮卡丘"]
+
+>>> # compute image feature
+>>> inputs = processor(images=image, return_tensors="pt")
+>>> image_features = model.get_image_features(**inputs)
+>>> image_features = image_features / image_features.norm(p=2, dim=-1, keepdim=True)  # normalize
+
+>>> # compute text features
+>>> inputs = processor(text=texts, padding=True, return_tensors="pt")
+>>> text_features = model.get_text_features(**inputs)
+>>> text_features = text_features / text_features.norm(p=2, dim=-1, keepdim=True)  # normalize
+
+>>> # compute image-text similarity scores
+>>> inputs = processor(text=texts, images=image, return_tensors="pt", padding=True)
+>>> outputs = model(**inputs)
+>>> logits_per_image = outputs.logits_per_image  # this is the image-text similarity score
+>>> probs = logits_per_image.softmax(dim=1)  # probs: [[1.2686e-03, 5.4499e-02, 6.7968e-04, 9.4355e-01]]
+```
+
+Currently, we release the following scales of pretrained Chinese-CLIP models at HF Model Hub:
+
+- [OFA-Sys/chinese-clip-vit-base-patch16](https://huggingface.co/OFA-Sys/chinese-clip-vit-base-patch16)
+- [OFA-Sys/chinese-clip-vit-large-patch14](https://huggingface.co/OFA-Sys/chinese-clip-vit-large-patch14)
+- [OFA-Sys/chinese-clip-vit-large-patch14-336px](https://huggingface.co/OFA-Sys/chinese-clip-vit-large-patch14-336px)
+- [OFA-Sys/chinese-clip-vit-huge-patch14](https://huggingface.co/OFA-Sys/chinese-clip-vit-huge-patch14)
+
+The Chinese-CLIP model was contributed by [OFA-Sys](https://huggingface.co/OFA-Sys). 
+
+## ChineseCLIPConfig
+
+[[autodoc]] ChineseCLIPConfig
+    - from_text_vision_configs
+
+## ChineseCLIPTextConfig
+
+[[autodoc]] ChineseCLIPTextConfig
+
+## ChineseCLIPVisionConfig
+
+[[autodoc]] ChineseCLIPVisionConfig
+
+## ChineseCLIPImageProcessor
+
+[[autodoc]] ChineseCLIPImageProcessor
+    - preprocess
+
+## ChineseCLIPFeatureExtractor
+
+[[autodoc]] ChineseCLIPFeatureExtractor
+
+## ChineseCLIPProcessor
+
+[[autodoc]] ChineseCLIPProcessor
+
+## ChineseCLIPModel
+
+[[autodoc]] ChineseCLIPModel
+    - forward
+    - get_text_features
+    - get_image_features
+
+## ChineseCLIPTextModel
+
+[[autodoc]] ChineseCLIPTextModel
+    - forward
+
+## ChineseCLIPVisionModel
+
+[[autodoc]] ChineseCLIPVisionModel
+    - forward
\ No newline at end of file

docs/source/en/serialization.mdx

@@ -55,6 +55,7 @@ Ready-made configurations include the following architectures:
 - BlenderbotSmall
 - BLOOM
 - CamemBERT
+- Chinese-CLIP
 - CLIP
 - CodeGen
 - Conditional DETR

src/transformers/__init__.py

@@ -171,6 +171,13 @@ _import_structure = {
     "models.byt5": ["ByT5Tokenizer"],
     "models.camembert": ["CAMEMBERT_PRETRAINED_CONFIG_ARCHIVE_MAP", "CamembertConfig"],
     "models.canine": ["CANINE_PRETRAINED_CONFIG_ARCHIVE_MAP", "CanineConfig", "CanineTokenizer"],
+    "models.chinese_clip": [
+        "CHINESE_CLIP_PRETRAINED_CONFIG_ARCHIVE_MAP",
+        "ChineseCLIPConfig",
+        "ChineseCLIPProcessor",
+        "ChineseCLIPTextConfig",
+        "ChineseCLIPVisionConfig",
+    ],
     "models.clip": [
         "CLIP_PRETRAINED_CONFIG_ARCHIVE_MAP",
         "CLIPConfig",
@@ -736,6 +743,7 @@ else:
     _import_structure["image_transforms"] = ["rescale", "resize", "to_pil_image"]
     _import_structure["image_utils"] = ["ImageFeatureExtractionMixin"]
     _import_structure["models.beit"].extend(["BeitFeatureExtractor", "BeitImageProcessor"])
+    _import_structure["models.chinese_clip"].extend(["ChineseCLIPFeatureExtractor", "ChineseCLIPImageProcessor"])
     _import_structure["models.clip"].extend(["CLIPFeatureExtractor", "CLIPImageProcessor"])
     _import_structure["models.conditional_detr"].extend(
         ["ConditionalDetrFeatureExtractor", "ConditionalDetrImageProcessor"]
@@ -1103,6 +1111,15 @@ else:
             "CLIPVisionModelWithProjection",
         ]
     )
+    _import_structure["models.chinese_clip"].extend(
+        [
+            "CHINESE_CLIP_PRETRAINED_MODEL_ARCHIVE_LIST",
+            "ChineseCLIPModel",
+            "ChineseCLIPPreTrainedModel",
+            "ChineseCLIPTextModel",
+            "ChineseCLIPVisionModel",
+        ]
+    )
     _import_structure["models.clipseg"].extend(
         [
             "CLIPSEG_PRETRAINED_MODEL_ARCHIVE_LIST",
@@ -3376,6 +3393,13 @@ if TYPE_CHECKING:
     from .models.byt5 import ByT5Tokenizer
     from .models.camembert import CAMEMBERT_PRETRAINED_CONFIG_ARCHIVE_MAP, CamembertConfig
     from .models.canine import CANINE_PRETRAINED_CONFIG_ARCHIVE_MAP, CanineConfig, CanineTokenizer
+    from .models.chinese_clip import (
+        CHINESE_CLIP_PRETRAINED_CONFIG_ARCHIVE_MAP,
+        ChineseCLIPConfig,
+        ChineseCLIPProcessor,
+        ChineseCLIPTextConfig,
+        ChineseCLIPVisionConfig,
+    )
     from .models.clip import (
         CLIP_PRETRAINED_CONFIG_ARCHIVE_MAP,
         CLIPConfig,
@@ -3874,6 +3898,7 @@ if TYPE_CHECKING:
         from .image_transforms import rescale, resize, to_pil_image
         from .image_utils import ImageFeatureExtractionMixin
         from .models.beit import BeitFeatureExtractor, BeitImageProcessor
+        from .models.chinese_clip import ChineseCLIPFeatureExtractor, ChineseCLIPImageProcessor
         from .models.clip import CLIPFeatureExtractor, CLIPImageProcessor
         from .models.conditional_detr import ConditionalDetrFeatureExtractor, ConditionalDetrImageProcessor
         from .models.convnext import ConvNextFeatureExtractor, ConvNextImageProcessor
@@ -4176,6 +4201,13 @@ if TYPE_CHECKING:
             CaninePreTrainedModel,
             load_tf_weights_in_canine,
         )
+        from .models.chinese_clip import (
+            CHINESE_CLIP_PRETRAINED_MODEL_ARCHIVE_LIST,
+            ChineseCLIPModel,
+            ChineseCLIPPreTrainedModel,
+            ChineseCLIPTextModel,
+            ChineseCLIPVisionModel,
+        )
         from .models.clip import (
             CLIP_PRETRAINED_MODEL_ARCHIVE_LIST,
             CLIPModel,

src/transformers/models/__init__.py

@@ -37,6 +37,7 @@ from . import (
     byt5,
     camembert,
     canine,
+    chinese_clip,
     clip,
     clipseg,
     codegen,

src/transformers/models/auto/configuration_auto.py

@@ -42,6 +42,7 @@ CONFIG_MAPPING_NAMES = OrderedDict(
         ("bloom", "BloomConfig"),
         ("camembert", "CamembertConfig"),
         ("canine", "CanineConfig"),
+        ("chinese_clip", "ChineseCLIPConfig"),
         ("clip", "CLIPConfig"),
         ("clipseg", "CLIPSegConfig"),
         ("codegen", "CodeGenConfig"),
@@ -192,6 +193,7 @@ CONFIG_ARCHIVE_MAP_MAPPING_NAMES = OrderedDict(
         ("bloom", "BLOOM_PRETRAINED_CONFIG_ARCHIVE_MAP"),
         ("camembert", "CAMEMBERT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
         ("canine", "CANINE_PRETRAINED_CONFIG_ARCHIVE_MAP"),
+        ("chinese_clip", "CHINESE_CLIP_PRETRAINED_CONFIG_ARCHIVE_MAP"),
         ("clip", "CLIP_PRETRAINED_CONFIG_ARCHIVE_MAP"),
         ("clipseg", "CLIPSEG_PRETRAINED_CONFIG_ARCHIVE_MAP"),
         ("codegen", "CODEGEN_PRETRAINED_CONFIG_ARCHIVE_MAP"),
@@ -334,6 +336,7 @@ MODEL_NAMES_MAPPING = OrderedDict(
         ("byt5", "ByT5"),
         ("camembert", "CamemBERT"),
         ("canine", "CANINE"),
+        ("chinese_clip", "Chinese-CLIP"),
         ("clip", "CLIP"),
         ("clipseg", "CLIPSeg"),
         ("codegen", "CodeGen"),

src/transformers/models/auto/feature_extraction_auto.py

@@ -39,6 +39,7 @@ FEATURE_EXTRACTOR_MAPPING_NAMES = OrderedDict(
     [
         ("audio-spectrogram-transformer", "ASTFeatureExtractor"),
         ("beit", "BeitFeatureExtractor"),
+        ("chinese_clip", "ChineseCLIPFeatureExtractor"),
         ("clip", "CLIPFeatureExtractor"),
         ("clipseg", "ViTFeatureExtractor"),
         ("conditional_detr", "ConditionalDetrFeatureExtractor"),

src/transformers/models/auto/image_processing_auto.py

@@ -38,6 +38,7 @@ logger = logging.get_logger(__name__)
 IMAGE_PROCESSOR_MAPPING_NAMES = OrderedDict(
     [
         ("beit", "BeitImageProcessor"),
+        ("chinese_clip", "ChineseCLIPImageProcessor"),
         ("clip", "CLIPImageProcessor"),
         ("clipseg", "ViTImageProcessor"),
         ("conditional_detr", "ConditionalDetrImageProcessor"),

src/transformers/models/auto/modeling_auto.py

@@ -41,6 +41,7 @@ MODEL_MAPPING_NAMES = OrderedDict(
         ("bloom", "BloomModel"),
         ("camembert", "CamembertModel"),
         ("canine", "CanineModel"),
+        ("chinese_clip", "ChineseCLIPModel"),
         ("clip", "CLIPModel"),
         ("clipseg", "CLIPSegModel"),
         ("codegen", "CodeGenModel"),
@@ -840,6 +841,7 @@ MODEL_FOR_AUDIO_XVECTOR_MAPPING_NAMES = OrderedDict(
 _MODEL_FOR_ZERO_SHOT_IMAGE_CLASSIFICATION_MAPPING_NAMES = OrderedDict(
     [
         # Model for Zero Shot Image Classification mapping
+        ("chinese_clip", "ChineseCLIPModel"),
         ("clip", "CLIPModel"),
         ("clipseg", "CLIPSegModel"),
     ]

src/transformers/models/auto/processing_auto.py

@@ -41,6 +41,7 @@ logger = logging.get_logger(__name__)
 
 PROCESSOR_MAPPING_NAMES = OrderedDict(
     [
+        ("chinese_clip", "ChineseCLIPProcessor"),
         ("clip", "CLIPProcessor"),
         ("clipseg", "CLIPSegProcessor"),
         ("flava", "FlavaProcessor"),

src/transformers/models/auto/tokenization_auto.py

@@ -86,6 +86,7 @@ else:
                 ),
             ),
             ("canine", ("CanineTokenizer", None)),
+            ("chinese_clip", ("BertTokenizer", "BertTokenizerFast" if is_tokenizers_available() else None)),
             (
                 "clip",
                 (

src/transformers/models/chinese_clip/__init__.py

+# flake8: noqa
+# There's no way to ignore "F401 '...' imported but unused" warnings in this
+# module, but to preserve other warnings. So, don't check this module at all.
+
+# Copyright 2022 The OFA-Sys Team Authors and The HuggingFace Team. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+from typing import TYPE_CHECKING
+
+from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available, is_vision_available
+
+
+_import_structure = {
+    "configuration_chinese_clip": [
+        "CHINESE_CLIP_PRETRAINED_CONFIG_ARCHIVE_MAP",
+        "ChineseCLIPConfig",
+        "ChineseCLIPOnnxConfig",
+        "ChineseCLIPTextConfig",
+        "ChineseCLIPVisionConfig",
+    ],
+    "processing_chinese_clip": ["ChineseCLIPProcessor"],
+}
+
+try:
+    if not is_vision_available():
+        raise OptionalDependencyNotAvailable()
+except OptionalDependencyNotAvailable:
+    pass
+else:
+    _import_structure["feature_extraction_chinese_clip"] = ["ChineseCLIPFeatureExtractor"]
+    _import_structure["image_processing_chinese_clip"] = ["ChineseCLIPImageProcessor"]
+
+try:
+    if not is_torch_available():
+        raise OptionalDependencyNotAvailable()
+except OptionalDependencyNotAvailable:
+    pass
+else:
+    _import_structure["modeling_chinese_clip"] = [
+        "CHINESE_CLIP_PRETRAINED_MODEL_ARCHIVE_LIST",
+        "ChineseCLIPModel",
+        "ChineseCLIPPreTrainedModel",
+        "ChineseCLIPTextModel",
+        "ChineseCLIPVisionModel",
+    ]
+
+if TYPE_CHECKING:
+    from .configuration_chinese_clip import (
+        CHINESE_CLIP_PRETRAINED_CONFIG_ARCHIVE_MAP,
+        ChineseCLIPConfig,
+        ChineseCLIPOnnxConfig,
+        ChineseCLIPTextConfig,
+        ChineseCLIPVisionConfig,
+    )
+    from .processing_chinese_clip import ChineseCLIPProcessor
+
+    try:
+        if not is_vision_available():
+            raise OptionalDependencyNotAvailable()
+    except OptionalDependencyNotAvailable:
+        pass
+    else:
+        from .feature_extraction_chinese_clip import ChineseCLIPFeatureExtractor, ChineseCLIPImageProcessor
+
+    try:
+        if not is_torch_available():
+            raise OptionalDependencyNotAvailable()
+    except OptionalDependencyNotAvailable:
+        pass
+    else:
+        from .modeling_chinese_clip import (
+            CHINESE_CLIP_PRETRAINED_MODEL_ARCHIVE_LIST,
+            ChineseCLIPModel,
+            ChineseCLIPPreTrainedModel,
+            ChineseCLIPTextModel,
+            ChineseCLIPVisionModel,
+        )
+
+else:
+    import sys
+
+    sys.modules[__name__] = _LazyModule(__name__, globals()["__file__"], _import_structure, module_spec=__spec__)

src/transformers/models/chinese_clip/configuration_chinese_clip.py

+# coding=utf-8
+# Copyright 2022 The OFA-Sys Team Authors and The HuggingFace Team. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+""" Chinese-CLIP model configuration"""
+
+import copy
+import os
+from collections import OrderedDict
+from typing import TYPE_CHECKING, Any, Mapping, Optional, Union
+
+
+if TYPE_CHECKING:
+    from ...processing_utils import ProcessorMixin
+    from ...utils import TensorType
+
+from ...configuration_utils import PretrainedConfig
+from ...onnx import OnnxConfig
+from ...utils import logging
+
+
+logger = logging.get_logger(__name__)
+
+CHINESE_CLIP_PRETRAINED_CONFIG_ARCHIVE_MAP = {
+    "OFA-Sys/chinese-clip-vit-base-patch16": (
+        "https://huggingface.co/OFA-Sys/chinese-clip-vit-base-patch16/resolve/main/config.json"
+    ),
+}
+
+
+class ChineseCLIPTextConfig(PretrainedConfig):
+    r"""
+    This is the configuration class to store the configuration of a [`ChineseCLIPModel`]. It is used to instantiate a
+    Chinese CLIP model according to the specified arguments, defining the model architecture. Instantiating a
+    configuration with the defaults will yield a similar configuration to that of the Chinese CLIP
+    [OFA-Sys/chinese-clip-vit-base-patch16](https:
+        //huggingface.co/OFA-Sys/chinese-clip-vit-base-patch16) architecture.
+
+    Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
+    documentation from [`PretrainedConfig`] for more information.
+
+
+    Args:
+        vocab_size (`int`, *optional*, defaults to 30522):
+            Vocabulary size of the CHINESE_CLIP model. Defines the number of different tokens that can be represented
+            by the `inputs_ids` passed when calling [`ChineseCLIPModel`].
+        hidden_size (`int`, *optional*, defaults to 768):
+            Dimensionality of the encoder layers and the pooler layer.
+        num_hidden_layers (`int`, *optional*, defaults to 12):
+            Number of hidden layers in the Transformer encoder.
+        num_attention_heads (`int`, *optional*, defaults to 12):
+            Number of attention heads for each attention layer in the Transformer encoder.
+        intermediate_size (`int`, *optional*, defaults to 3072):
+            Dimensionality of the "intermediate" (often named feed-forward) layer in the Transformer encoder.
+        hidden_act (`str` or `Callable`, *optional*, defaults to `"gelu"`):
+            The non-linear activation function (function or string) in the encoder and pooler. If string, `"gelu"`,
+            `"relu"`, `"silu"` and `"gelu_new"` are supported.
+        hidden_dropout_prob (`float`, *optional*, defaults to 0.1):
+            The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
+        attention_probs_dropout_prob (`float`, *optional*, defaults to 0.1):
+            The dropout ratio for the attention probabilities.
+        max_position_embeddings (`int`, *optional*, defaults to 512):
+            The maximum sequence length that this model might ever be used with. Typically set this to something large
+            just in case (e.g., 512 or 1024 or 2048).
+        type_vocab_size (`int`, *optional*, defaults to 2):
+            The vocabulary size of the `token_type_ids` passed when calling [`ChineseCLIPModel`].
+        initializer_range (`float`, *optional*, defaults to 0.02):
+            The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
+        layer_norm_eps (`float`, *optional*, defaults to 1e-12):
+            The epsilon used by the layer normalization layers.
+        position_embedding_type (`str`, *optional*, defaults to `"absolute"`):
+            Type of position embedding. Choose one of `"absolute"`, `"relative_key"`, `"relative_key_query"`. For
+            positional embeddings use `"absolute"`. For more information on `"relative_key"`, please refer to
+            [Self-Attention with Relative Position Representations (Shaw et al.)](https://arxiv.org/abs/1803.02155).
+            For more information on `"relative_key_query"`, please refer to *Method 4* in [Improve Transformer Models
+            with Better Relative Position Embeddings (Huang et al.)](https://arxiv.org/abs/2009.13658).
+        use_cache (`bool`, *optional*, defaults to `True`):
+            Whether or not the model should return the last key/values attentions (not used by all models). Only
+            relevant if `config.is_decoder=True`.
+        classifier_dropout (`float`, *optional*):
+            The dropout ratio for the classification head.
+
+    Example:
+
+    ```python
+    >>> from transformers import ChineseCLIPTextConfig, ChineseCLIPTextModel
+
+    >>> # Initializing a ChineseCLIPTextConfig with OFA-Sys/chinese-clip-vit-base-patch16 style configuration
+    >>> configuration = ChineseCLIPTextConfig()
+
+    >>> # Initializing a ChineseCLIPTextModel (with random weights) from the OFA-Sys/chinese-clip-vit-base-patch16 style configuration
+    >>> model = ChineseCLIPTextModel(configuration)
+
+    >>> # Accessing the model configuration
+    >>> configuration = model.config
+    ```"""
+    model_type = "chinese_clip_text_model"
+
+    def __init__(
+        self,
+        vocab_size=30522,
+        hidden_size=768,
+        num_hidden_layers=12,
+        num_attention_heads=12,
+        intermediate_size=3072,
+        hidden_act="gelu",
+        hidden_dropout_prob=0.1,
+        attention_probs_dropout_prob=0.1,
+        max_position_embeddings=512,
+        type_vocab_size=2,
+        initializer_range=0.02,
+        initializer_factor=1.0,
+        layer_norm_eps=1e-12,
+        pad_token_id=0,
+        position_embedding_type="absolute",
+        use_cache=True,
+        classifier_dropout=None,
+        **kwargs
+    ):
+        super().__init__(pad_token_id=pad_token_id, **kwargs)
+
+        self.vocab_size = vocab_size
+        self.hidden_size = hidden_size
+        self.num_hidden_layers = num_hidden_layers
+        self.num_attention_heads = num_attention_heads
+        self.hidden_act = hidden_act
+        self.intermediate_size = intermediate_size
+        self.hidden_dropout_prob = hidden_dropout_prob
+        self.attention_probs_dropout_prob = attention_probs_dropout_prob
+        self.max_position_embeddings = max_position_embeddings
+        self.type_vocab_size = type_vocab_size
+        self.initializer_range = initializer_range
+        self.initializer_factor = initializer_factor
+        self.layer_norm_eps = layer_norm_eps
+        self.position_embedding_type = position_embedding_type
+        self.use_cache = use_cache
+        self.classifier_dropout = classifier_dropout
+
+    @classmethod
+    def from_pretrained(cls, pretrained_model_name_or_path: Union[str, os.PathLike], **kwargs) -> "PretrainedConfig":
+
+        config_dict, kwargs = cls.get_config_dict(pretrained_model_name_or_path, **kwargs)
+
+        # get the vision config dict if we are loading from ChineseCLIPConfig
+        if config_dict.get("model_type") == "chinese_clip":
+            config_dict = config_dict["text_config"]
+
+        if "model_type" in config_dict and hasattr(cls, "model_type") and config_dict["model_type"] != cls.model_type:
+            logger.warning(
+                f"You are using a model of type {config_dict['model_type']} to instantiate a model of type "
+                f"{cls.model_type}. This is not supported for all configurations of models and can yield errors."
+            )
+
+        return cls.from_dict(config_dict, **kwargs)
+
+
+class ChineseCLIPVisionConfig(PretrainedConfig):
+    r"""
+    This is the configuration class to store the configuration of a [`ChineseCLIPModel`]. It is used to instantiate an
+    ChineseCLIP model according to the specified arguments, defining the model architecture. Instantiating a
+    configuration with the defaults will yield a similar configuration to that of the ChineseCLIP
+    [OFA-Sys/chinese-clip-vit-base-patch16](https:
+        //huggingface.co/OFA-Sys/chinese-clip-vit-base-patch16) architecture.
+
+    Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
+    documentation from [`PretrainedConfig`] for more information.
+
+
+    Args:
+        hidden_size (`int`, *optional*, defaults to 768):
+            Dimensionality of the encoder layers and the pooler layer.
+        intermediate_size (`int`, *optional*, defaults to 3072):
+            Dimensionality of the "intermediate" (i.e., feed-forward) layer in the Transformer encoder.
+        num_hidden_layers (`int`, *optional*, defaults to 12):
+            Number of hidden layers in the Transformer encoder.
+        num_attention_heads (`int`, *optional*, defaults to 12):
+            Number of attention heads for each attention layer in the Transformer encoder.
+        image_size (`int`, *optional*, defaults to 224):
+            The size (resolution) of each image.
+        patch_size (`int`, *optional*, defaults to 32):
+            The size (resolution) of each patch.
+        hidden_act (`str` or `function`, *optional*, defaults to `"quick_gelu"`):
+            The non-linear activation function (function or string) in the encoder and pooler. If string, `"gelu"`,
+            `"relu"`, `"selu"` and `"gelu_new"` ``"quick_gelu"` are supported. layer_norm_eps (`float`, *optional*,
+            defaults to 1e-5): The epsilon used by the layer normalization layers.
+        dropout (`float`, *optional*, defaults to 0.0):
+            The dropout probabilitiy for all fully connected layers in the embeddings, encoder, and pooler.
+        attention_dropout (`float`, *optional*, defaults to 0.0):
+            The dropout ratio for the attention probabilities.
+        initializer_range (`float`, *optional*, defaults to 0.02):
+            The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
+        initializer_factor (`float``, *optional*, defaults to 1):
+            A factor for initializing all weight matrices (should be kept to 1, used internally for initialization
+            testing).
+    Example:
+    ```python
+    >>> from transformers import ChineseCLIPVisionConfig, ChineseCLIPVisionModel
+
+    >>> # Initializing a ChineseCLIPVisionConfig with OFA-Sys/chinese-clip-vit-base-patch16 style configuration
+    >>> configuration = ChineseCLIPVisionConfig()
+
+    >>> # Initializing a ChineseCLIPVisionModel (with random weights) from the OFA-Sys/chinese-clip-vit-base-patch16 style configuration
+    >>> model = ChineseCLIPVisionModel(configuration)
+
+    >>> # Accessing the model configuration
+    >>> configuration = model.config
+    ```"""
+
+    model_type = "chinese_clip_vision_model"
+
+    def __init__(
+        self,
+        hidden_size=768,
+        intermediate_size=3072,
+        projection_dim=512,
+        num_hidden_layers=12,
+        num_attention_heads=12,
+        num_channels=3,
+        image_size=224,
+        patch_size=32,
+        hidden_act="quick_gelu",
+        layer_norm_eps=0.00001,
+        dropout=0.0,
+        attention_dropout=0.0,
+        initializer_range=0.02,
+        initializer_factor=1.0,
+        **kwargs
+    ):
+        super().__init__(**kwargs)
+
+        self.hidden_size = hidden_size
+        self.intermediate_size = intermediate_size
+        self.projection_dim = projection_dim
+        self.dropout = dropout
+        self.num_hidden_layers = num_hidden_layers
+        self.num_attention_heads = num_attention_heads
+        self.num_channels = num_channels
+        self.patch_size = patch_size
+        self.image_size = image_size
+        self.initializer_range = initializer_range
+        self.initializer_factor = initializer_factor
+        self.attention_dropout = attention_dropout
+        self.layer_norm_eps = layer_norm_eps
+        self.hidden_act = hidden_act
+
+    @classmethod
+    def from_pretrained(cls, pretrained_model_name_or_path: Union[str, os.PathLike], **kwargs) -> "PretrainedConfig":
+
+        config_dict, kwargs = cls.get_config_dict(pretrained_model_name_or_path, **kwargs)
+
+        # get the vision config dict if we are loading from ChineseCLIPConfig
+        if config_dict.get("model_type") == "chinese_clip":
+            config_dict = config_dict["vision_config"]
+
+        if "model_type" in config_dict and hasattr(cls, "model_type") and config_dict["model_type"] != cls.model_type:
+            logger.warning(
+                f"You are using a model of type {config_dict['model_type']} to instantiate a model of type "
+                f"{cls.model_type}. This is not supported for all configurations of models and can yield errors."
+            )
+
+        return cls.from_dict(config_dict, **kwargs)
+
+
+class ChineseCLIPConfig(PretrainedConfig):
+    r"""
+    [`ChineseCLIPConfig`] is the configuration class to store the configuration of a [`ChineseCLIPModel`]. It is used
+    to instantiate Chinese-CLIP model according to the specified arguments, defining the text model and vision model
+    configs. Instantiating a configuration with the defaults will yield a similar configuration to that of the
+    Chinese-CLIP [OFA-Sys/chinese-clip-vit-base-patch16](https://huggingface.co/OFA-Sys/chinese-clip-vit-base-patch16)
+    architecture.
+
+    Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
+    documentation from [`PretrainedConfig`] for more information.
+
+    Args:
+        text_config (`dict`, *optional*):
+            Dictionary of configuration options used to initialize [`ChineseCLIPTextConfig`].
+        vision_config (`dict`, *optional*):
+            Dictionary of configuration options used to initialize [`ChineseCLIPVisionConfig`].
+        projection_dim (`int`, *optional*, defaults to 512):
+            Dimentionality of text and vision projection layers.
+        logit_scale_init_value (`float`, *optional*, defaults to 2.6592):
+            The inital value of the *logit_scale* paramter. Default is used as per the original ChineseCLIP
+            implementation.
+        kwargs (*optional*):
+            Dictionary of keyword arguments.
+
+    Example:
+
+    ```python
+    >>> from transformers import ChineseCLIPConfig, ChineseCLIPModel
+
+    >>> # Initializing a ChineseCLIPConfig with OFA-Sys/chinese-clip-vit-base-patch16 style configuration
+    >>> configuration = ChineseCLIPConfig()
+
+    >>> # Initializing a ChineseCLIPModel (with random weights) from the OFA-Sys/chinese-clip-vit-base-patch16 style configuration
+    >>> model = ChineseCLIPModel(configuration)
+
+    >>> # Accessing the model configuration
+    >>> configuration = model.config
+
+    >>> # We can also initialize a ChineseCLIPConfig from a ChineseCLIPTextConfig and a ChineseCLIPVisionConfig
+
+    >>> # Initializing a ChineseCLIPTextConfig and ChineseCLIPVisionConfig configuration
+    >>> config_text = ChineseCLIPTextConfig()
+    >>> config_vision = ChineseCLIPVisionConfig()
+
+    >>> config = ChineseCLIPConfig.from_text_vision_configs(config_text, config_vision)
+    ```"""
+
+    model_type = "chinese_clip"
+    is_composition = True
+
+    def __init__(
+        self, text_config=None, vision_config=None, projection_dim=512, logit_scale_init_value=2.6592, **kwargs
+    ):
+        super().__init__(**kwargs)
+
+        # If `_config_dict` exist, we use them for the backward compatibility.
+        text_config_dict = kwargs.pop("text_config_dict", None)
+        vision_config_dict = kwargs.pop("vision_config_dict", None)
+        if text_config_dict is not None:
+            text_config = text_config_dict
+        if vision_config_dict is not None:
+            vision_config = vision_config_dict
+
+        if text_config is None:
+            text_config = {}
+            logger.info("text_config is None. Initializing the ChineseCLIPTextConfig with default values.")
+
+        if vision_config is None:
+            vision_config = {}
+            logger.info("vision_config is None. initializing the ChineseCLIPVisionConfig with default values.")
+
+        self.text_config = ChineseCLIPTextConfig(**text_config)
+        self.vision_config = ChineseCLIPVisionConfig(**vision_config)
+
+        self.projection_dim = projection_dim
+        self.logit_scale_init_value = logit_scale_init_value
+        self.initializer_factor = 1.0
+        self.initializer_range = 0.02
+
+    @classmethod
+    def from_text_vision_configs(
+        cls, text_config: ChineseCLIPTextConfig, vision_config: ChineseCLIPVisionConfig, **kwargs
+    ):
+        r"""
+        Instantiate a [`ChineseCLIPConfig`] (or a derived class) from Chinese-CLIP text model configuration and
+        Chinese-CLIP vision model configuration. Returns:
+            [`ChineseCLIPConfig`]: An instance of a configuration object
+        """
+
+        return cls(text_config=text_config.to_dict(), vision_config=vision_config.to_dict(), **kwargs)
+
+    def to_dict(self):
+        """
+        Serializes this instance to a Python dictionary. Override the default [`~PretrainedConfig.to_dict`].
+
+        Returns:
+            `Dict[str, any]`: Dictionary of all the attributes that make up this configuration instance,
+        """
+        output = copy.deepcopy(self.__dict__)
+        output["text_config"] = self.text_config.to_dict()
+        output["vision_config"] = self.vision_config.to_dict()
+        output["model_type"] = self.__class__.model_type
+        return output
+
+
+class ChineseCLIPOnnxConfig(OnnxConfig):
+    @property
+    def inputs(self) -> Mapping[str, Mapping[int, str]]:
+        return OrderedDict(
+            [
+                ("input_ids", {0: "batch", 1: "sequence"}),
+                ("pixel_values", {0: "batch", 1: "num_channels", 2: "height", 3: "width"}),
+                ("attention_mask", {0: "batch", 1: "sequence"}),
+            ]
+        )
+
+    @property
+    def outputs(self) -> Mapping[str, Mapping[int, str]]:
+        return OrderedDict(
+            [
+                ("logits_per_image", {0: "batch"}),
+                ("logits_per_text", {0: "batch"}),
+                ("text_embeds", {0: "batch"}),
+                ("image_embeds", {0: "batch"}),
+            ]
+        )
+
+    @property
+    def atol_for_validation(self) -> float:
+        return 1e-4
+
+    def generate_dummy_inputs(
+        self,
+        processor: "ProcessorMixin",
+        batch_size: int = -1,
+        seq_length: int = -1,
+        framework: Optional["TensorType"] = None,
+    ) -> Mapping[str, Any]:
+
+        text_input_dict = super().generate_dummy_inputs(
+            processor.tokenizer, batch_size=batch_size, seq_length=seq_length, framework=framework
+        )
+        image_input_dict = super().generate_dummy_inputs(
+            processor.feature_extractor, batch_size=batch_size, framework=framework
+        )
+        return {**text_input_dict, **image_input_dict}
+
+    @property
+    def default_onnx_opset(self) -> int:
+        return 14