Add SigLIP (#26522)

* Add first draft

* Use appropriate gelu function

* More improvements

* More improvements

* More improvements

* Convert checkpoint

* More improvements

* Improve docs, remove print statements

* More improvements

* Add link

* remove unused masking function

* begin tokenizer

* do_lower_case

* debug

* set split_special_tokens=True

* Remove script

* Fix style

* Fix rebase

* Use same design as CLIP

* Add fast tokenizer

* Add SiglipTokenizer to init, remove extra_ids

* Improve conversion script

* Use smaller inputs in conversion script

* Update conversion script

* More improvements

* Add processor to conversion script

* Add tests

* Remove print statements

* Add tokenizer tests

* Fix more tests

* More improvements related to weight initialization

* More improvements

* Make more tests pass

* More improvements

* More improvements

* Add copied from

* Add canonicalize_text

* Enable fast tokenizer tests

* More improvements

* Fix most slow tokenizer tests

* Address comments

* Fix style

* Remove script

* Address some comments

* Add copied from to tests

* Add more copied from

* Add more copied from

* Add more copied from

* Remove is_flax_available

* More updates

* Address comment

* Remove SiglipTokenizerFast for now

* Add caching

* Remove umt5 test

* Add canonicalize_text inside _tokenize, thanks Arthur

* Fix image processor tests

* Skip tests which are not applicable

* Skip test_initialization

* More improvements

* Compare pixel values

* Fix doc tests, add integration test

* Add do_normalize

* Remove causal mask and leverage ignore copy

* Fix attention_mask

* Fix remaining tests

* Fix dummies

* Rename temperature and bias

* Address comments

* Add copied from to tokenizer tests

* Add SiglipVisionModel to auto mapping

* Add copied from to image processor tests

* Improve doc

* Remove SiglipVisionModel from index

* Address comments

* Improve docs

* Simplify config

* Add first draft

* Make it like mistral

* More improvements

* Fix attention_mask

* Fix output_attentions

* Add note in docs

* Convert multilingual model

* Convert large checkpoint

* Convert more checkpoints

* Add pipeline support, correct image_mean and image_std

* Use padding=max_length by default

* Make processor like llava

* Add code snippet

* Convert more checkpoints

* Set keep_punctuation_string=None as in OpenCLIP

* Set normalized=False for special tokens

* Fix doc test

* Update integration test

* Add figure

* Update organization

* Happy new year

* Use AutoModel everywhere

---------
Co-authored-by: patil-suraj <surajp815@gmail.com>

README.md

@@ -475,6 +475,7 @@ Current number of checkpoints: ![](https://img.shields.io/endpoint?url=https://h
 1. **[Segment Anything](https://huggingface.co/docs/transformers/model_doc/sam)** (from Meta AI) released with the paper [Segment Anything](https://arxiv.org/pdf/2304.02643v1.pdf) by Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alex Berg, Wan-Yen Lo, Piotr Dollar, Ross Girshick.
 1. **[SEW](https://huggingface.co/docs/transformers/model_doc/sew)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
 1. **[SEW-D](https://huggingface.co/docs/transformers/model_doc/sew_d)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
+1. **[SigLIP](https://huggingface.co/docs/transformers/main/model_doc/siglip)** (from Google AI) released with the paper [Sigmoid Loss for Language Image Pre-Training](https://arxiv.org/abs/2303.15343) by Xiaohua Zhai, Basil Mustafa, Alexander Kolesnikov, Lucas Beyer.
 1. **[SpeechT5](https://huggingface.co/docs/transformers/model_doc/speecht5)** (from Microsoft Research) released with the paper [SpeechT5: Unified-Modal Encoder-Decoder Pre-Training for Spoken Language Processing](https://arxiv.org/abs/2110.07205) by Junyi Ao, Rui Wang, Long Zhou, Chengyi Wang, Shuo Ren, Yu Wu, Shujie Liu, Tom Ko, Qing Li, Yu Zhang, Zhihua Wei, Yao Qian, Jinyu Li, Furu Wei.
 1. **[SpeechToTextTransformer](https://huggingface.co/docs/transformers/model_doc/speech_to_text)** (from Facebook), released together with the paper [fairseq S2T: Fast Speech-to-Text Modeling with fairseq](https://arxiv.org/abs/2010.05171) by Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Dmytro Okhonko, Juan Pino.
 1. **[SpeechToTextTransformer2](https://huggingface.co/docs/transformers/model_doc/speech_to_text_2)** (from Facebook), released together with the paper [Large-Scale Self- and Semi-Supervised Learning for Speech Translation](https://arxiv.org/abs/2104.06678) by Changhan Wang, Anne Wu, Juan Pino, Alexei Baevski, Michael Auli, Alexis Conneau.

README_es.md

@@ -450,6 +450,7 @@ Número actual de puntos de control: ![](https://img.shields.io/endpoint?url=htt
 1. **[Segment Anything](https://huggingface.co/docs/transformers/model_doc/sam)** (from Meta AI) released with the paper [Segment Anything](https://arxiv.org/pdf/2304.02643v1.pdf) by Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alex Berg, Wan-Yen Lo, Piotr Dollar, Ross Girshick.
 1. **[SEW](https://huggingface.co/docs/transformers/model_doc/sew)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
 1. **[SEW-D](https://huggingface.co/docs/transformers/model_doc/sew_d)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
+1. **[SigLIP](https://huggingface.co/docs/transformers/main/model_doc/siglip)** (from Google AI) released with the paper [Sigmoid Loss for Language Image Pre-Training](https://arxiv.org/abs/2303.15343) by Xiaohua Zhai, Basil Mustafa, Alexander Kolesnikov, Lucas Beyer.
 1. **[SpeechT5](https://huggingface.co/docs/transformers/model_doc/speecht5)** (from Microsoft Research) released with the paper [SpeechT5: Unified-Modal Encoder-Decoder Pre-Training for Spoken Language Processing](https://arxiv.org/abs/2110.07205) by Junyi Ao, Rui Wang, Long Zhou, Chengyi Wang, Shuo Ren, Yu Wu, Shujie Liu, Tom Ko, Qing Li, Yu Zhang, Zhihua Wei, Yao Qian, Jinyu Li, Furu Wei.
 1. **[SpeechToTextTransformer](https://huggingface.co/docs/transformers/model_doc/speech_to_text)** (from Facebook), released together with the paper [fairseq S2T: Fast Speech-to-Text Modeling with fairseq](https://arxiv.org/abs/2010.05171) by Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Dmytro Okhonko, Juan Pino.
 1. **[SpeechToTextTransformer2](https://huggingface.co/docs/transformers/model_doc/speech_to_text_2)** (from Facebook), released together with the paper [Large-Scale Self- and Semi-Supervised Learning for Speech Translation](https://arxiv.org/abs/2104.06678) by Changhan Wang, Anne Wu, Juan Pino, Alexei Baevski, Michael Auli, Alexis Conneau.

README_hd.md

@@ -424,6 +424,7 @@ conda install conda-forge::transformers
 1. **[Segment Anything](https://huggingface.co/docs/transformers/model_doc/sam)** (Meta AI से) Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alex Berg, Wan-Yen Lo, Piotr Dollar, Ross Girshick. द्वाराअनुसंधान पत्र [Segment Anything](https://arxiv.org/pdf/2304.02643v1.pdf) के साथ जारी किया गया
 1. **[SEW](https://huggingface.co/docs/transformers/model_doc/sew)** (ASAPP से) साथ देने वाला पेपर [भाषण पहचान के लिए अनसुपरवाइज्ड प्री-ट्रेनिंग में परफॉर्मेंस-एफिशिएंसी ट्रेड-ऑफ्स](https ://arxiv.org/abs/2109.06870) फेलिक्स वू, क्वांगयुन किम, जिंग पैन, क्यू हान, किलियन क्यू. वेनबर्गर, योव आर्टज़ी द्वारा।
 1. **[SEW-D](https://huggingface.co/docs/transformers/model_doc/sew_d)** (ASAPP से) साथ में पेपर [भाषण पहचान के लिए अनसुपरवाइज्ड प्री-ट्रेनिंग में परफॉर्मेंस-एफिशिएंसी ट्रेड-ऑफ्स] (https://arxiv.org/abs/2109.06870) फेलिक्स वू, क्वांगयुन किम, जिंग पैन, क्यू हान, किलियन क्यू. वेनबर्गर, योआव आर्टज़ी द्वारा पोस्ट किया गया।
+1. **[SigLIP](https://huggingface.co/docs/transformers/main/model_doc/siglip)** (Google AI से) Xiaohua Zhai, Basil Mustafa, Alexander Kolesnikov, Lucas Beyer. द्वाराअनुसंधान पत्र [Sigmoid Loss for Language Image Pre-Training](https://arxiv.org/abs/2303.15343) के साथ जारी किया गया
 1. **[SpeechT5](https://huggingface.co/docs/transformers/model_doc/speecht5)** (from Microsoft Research) released with the paper [SpeechT5: Unified-Modal Encoder-Decoder Pre-Training for Spoken Language Processing](https://arxiv.org/abs/2110.07205) by Junyi Ao, Rui Wang, Long Zhou, Chengyi Wang, Shuo Ren, Yu Wu, Shujie Liu, Tom Ko, Qing Li, Yu Zhang, Zhihua Wei, Yao Qian, Jinyu Li, Furu Wei.
 1. **[SpeechToTextTransformer](https://huggingface.co/docs/transformers/model_doc/speech_to_text)** (फेसबुक से), साथ में पेपर [फेयरसेक S2T: फास्ट स्पीच-टू-टेक्स्ट मॉडलिंग विद फेयरसेक](https: //arxiv.org/abs/2010.05171) चांगहान वांग, यूं तांग, जुताई मा, ऐनी वू, दिमित्रो ओखोनको, जुआन पिनो द्वारा पोस्ट किया गया。
 1. **[SpeechToTextTransformer2](https://huggingface.co/docs/transformers/model_doc/speech_to_text_2)** (फेसबुक से) साथ में पेपर [लार्ज-स्केल सेल्फ- एंड सेमी-सुपरवाइज्ड लर्निंग फॉर स्पीच ट्रांसलेशन](https://arxiv.org/abs/2104.06678) चांगहान वांग, ऐनी वू, जुआन पिनो, एलेक्सी बेवस्की, माइकल औली, एलेक्सिस द्वारा Conneau द्वारा पोस्ट किया गया।

README_ja.md

@@ -484,6 +484,7 @@ Flax、PyTorch、TensorFlowをcondaでインストールする方法は、それ
 1. **[Segment Anything](https://huggingface.co/docs/transformers/model_doc/sam)** (Meta AI から) Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alex Berg, Wan-Yen Lo, Piotr Dollar, Ross Girshick. から公開された研究論文 [Segment Anything](https://arxiv.org/pdf/2304.02643v1.pdf)
 1. **[SEW](https://huggingface.co/docs/transformers/model_doc/sew)** (ASAPP から) Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi から公開された研究論文: [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870)
 1. **[SEW-D](https://huggingface.co/docs/transformers/model_doc/sew_d)** (ASAPP から) Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi から公開された研究論文: [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870)
+1. **[SigLIP](https://huggingface.co/docs/transformers/main/model_doc/siglip)** (Google AI から) Xiaohua Zhai, Basil Mustafa, Alexander Kolesnikov, Lucas Beyer. から公開された研究論文 [Sigmoid Loss for Language Image Pre-Training](https://arxiv.org/abs/2303.15343)
 1. **[SpeechT5](https://huggingface.co/docs/transformers/model_doc/speecht5)** (Microsoft Research から) Junyi Ao, Rui Wang, Long Zhou, Chengyi Wang, Shuo Ren, Yu Wu, Shujie Liu, Tom Ko, Qing Li, Yu Zhang, Zhihua Wei, Yao Qian, Jinyu Li, Furu Wei. から公開された研究論文 [SpeechT5: Unified-Modal Encoder-Decoder Pre-Training for Spoken Language Processing](https://arxiv.org/abs/2110.07205)
 1. **[SpeechToTextTransformer](https://huggingface.co/docs/transformers/model_doc/speech_to_text)** (Facebook から), Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Dmytro Okhonko, Juan Pino から公開された研究論文: [fairseq S2T: Fast Speech-to-Text Modeling with fairseq](https://arxiv.org/abs/2010.05171)
 1. **[SpeechToTextTransformer2](https://huggingface.co/docs/transformers/model_doc/speech_to_text_2)** (Facebook から), Changhan Wang, Anne Wu, Juan Pino, Alexei Baevski, Michael Auli, Alexis Conneau から公開された研究論文: [Large-Scale Self- and Semi-Supervised Learning for Speech Translation](https://arxiv.org/abs/2104.06678)

README_ko.md

@@ -399,6 +399,7 @@ Flax, PyTorch, TensorFlow 설치 페이지에서 이들을 conda로 설치하는
 1. **[Segment Anything](https://huggingface.co/docs/transformers/model_doc/sam)** (Meta AI 에서 제공)은 Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alex Berg, Wan-Yen Lo, Piotr Dollar, Ross Girshick.의 [Segment Anything](https://arxiv.org/pdf/2304.02643v1.pdf)논문과 함께 발표했습니다.
 1. **[SEW](https://huggingface.co/docs/transformers/model_doc/sew)** (ASAPP 에서) Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi 의 [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) 논문과 함께 발표했습니다.
 1. **[SEW-D](https://huggingface.co/docs/transformers/model_doc/sew_d)** (ASAPP 에서) Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi 의 [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) 논문과 함께 발표했습니다.
+1. **[SigLIP](https://huggingface.co/docs/transformers/main/model_doc/siglip)** (Google AI 에서 제공)은 Xiaohua Zhai, Basil Mustafa, Alexander Kolesnikov, Lucas Beyer.의 [Sigmoid Loss for Language Image Pre-Training](https://arxiv.org/abs/2303.15343)논문과 함께 발표했습니다.
 1. **[SpeechT5](https://huggingface.co/docs/transformers/model_doc/speecht5)** (Microsoft Research 에서 제공)은 Junyi Ao, Rui Wang, Long Zhou, Chengyi Wang, Shuo Ren, Yu Wu, Shujie Liu, Tom Ko, Qing Li, Yu Zhang, Zhihua Wei, Yao Qian, Jinyu Li, Furu Wei.의 [SpeechT5: Unified-Modal Encoder-Decoder Pre-Training for Spoken Language Processing](https://arxiv.org/abs/2110.07205)논문과 함께 발표했습니다.
 1. **[SpeechToTextTransformer](https://huggingface.co/docs/transformers/model_doc/speech_to_text)** (Facebook 에서) Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Dmytro Okhonko, Juan Pino 의 [fairseq S2T: Fast Speech-to-Text Modeling with fairseq](https://arxiv.org/abs/2010.05171) 논문과 함께 발표했습니다.
 1. **[SpeechToTextTransformer2](https://huggingface.co/docs/transformers/model_doc/speech_to_text_2)** (Facebook 에서) Changhan Wang, Anne Wu, Juan Pino, Alexei Baevski, Michael Auli, Alexis Conneau 의 [Large-Scale Self- and Semi-Supervised Learning for Speech Translation](https://arxiv.org/abs/2104.06678) 논문과 함께 발표했습니다.

README_zh-hans.md

@@ -423,6 +423,7 @@ conda install conda-forge::transformers
 1. **[Segment Anything](https://huggingface.co/docs/transformers/model_doc/sam)** (来自 Meta AI) 伴随论文 [Segment Anything](https://arxiv.org/pdf/2304.02643v1.pdf) 由 Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alex Berg, Wan-Yen Lo, Piotr Dollar, Ross Girshick 发布。
 1. **[SEW](https://huggingface.co/docs/transformers/model_doc/sew)** (来自 ASAPP) 伴随论文 [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) 由 Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi 发布。
 1. **[SEW-D](https://huggingface.co/docs/transformers/model_doc/sew_d)** (来自 ASAPP) 伴随论文 [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) 由 Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi 发布。
+1. **[SigLIP](https://huggingface.co/docs/transformers/main/model_doc/siglip)** (来自 Google AI) 伴随论文 [Sigmoid Loss for Language Image Pre-Training](https://arxiv.org/abs/2303.15343) 由 Xiaohua Zhai, Basil Mustafa, Alexander Kolesnikov, Lucas Beyer 发布。
 1. **[SpeechT5](https://huggingface.co/docs/transformers/model_doc/speecht5)** (来自 Microsoft Research) 伴随论文 [SpeechT5: Unified-Modal Encoder-Decoder Pre-Training for Spoken Language Processing](https://arxiv.org/abs/2110.07205) 由 Junyi Ao, Rui Wang, Long Zhou, Chengyi Wang, Shuo Ren, Yu Wu, Shujie Liu, Tom Ko, Qing Li, Yu Zhang, Zhihua Wei, Yao Qian, Jinyu Li, Furu Wei 发布。
 1. **[SpeechToTextTransformer](https://huggingface.co/docs/transformers/model_doc/speech_to_text)** (来自 Facebook), 伴随论文 [fairseq S2T: Fast Speech-to-Text Modeling with fairseq](https://arxiv.org/abs/2010.05171) 由 Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Dmytro Okhonko, Juan Pino 发布。
 1. **[SpeechToTextTransformer2](https://huggingface.co/docs/transformers/model_doc/speech_to_text_2)** (来自 Facebook) 伴随论文 [Large-Scale Self- and Semi-Supervised Learning for Speech Translation](https://arxiv.org/abs/2104.06678) 由 Changhan Wang, Anne Wu, Juan Pino, Alexei Baevski, Michael Auli, Alexis Conneau 发布。

README_zh-hant.md

@@ -435,6 +435,7 @@ conda install conda-forge::transformers
 1. **[Segment Anything](https://huggingface.co/docs/transformers/model_doc/sam)** (from Meta AI) released with the paper [Segment Anything](https://arxiv.org/pdf/2304.02643v1.pdf) by Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alex Berg, Wan-Yen Lo, Piotr Dollar, Ross Girshick.
 1. **[SEW](https://huggingface.co/docs/transformers/model_doc/sew)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
 1. **[SEW-D](https://huggingface.co/docs/transformers/model_doc/sew_d)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
+1. **[SigLIP](https://huggingface.co/docs/transformers/main/model_doc/siglip)** (from Google AI) released with the paper [Sigmoid Loss for Language Image Pre-Training](https://arxiv.org/abs/2303.15343) by Xiaohua Zhai, Basil Mustafa, Alexander Kolesnikov, Lucas Beyer.
 1. **[SpeechT5](https://huggingface.co/docs/transformers/model_doc/speecht5)** (from Microsoft Research) released with the paper [SpeechT5: Unified-Modal Encoder-Decoder Pre-Training for Spoken Language Processing](https://arxiv.org/abs/2110.07205) by Junyi Ao, Rui Wang, Long Zhou, Chengyi Wang, Shuo Ren, Yu Wu, Shujie Liu, Tom Ko, Qing Li, Yu Zhang, Zhihua Wei, Yao Qian, Jinyu Li, Furu Wei.
 1. **[SpeechToTextTransformer](https://huggingface.co/docs/transformers/model_doc/speech_to_text)** (from Facebook), released together with the paper [fairseq S2T: Fast Speech-to-Text Modeling with fairseq](https://arxiv.org/abs/2010.05171) by Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Dmytro Okhonko, Juan Pino.
 1. **[SpeechToTextTransformer2](https://huggingface.co/docs/transformers/model_doc/speech_to_text_2)** (from Facebook) released with the paper [Large-Scale Self- and Semi-Supervised Learning for Speech Translation](https://arxiv.org/abs/2104.06678) by Changhan Wang, Anne Wu, Juan Pino, Alexei Baevski, Michael Auli, Alexis Conneau.

docs/source/en/_toctree.yml

@@ -733,6 +733,8 @@
         title: Pix2Struct
       - local: model_doc/sam
         title: Segment Anything
+      - local: model_doc/siglip
+        title: SigLIP
       - local: model_doc/speech-encoder-decoder
         title: Speech Encoder Decoder Models
       - local: model_doc/tapas

docs/source/en/index.md

@@ -250,6 +250,7 @@ Flax), PyTorch, and/or TensorFlow.
 |                     [SegFormer](model_doc/segformer)                     |       ✅        |         ✅         |      ❌      |
 |                           [SEW](model_doc/sew)                           |       ✅        |         ❌         |      ❌      |
 |                         [SEW-D](model_doc/sew-d)                         |       ✅        |         ❌         |      ❌      |
+|                        [SigLIP](model_doc/siglip)                        |       ✅        |         ❌         |      ❌      |
 |        [Speech Encoder decoder](model_doc/speech-encoder-decoder)        |       ✅        |         ❌         |      ✅      |
 |                 [Speech2Text](model_doc/speech_to_text)                  |       ✅        |         ✅         |      ❌      |
 |                      [SpeechT5](model_doc/speecht5)                      |       ✅        |         ❌         |      ❌      |

docs/source/en/model_doc/siglip.md

+<!--Copyright 2023 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.
+
+⚠️ Note that this file is in Markdown but contain specific syntax for our doc-builder (similar to MDX) that may not be
+rendered properly in your Markdown viewer.
+
+-->
+
+# SigLIP
+
+## Overview
+
+The SigLIP model was proposed in [Sigmoid Loss for Language Image Pre-Training](https://arxiv.org/abs/2303.15343) by Xiaohua Zhai, Basil Mustafa, Alexander Kolesnikov, Lucas Beyer. SigLIP proposes to replace the loss function used in [CLIP](clip) by a simple pairwise sigmoid loss. This results in better performance in terms of zero-shot classification accuracy on ImageNet.
+
+The abstract from the paper is the following:
+
+*We propose a simple pairwise Sigmoid loss for Language-Image Pre-training (SigLIP). Unlike standard contrastive learning with softmax normalization, the sigmoid loss operates solely on image-text pairs and does not require a global view of the pairwise similarities for normalization. The sigmoid loss simultaneously allows further scaling up the batch size, while also performing better at smaller batch sizes. Combined with Locked-image Tuning, with only four TPUv4 chips, we train a SigLiT model that achieves 84.5% ImageNet zero-shot accuracy in two days. The disentanglement of the batch size from the loss further allows us to study the impact of examples vs pairs and negative to positive ratio. Finally, we push the batch size to the extreme, up to one million, and find that the benefits of growing batch size quickly diminish, with a more reasonable batch size of 32k being sufficient.*
+
+## Usage tips
+
+- Usage of SigLIP is similar to [CLIP](clip). The main difference is the training loss, which does not require a global view of all the pairwise similarities of images and texts within a batch. One needs to apply the sigmoid activation function to the logits, rather than the softmax.
+- Training is not yet supported. If you want to fine-tune SigLIP or train from scratch, refer to the loss function from [OpenCLIP](https://github.com/mlfoundations/open_clip/blob/73ad04ae7fb93ede1c02dc9040a828634cb1edf1/src/open_clip/loss.py#L307), which leverages various `torch.distributed` utilities.
+- When using the standalone [`SiglipTokenizer`], make sure to pass `padding="max_length"` as that's how the model was trained. The multimodal [`SiglipProcessor`] takes care of this behind the scenes.
+
+<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers/model_doc/siglip_table.jpeg"
+alt="drawing" width="600"/>
+
+<small> SigLIP evaluation results compared to CLIP. Taken from the <a href="https://arxiv.org/abs/2303.15343">original paper</a>.</small>
+
+This model was contributed by [nielsr](https://huggingface.co/nielsr).
+The original code can be found [here](https://github.com/google-research/big_vision/tree/main).
+
+## Usage example
+
+There are 2 main ways to use SigLIP: either using the pipeline API, which abstracts away all the complexity for you, or by using the `SiglipModel` class yourself.
+
+### Pipeline API
+
+The pipeline allows to use the model in a few lines of code:
+
+```python
+>>> from transformers import pipeline
+>>> from PIL import Image
+>>> import requests
+
+>>> # load pipe
+>>> image_classifier = pipeline(task="zero-shot-image-classification", model="google/siglip-base-patch16-224")
+
+>>> # load image
+>>> url = 'http://images.cocodataset.org/val2017/000000039769.jpg'
+>>> image = Image.open(requests.get(url, stream=True).raw)
+
+>>> # inference
+>>> outputs = image_classifier(image, candidate_labels=["2 cats", "a plane", "a remote"])
+>>> outputs = [{"score": round(output["score"], 4), "label": output["label"] } for output in outputs]
+>>> print(outputs)
+[{'score': 0.1979, 'label': '2 cats'}, {'score': 0.0, 'label': 'a remote'}, {'score': 0.0, 'label': 'a plane'}]
+```
+
+### Using the model yourself
+
+If you want to do the pre- and postprocessing yourself, here's how to do that:
+
+```python
+>>> from PIL import Image
+>>> import requests
+>>> from transformers import AutoProcessor, AutoModel
+>>> import torch
+
+>>> model = AutoModel.from_pretrained("google/siglip-base-patch16-224")
+>>> processor = AutoProcessor.from_pretrained("google/siglip-base-patch16-224")
+
+>>> url = "http://images.cocodataset.org/val2017/000000039769.jpg"
+>>> image = Image.open(requests.get(url, stream=True).raw)
+
+>>> texts = ["a photo of 2 cats", "a photo of 2 dogs"]
+>>> inputs = processor(text=texts, images=image, return_tensors="pt")
+
+>>> with torch.no_grad():
+...     outputs = model(**inputs)
+
+>>> logits_per_image = outputs.logits_per_image
+>>> probs = torch.sigmoid(logits_per_image) # these are the probabilities
+>>> print(f"{probs[0][0]:.1%} that image 0 is '{texts[0]}'")
+31.9% that image 0 is 'a photo of 2 cats'
+```
+
+## SiglipConfig
+
+[[autodoc]] SiglipConfig
+    - from_text_vision_configs
+
+## SiglipTextConfig
+
+[[autodoc]] SiglipTextConfig
+
+## SiglipVisionConfig
+
+[[autodoc]] SiglipVisionConfig
+
+## SiglipTokenizer
+
+[[autodoc]] SiglipTokenizer
+    - build_inputs_with_special_tokens
+    - get_special_tokens_mask
+    - create_token_type_ids_from_sequences
+    - save_vocabulary
+
+## SiglipImageProcessor
+
+[[autodoc]] SiglipImageProcessor
+    - preprocess
+
+## SiglipProcessor
+
+[[autodoc]] SiglipProcessor
+
+## SiglipModel
+
+[[autodoc]] SiglipModel
+    - forward
+    - get_text_features
+    - get_image_features
+
+## SiglipTextModel
+
+[[autodoc]] SiglipTextModel
+    - forward
+
+## SiglipVisionModel
+
+[[autodoc]] SiglipVisionModel
+    - forward

src/transformers/__init__.py

@@ -762,6 +762,14 @@ _import_structure = {
     "models.segformer": ["SEGFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP", "SegformerConfig"],
     "models.sew": ["SEW_PRETRAINED_CONFIG_ARCHIVE_MAP", "SEWConfig"],
     "models.sew_d": ["SEW_D_PRETRAINED_CONFIG_ARCHIVE_MAP", "SEWDConfig"],
+    "models.siglip": [
+        "SIGLIP_PRETRAINED_CONFIG_ARCHIVE_MAP",
+        "SiglipConfig",
+        "SiglipProcessor",
+        "SiglipTextConfig",
+        "SiglipTokenizer",
+        "SiglipVisionConfig",
+    ],
     "models.speech_encoder_decoder": ["SpeechEncoderDecoderConfig"],
     "models.speech_to_text": [
         "SPEECH_TO_TEXT_PRETRAINED_CONFIG_ARCHIVE_MAP",
@@ -1290,6 +1298,7 @@ else:
     _import_structure["models.pvt"].extend(["PvtImageProcessor"])
     _import_structure["models.sam"].extend(["SamImageProcessor"])
     _import_structure["models.segformer"].extend(["SegformerFeatureExtractor", "SegformerImageProcessor"])
+    _import_structure["models.siglip"].append("SiglipImageProcessor")
     _import_structure["models.swin2sr"].append("Swin2SRImageProcessor")
     _import_structure["models.tvlt"].append("TvltImageProcessor")
     _import_structure["models.tvp"].append("TvpImageProcessor")
@@ -3155,6 +3164,15 @@ else:
             "SEWDPreTrainedModel",
         ]
     )
+    _import_structure["models.siglip"].extend(
+        [
+            "SIGLIP_PRETRAINED_MODEL_ARCHIVE_LIST",
+            "SiglipModel",
+            "SiglipPreTrainedModel",
+            "SiglipTextModel",
+            "SiglipVisionModel",
+        ]
+    )
     _import_structure["models.speech_encoder_decoder"].extend(["SpeechEncoderDecoderModel"])
     _import_structure["models.speech_to_text"].extend(
         [
@@ -5441,6 +5459,14 @@ if TYPE_CHECKING:
     )
     from .models.sew import SEW_PRETRAINED_CONFIG_ARCHIVE_MAP, SEWConfig
     from .models.sew_d import SEW_D_PRETRAINED_CONFIG_ARCHIVE_MAP, SEWDConfig
+    from .models.siglip import (
+        SIGLIP_PRETRAINED_CONFIG_ARCHIVE_MAP,
+        SiglipConfig,
+        SiglipProcessor,
+        SiglipTextConfig,
+        SiglipTokenizer,
+        SiglipVisionConfig,
+    )
     from .models.speech_encoder_decoder import SpeechEncoderDecoderConfig
     from .models.speech_to_text import (
         SPEECH_TO_TEXT_PRETRAINED_CONFIG_ARCHIVE_MAP,
@@ -5966,6 +5992,7 @@ if TYPE_CHECKING:
         from .models.pvt import PvtImageProcessor
         from .models.sam import SamImageProcessor
         from .models.segformer import SegformerFeatureExtractor, SegformerImageProcessor
+        from .models.siglip import SiglipImageProcessor
         from .models.swin2sr import Swin2SRImageProcessor
         from .models.tvlt import TvltImageProcessor
         from .models.tvp import TvpImageProcessor
@@ -7508,6 +7535,13 @@ if TYPE_CHECKING:
             SEWDModel,
             SEWDPreTrainedModel,
         )
+        from .models.siglip import (
+            SIGLIP_PRETRAINED_MODEL_ARCHIVE_LIST,
+            SiglipModel,
+            SiglipPreTrainedModel,
+            SiglipTextModel,
+            SiglipVisionModel,
+        )
         from .models.speech_encoder_decoder import SpeechEncoderDecoderModel
         from .models.speech_to_text import (
             SPEECH_TO_TEXT_PRETRAINED_MODEL_ARCHIVE_LIST,

src/transformers/models/__init__.py

@@ -193,6 +193,7 @@ from . import (
     segformer,
     sew,
     sew_d,
+    siglip,
     speech_encoder_decoder,
     speech_to_text,
     speech_to_text_2,

src/transformers/models/auto/configuration_auto.py

@@ -200,6 +200,8 @@ CONFIG_MAPPING_NAMES = OrderedDict(
         ("segformer", "SegformerConfig"),
         ("sew", "SEWConfig"),
         ("sew-d", "SEWDConfig"),
+        ("siglip", "SiglipConfig"),
+        ("siglip_vision_model", "SiglipVisionConfig"),
         ("speech-encoder-decoder", "SpeechEncoderDecoderConfig"),
         ("speech_to_text", "Speech2TextConfig"),
         ("speech_to_text_2", "Speech2Text2Config"),
@@ -419,6 +421,7 @@ CONFIG_ARCHIVE_MAP_MAPPING_NAMES = OrderedDict(
         ("segformer", "SEGFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP"),
         ("sew", "SEW_PRETRAINED_CONFIG_ARCHIVE_MAP"),
         ("sew-d", "SEW_D_PRETRAINED_CONFIG_ARCHIVE_MAP"),
+        ("siglip", "SIGLIP_PRETRAINED_CONFIG_ARCHIVE_MAP"),
         ("speech_to_text", "SPEECH_TO_TEXT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
         ("speech_to_text_2", "SPEECH_TO_TEXT_2_PRETRAINED_CONFIG_ARCHIVE_MAP"),
         ("speecht5", "SPEECHT5_PRETRAINED_CONFIG_ARCHIVE_MAP"),
@@ -664,6 +667,8 @@ MODEL_NAMES_MAPPING = OrderedDict(
         ("segformer", "SegFormer"),
         ("sew", "SEW"),
         ("sew-d", "SEW-D"),
+        ("siglip", "SigLIP"),
+        ("siglip_vision_model", "SiglipVisionModel"),
         ("speech-encoder-decoder", "Speech Encoder decoder"),
         ("speech_to_text", "Speech2Text"),
         ("speech_to_text_2", "Speech2Text2"),
@@ -755,6 +760,7 @@ SPECIAL_MODEL_TYPE_TO_MODULE_NAME = OrderedDict(
         ("maskformer-swin", "maskformer"),
         ("xclip", "x_clip"),
         ("clip_vision_model", "clip"),
+        ("siglip_vision_model", "siglip"),
     ]
 )
 

src/transformers/models/auto/image_processing_auto.py

@@ -97,6 +97,7 @@ IMAGE_PROCESSOR_MAPPING_NAMES = OrderedDict(
         ("resnet", "ConvNextImageProcessor"),
         ("sam", "SamImageProcessor"),
         ("segformer", "SegformerImageProcessor"),
+        ("siglip", "SiglipImageProcessor"),
         ("swiftformer", "ViTImageProcessor"),
         ("swin", "ViTImageProcessor"),
         ("swin2sr", "Swin2SRImageProcessor"),

src/transformers/models/auto/modeling_auto.py

@@ -193,6 +193,8 @@ MODEL_MAPPING_NAMES = OrderedDict(
         ("segformer", "SegformerModel"),
         ("sew", "SEWModel"),
         ("sew-d", "SEWDModel"),
+        ("siglip", "SiglipModel"),
+        ("siglip_vision_model", "SiglipVisionModel"),
         ("speech_to_text", "Speech2TextModel"),
         ("speecht5", "SpeechT5Model"),
         ("splinter", "SplinterModel"),
@@ -1102,6 +1104,7 @@ MODEL_FOR_ZERO_SHOT_IMAGE_CLASSIFICATION_MAPPING_NAMES = OrderedDict(
         ("chinese_clip", "ChineseCLIPModel"),
         ("clip", "CLIPModel"),
         ("clipseg", "CLIPSegModel"),
+        ("siglip", "SiglipModel"),
     ]
 )
 

src/transformers/models/auto/processing_auto.py

@@ -77,6 +77,7 @@ PROCESSOR_MAPPING_NAMES = OrderedDict(
         ("seamless_m4t", "SeamlessM4TProcessor"),
         ("sew", "Wav2Vec2Processor"),
         ("sew-d", "Wav2Vec2Processor"),
+        ("siglip", "SiglipProcessor"),
         ("speech_to_text", "Speech2TextProcessor"),
         ("speech_to_text_2", "Speech2Text2Processor"),
         ("speecht5", "SpeechT5Processor"),

src/transformers/models/auto/tokenization_auto.py

@@ -372,6 +372,7 @@ else:
                     "SeamlessM4TTokenizerFast" if is_tokenizers_available() else None,
                 ),
             ),
+            ("siglip", ("SiglipTokenizer", None)),
             ("speech_to_text", ("Speech2TextTokenizer" if is_sentencepiece_available() else None, None)),
             ("speech_to_text_2", ("Speech2Text2Tokenizer", None)),
             ("speecht5", ("SpeechT5Tokenizer" if is_sentencepiece_available() else None, None)),

src/transformers/models/siglip/__init__.py

+# Copyright 2024 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_siglip": [
+        "SIGLIP_PRETRAINED_CONFIG_ARCHIVE_MAP",
+        "SiglipConfig",
+        "SiglipTextConfig",
+        "SiglipVisionConfig",
+    ],
+    "processing_siglip": ["SiglipProcessor"],
+    "tokenization_siglip": ["SiglipTokenizer"],
+}
+
+try:
+    if not is_vision_available():
+        raise OptionalDependencyNotAvailable()
+except OptionalDependencyNotAvailable:
+    pass
+else:
+    _import_structure["image_processing_siglip"] = ["SiglipImageProcessor"]
+
+try:
+    if not is_torch_available():
+        raise OptionalDependencyNotAvailable()
+except OptionalDependencyNotAvailable:
+    pass
+else:
+    _import_structure["modeling_siglip"] = [
+        "SIGLIP_PRETRAINED_MODEL_ARCHIVE_LIST",
+        "SiglipModel",
+        "SiglipPreTrainedModel",
+        "SiglipTextModel",
+        "SiglipVisionModel",
+    ]
+
+
+if TYPE_CHECKING:
+    from .configuration_siglip import (
+        SIGLIP_PRETRAINED_CONFIG_ARCHIVE_MAP,
+        SiglipConfig,
+        SiglipTextConfig,
+        SiglipVisionConfig,
+    )
+    from .processing_siglip import SiglipProcessor
+    from .tokenization_siglip import SiglipTokenizer
+
+    try:
+        if not is_vision_available():
+            raise OptionalDependencyNotAvailable()
+    except OptionalDependencyNotAvailable:
+        pass
+    else:
+        from .image_processing_siglip import SiglipImageProcessor
+
+    try:
+        if not is_torch_available():
+            raise OptionalDependencyNotAvailable()
+    except OptionalDependencyNotAvailable:
+        pass
+    else:
+        from .modeling_siglip import (
+            SIGLIP_PRETRAINED_MODEL_ARCHIVE_LIST,
+            SiglipModel,
+            SiglipPreTrainedModel,
+            SiglipTextModel,
+            SiglipVisionModel,
+        )
+
+
+else:
+    import sys
+
+    sys.modules[__name__] = _LazyModule(__name__, globals()["__file__"], _import_structure, module_spec=__spec__)

src/transformers/models/siglip/configuration_siglip.py

+# coding=utf-8
+# Copyright 2024 The HuggingFace Inc. 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.
+""" Siglip model configuration"""
+
+import os
+from typing import Union
+
+from ...configuration_utils import PretrainedConfig
+from ...utils import logging
+
+
+logger = logging.get_logger(__name__)
+
+SIGLIP_PRETRAINED_CONFIG_ARCHIVE_MAP = {
+    "google/siglip-base-patch16-224": "https://huggingface.co/google/siglip-base-patch16-224/resolve/main/config.json",
+}
+
+
+class SiglipTextConfig(PretrainedConfig):
+    r"""
+    This is the configuration class to store the configuration of a [`SiglipTextModel`]. It is used to instantiate a
+    Siglip text encoder according to the specified arguments, defining the model architecture. Instantiating a
+    configuration with the defaults will yield a similar configuration to that of the text encoder of the Siglip
+    [google/siglip-base-patch16-224](https://huggingface.co/google/siglip-base-patch16-224) 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 32000):
+            Vocabulary size of the Siglip text model. Defines the number of different tokens that can be represented by
+            the `inputs_ids` passed when calling [`SiglipModel`].
+        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.
+        max_position_embeddings (`int`, *optional*, defaults to 64):
+            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).
+        hidden_act (`str` or `function`, *optional*, defaults to `"gelu_pytorch_tanh"`):
+            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-06):
+            The epsilon used by the layer normalization layers.
+        attention_dropout (`float`, *optional*, defaults to 0.0):
+            The dropout ratio for the attention probabilities.
+        pad_token_id (`int`, *optional*, defaults to 1):
+            The id of the padding token in the vocabulary.
+        bos_token_id (`int`, *optional*, defaults to 49406):
+            The id of the beginning-of-sequence token in the vocabulary.
+        eos_token_id (`int`, *optional*, defaults to 49407):
+            The id of the end-of-sequence token in the vocabulary.
+
+    Example:
+
+    ```python
+    >>> from transformers import SiglipTextConfig, SiglipTextModel
+
+    >>> # Initializing a SiglipTextConfig with google/siglip-base-patch16-224 style configuration
+    >>> configuration = SiglipTextConfig()
+
+    >>> # Initializing a SiglipTextModel (with random weights) from the google/siglip-base-patch16-224 style configuration
+    >>> model = SiglipTextModel(configuration)
+
+    >>> # Accessing the model configuration
+    >>> configuration = model.config
+    ```"""
+
+    model_type = "siglip_text_model"
+
+    def __init__(
+        self,
+        vocab_size=32000,
+        hidden_size=768,
+        intermediate_size=3072,
+        num_hidden_layers=12,
+        num_attention_heads=12,
+        max_position_embeddings=64,
+        hidden_act="gelu_pytorch_tanh",
+        layer_norm_eps=1e-6,
+        attention_dropout=0.0,
+        # This differs from `CLIPTokenizer`'s default and from openai/siglip
+        # See https://github.com/huggingface/transformers/pull/24773#issuecomment-1632287538
+        pad_token_id=1,
+        bos_token_id=49406,
+        eos_token_id=49407,
+        **kwargs,
+    ):
+        super().__init__(pad_token_id=pad_token_id, bos_token_id=bos_token_id, eos_token_id=eos_token_id, **kwargs)
+
+        self.vocab_size = vocab_size
+        self.hidden_size = hidden_size
+        self.intermediate_size = intermediate_size
+        self.num_hidden_layers = num_hidden_layers
+        self.num_attention_heads = num_attention_heads
+        self.max_position_embeddings = max_position_embeddings
+        self.layer_norm_eps = layer_norm_eps
+        self.hidden_act = hidden_act
+        self.attention_dropout = attention_dropout
+
+    @classmethod
+    def from_pretrained(cls, pretrained_model_name_or_path: Union[str, os.PathLike], **kwargs) -> "PretrainedConfig":
+        cls._set_token_in_kwargs(kwargs)
+
+        config_dict, kwargs = cls.get_config_dict(pretrained_model_name_or_path, **kwargs)
+
+        # get the text config dict if we are loading from SiglipConfig
+        if config_dict.get("model_type") == "siglip":
+            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 SiglipVisionConfig(PretrainedConfig):
+    r"""
+    This is the configuration class to store the configuration of a [`SiglipVisionModel`]. It is used to instantiate a
+    Siglip vision encoder according to the specified arguments, defining the model architecture. Instantiating a
+    configuration with the defaults will yield a similar configuration to that of the vision encoder of the Siglip
+    [google/siglip-base-patch16-224](https://huggingface.co/google/siglip-base-patch16-224) 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.
+        num_channels (`int`, *optional*, defaults to 3):
+            Number of channels in the input images.
+        image_size (`int`, *optional*, defaults to 224):
+            The size (resolution) of each image.
+        patch_size (`int`, *optional*, defaults to 16):
+            The size (resolution) of each patch.
+        hidden_act (`str` or `function`, *optional*, defaults to `"gelu_pytorch_tanh"`):
+            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-06):
+            The epsilon used by the layer normalization layers.
+        attention_dropout (`float`, *optional*, defaults to 0.0):
+            The dropout ratio for the attention probabilities.
+
+    Example:
+
+    ```python
+    >>> from transformers import SiglipVisionConfig, SiglipVisionModel
+
+    >>> # Initializing a SiglipVisionConfig with google/siglip-base-patch16-224 style configuration
+    >>> configuration = SiglipVisionConfig()
+
+    >>> # Initializing a SiglipVisionModel (with random weights) from the google/siglip-base-patch16-224 style configuration
+    >>> model = SiglipVisionModel(configuration)
+
+    >>> # Accessing the model configuration
+    >>> configuration = model.config
+    ```"""
+
+    model_type = "siglip_vision_model"
+
+    def __init__(
+        self,
+        hidden_size=768,
+        intermediate_size=3072,
+        num_hidden_layers=12,
+        num_attention_heads=12,
+        num_channels=3,
+        image_size=224,
+        patch_size=16,
+        hidden_act="gelu_pytorch_tanh",
+        layer_norm_eps=1e-6,
+        attention_dropout=0.0,
+        **kwargs,
+    ):
+        super().__init__(**kwargs)
+
+        self.hidden_size = hidden_size
+        self.intermediate_size = intermediate_size
+        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.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":
+        cls._set_token_in_kwargs(kwargs)
+
+        config_dict, kwargs = cls.get_config_dict(pretrained_model_name_or_path, **kwargs)
+
+        # get the vision config dict if we are loading from SiglipConfig
+        if config_dict.get("model_type") == "siglip":
+            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 SiglipConfig(PretrainedConfig):
+    r"""
+    [`SiglipConfig`] is the configuration class to store the configuration of a [`SiglipModel`]. It is used to
+    instantiate a Siglip 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 Siglip
+    [google/siglip-base-patch16-224](https://huggingface.co/google/siglip-base-patch16-224) 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 [`SiglipTextConfig`].
+        vision_config (`dict`, *optional*):
+            Dictionary of configuration options used to initialize [`SiglipVisionConfig`].
+        kwargs (*optional*):
+            Dictionary of keyword arguments.
+
+    Example:
+
+    ```python
+    >>> from transformers import SiglipConfig, SiglipModel
+
+    >>> # Initializing a SiglipConfig with google/siglip-base-patch16-224 style configuration
+    >>> configuration = SiglipConfig()
+
+    >>> # Initializing a SiglipModel (with random weights) from the google/siglip-base-patch16-224 style configuration
+    >>> model = SiglipModel(configuration)
+
+    >>> # Accessing the model configuration
+    >>> configuration = model.config
+
+    >>> # We can also initialize a SiglipConfig from a SiglipTextConfig and a SiglipVisionConfig
+    >>> from transformers import SiglipTextConfig, SiglipVisionConfig
+
+    >>> # Initializing a SiglipText and SiglipVision configuration
+    >>> config_text = SiglipTextConfig()
+    >>> config_vision = SiglipVisionConfig()
+
+    >>> config = SiglipConfig.from_text_vision_configs(config_text, config_vision)
+    ```"""
+
+    model_type = "siglip"
+
+    def __init__(self, text_config=None, vision_config=None, **kwargs):
+        super().__init__(**kwargs)
+
+        if text_config is None:
+            text_config = {}
+            logger.info("`text_config` is `None`. Initializing the `SiglipTextConfig` with default values.")
+
+        if vision_config is None:
+            vision_config = {}
+            logger.info("`vision_config` is `None`. initializing the `SiglipVisionConfig` with default values.")
+
+        self.text_config = SiglipTextConfig(**text_config)
+        self.vision_config = SiglipVisionConfig(**vision_config)
+
+        self.initializer_factor = 1.0
+
+    @classmethod
+    def from_text_vision_configs(cls, text_config: SiglipTextConfig, vision_config: SiglipVisionConfig, **kwargs):
+        r"""
+        Instantiate a [`SiglipConfig`] (or a derived class) from siglip text model configuration and siglip vision
+        model configuration.
+
+        Returns:
+            [`SiglipConfig`]: An instance of a configuration object
+        """
+
+        return cls(text_config=text_config.to_dict(), vision_config=vision_config.to_dict(), **kwargs)