Fix inclusion of non py files in package (#21546)

* Fix inclusion of non py files in package

* No need for the **

README.md

@@ -271,7 +271,7 @@ Current number of checkpoints: ![](https://img.shields.io/endpoint?url=https://h
 🤗 Transformers currently provides the following architectures (see [here](https://huggingface.co/docs/transformers/model_summary) for a high-level summary of each them):
 
 1. **[ALBERT](https://huggingface.co/docs/transformers/model_doc/albert)** (from Google Research and the Toyota Technological Institute at Chicago) released with the paper [ALBERT: A Lite BERT for Self-supervised Learning of Language Representations](https://arxiv.org/abs/1909.11942), by Zhenzhong Lan, Mingda Chen, Sebastian Goodman, Kevin Gimpel, Piyush Sharma, Radu Soricut.
-1. **[AltCLIP](https://huggingface.co/docs/transformers/main/model_doc/altclip)** (from BAAI) released with the paper [AltCLIP: Altering the Language Encoder in CLIP for Extended Language Capabilities](https://arxiv.org/abs/2211.06679) by Chen, Zhongzhi and Liu, Guang and Zhang, Bo-Wen and Ye, Fulong and Yang, Qinghong and Wu, Ledell.
+1. **[AltCLIP](https://huggingface.co/docs/transformers/model_doc/altclip)** (from BAAI) released with the paper [AltCLIP: Altering the Language Encoder in CLIP for Extended Language Capabilities](https://arxiv.org/abs/2211.06679) by Chen, Zhongzhi and Liu, Guang and Zhang, Bo-Wen and Ye, Fulong and Yang, Qinghong and Wu, Ledell.
 1. **[Audio Spectrogram Transformer](https://huggingface.co/docs/transformers/model_doc/audio-spectrogram-transformer)** (from MIT) released with the paper [AST: Audio Spectrogram Transformer](https://arxiv.org/abs/2104.01778) by Yuan Gong, Yu-An Chung, James Glass.
 1. **[BART](https://huggingface.co/docs/transformers/model_doc/bart)** (from Facebook) released with the paper [BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension](https://arxiv.org/abs/1910.13461) by Mike Lewis, Yinhan Liu, Naman Goyal, Marjan Ghazvininejad, Abdelrahman Mohamed, Omer Levy, Ves Stoyanov and Luke Zettlemoyer.
 1. **[BARThez](https://huggingface.co/docs/transformers/model_doc/barthez)** (from École polytechnique) released with the paper [BARThez: a Skilled Pretrained French Sequence-to-Sequence Model](https://arxiv.org/abs/2010.12321) by Moussa Kamal Eddine, Antoine J.-P. Tixier, Michalis Vazirgiannis.
@@ -282,11 +282,11 @@ Current number of checkpoints: ![](https://img.shields.io/endpoint?url=https://h
 1. **[BERTweet](https://huggingface.co/docs/transformers/model_doc/bertweet)** (from VinAI Research) released with the paper [BERTweet: A pre-trained language model for English Tweets](https://aclanthology.org/2020.emnlp-demos.2/) by Dat Quoc Nguyen, Thanh Vu and Anh Tuan Nguyen.
 1. **[BigBird-Pegasus](https://huggingface.co/docs/transformers/model_doc/bigbird_pegasus)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
 1. **[BigBird-RoBERTa](https://huggingface.co/docs/transformers/model_doc/big_bird)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
-1. **[BioGpt](https://huggingface.co/docs/transformers/main/model_doc/biogpt)** (from Microsoft Research AI4Science) released with the paper [BioGPT: generative pre-trained transformer for biomedical text generation and mining](https://academic.oup.com/bib/advance-article/doi/10.1093/bib/bbac409/6713511?guestAccessKey=a66d9b5d-4f83-4017-bb52-405815c907b9) by Renqian Luo, Liai Sun, Yingce Xia, Tao Qin, Sheng Zhang, Hoifung Poon and Tie-Yan Liu.
-1. **[BiT](https://huggingface.co/docs/transformers/main/model_doc/bit)** (from Google AI) released with the paper [Big Transfer (BiT): General Visual Representation Learning](https://arxiv.org/abs/1912.11370) by Alexander Kolesnikov, Lucas Beyer, Xiaohua Zhai, Joan Puigcerver, Jessica Yung, Sylvain Gelly, Neil Houlsby.
+1. **[BioGpt](https://huggingface.co/docs/transformers/model_doc/biogpt)** (from Microsoft Research AI4Science) released with the paper [BioGPT: generative pre-trained transformer for biomedical text generation and mining](https://academic.oup.com/bib/advance-article/doi/10.1093/bib/bbac409/6713511?guestAccessKey=a66d9b5d-4f83-4017-bb52-405815c907b9) by Renqian Luo, Liai Sun, Yingce Xia, Tao Qin, Sheng Zhang, Hoifung Poon and Tie-Yan Liu.
+1. **[BiT](https://huggingface.co/docs/transformers/model_doc/bit)** (from Google AI) released with the paper [Big Transfer (BiT): General Visual Representation Learning](https://arxiv.org/abs/1912.11370) by Alexander Kolesnikov, Lucas Beyer, Xiaohua Zhai, Joan Puigcerver, Jessica Yung, Sylvain Gelly, Neil Houlsby.
 1. **[Blenderbot](https://huggingface.co/docs/transformers/model_doc/blenderbot)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
 1. **[BlenderbotSmall](https://huggingface.co/docs/transformers/model_doc/blenderbot-small)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
-1. **[BLIP](https://huggingface.co/docs/transformers/main/model_doc/blip)** (from Salesforce) released with the paper [BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generation](https://arxiv.org/abs/2201.12086) by Junnan Li, Dongxu Li, Caiming Xiong, Steven Hoi.
+1. **[BLIP](https://huggingface.co/docs/transformers/model_doc/blip)** (from Salesforce) released with the paper [BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generation](https://arxiv.org/abs/2201.12086) by Junnan Li, Dongxu Li, Caiming Xiong, Steven Hoi.
 1. **[BLOOM](https://huggingface.co/docs/transformers/model_doc/bloom)** (from BigScience workshop) released by the [BigScience Workshop](https://bigscience.huggingface.co/).
 1. **[BORT](https://huggingface.co/docs/transformers/model_doc/bort)** (from Alexa) released with the paper [Optimal Subarchitecture Extraction For BERT](https://arxiv.org/abs/2010.10499) by Adrian de Wynter and Daniel J. Perry.
 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.
@@ -316,7 +316,7 @@ Current number of checkpoints: ![](https://img.shields.io/endpoint?url=https://h
 1. **[Donut](https://huggingface.co/docs/transformers/model_doc/donut)** (from NAVER), released together with the paper [OCR-free Document Understanding Transformer](https://arxiv.org/abs/2111.15664) by Geewook Kim, Teakgyu Hong, Moonbin Yim, Jeongyeon Nam, Jinyoung Park, Jinyeong Yim, Wonseok Hwang, Sangdoo Yun, Dongyoon Han, Seunghyun Park.
 1. **[DPR](https://huggingface.co/docs/transformers/model_doc/dpr)** (from Facebook) released with the paper [Dense Passage Retrieval for Open-Domain Question Answering](https://arxiv.org/abs/2004.04906) by Vladimir Karpukhin, Barlas Oğuz, Sewon Min, Patrick Lewis, Ledell Wu, Sergey Edunov, Danqi Chen, and Wen-tau Yih.
 1. **[DPT](https://huggingface.co/docs/transformers/master/model_doc/dpt)** (from Intel Labs) released with the paper [Vision Transformers for Dense Prediction](https://arxiv.org/abs/2103.13413) by René Ranftl, Alexey Bochkovskiy, Vladlen Koltun.
-1. **[EfficientFormer](https://huggingface.co/docs/transformers/main/model_doc/efficientformer)** (from Snap Research) released with the paper [EfficientFormer: Vision Transformers at MobileNetSpeed](https://arxiv.org/abs/2206.01191) by Yanyu Li, Geng Yuan, Yang Wen, Ju Hu, Georgios Evangelidis, Sergey Tulyakov, Yanzhi Wang, Jian Ren.
+1. **[EfficientFormer](https://huggingface.co/docs/transformers/model_doc/efficientformer)** (from Snap Research) released with the paper [EfficientFormer: Vision Transformers at MobileNetSpeed](https://arxiv.org/abs/2206.01191) by Yanyu Li, Geng Yuan, Yang Wen, Ju Hu, Georgios Evangelidis, Sergey Tulyakov, Yanzhi Wang, Jian Ren.
 1. **[ELECTRA](https://huggingface.co/docs/transformers/model_doc/electra)** (from Google Research/Stanford University) released with the paper [ELECTRA: Pre-training text encoders as discriminators rather than generators](https://arxiv.org/abs/2003.10555) by Kevin Clark, Minh-Thang Luong, Quoc V. Le, Christopher D. Manning.
 1. **[EncoderDecoder](https://huggingface.co/docs/transformers/model_doc/encoder-decoder)** (from Google Research) released with the paper [Leveraging Pre-trained Checkpoints for Sequence Generation Tasks](https://arxiv.org/abs/1907.12461) by Sascha Rothe, Shashi Narayan, Aliaksei Severyn.
 1. **[ERNIE](https://huggingface.co/docs/transformers/model_doc/ernie)** (from Baidu) released with the paper [ERNIE: Enhanced Representation through Knowledge Integration](https://arxiv.org/abs/1904.09223) by Yu Sun, Shuohuan Wang, Yukun Li, Shikun Feng, Xuyi Chen, Han Zhang, Xin Tian, Danxiang Zhu, Hao Tian, Hua Wu.
@@ -326,7 +326,7 @@ Current number of checkpoints: ![](https://img.shields.io/endpoint?url=https://h
 1. **[FLAVA](https://huggingface.co/docs/transformers/model_doc/flava)** (from Facebook AI) released with the paper [FLAVA: A Foundational Language And Vision Alignment Model](https://arxiv.org/abs/2112.04482) by Amanpreet Singh, Ronghang Hu, Vedanuj Goswami, Guillaume Couairon, Wojciech Galuba, Marcus Rohrbach, and Douwe Kiela.
 1. **[FNet](https://huggingface.co/docs/transformers/model_doc/fnet)** (from Google Research) released with the paper [FNet: Mixing Tokens with Fourier Transforms](https://arxiv.org/abs/2105.03824) by James Lee-Thorp, Joshua Ainslie, Ilya Eckstein, Santiago Ontanon.
 1. **[Funnel Transformer](https://huggingface.co/docs/transformers/model_doc/funnel)** (from CMU/Google Brain) released with the paper [Funnel-Transformer: Filtering out Sequential Redundancy for Efficient Language Processing](https://arxiv.org/abs/2006.03236) by Zihang Dai, Guokun Lai, Yiming Yang, Quoc V. Le.
-1. **[GIT](https://huggingface.co/docs/transformers/main/model_doc/git)** (from Microsoft Research) released with the paper [GIT: A Generative Image-to-text Transformer for Vision and Language](https://arxiv.org/abs/2205.14100) by Jianfeng Wang, Zhengyuan Yang, Xiaowei Hu, Linjie Li, Kevin Lin, Zhe Gan, Zicheng Liu, Ce Liu, Lijuan Wang.
+1. **[GIT](https://huggingface.co/docs/transformers/model_doc/git)** (from Microsoft Research) released with the paper [GIT: A Generative Image-to-text Transformer for Vision and Language](https://arxiv.org/abs/2205.14100) by Jianfeng Wang, Zhengyuan Yang, Xiaowei Hu, Linjie Li, Kevin Lin, Zhe Gan, Zicheng Liu, Ce Liu, Lijuan Wang.
 1. **[GLPN](https://huggingface.co/docs/transformers/model_doc/glpn)** (from KAIST) released with the paper [Global-Local Path Networks for Monocular Depth Estimation with Vertical CutDepth](https://arxiv.org/abs/2201.07436) by Doyeon Kim, Woonghyun Ga, Pyungwhan Ahn, Donggyu Joo, Sehwan Chun, Junmo Kim.
 1. **[GPT](https://huggingface.co/docs/transformers/model_doc/openai-gpt)** (from OpenAI) released with the paper [Improving Language Understanding by Generative Pre-Training](https://blog.openai.com/language-unsupervised/) by Alec Radford, Karthik Narasimhan, Tim Salimans and Ilya Sutskever.
 1. **[GPT Neo](https://huggingface.co/docs/transformers/model_doc/gpt_neo)** (from EleutherAI) released in the repository [EleutherAI/gpt-neo](https://github.com/EleutherAI/gpt-neo) by Sid Black, Stella Biderman, Leo Gao, Phil Wang and Connor Leahy.
@@ -334,8 +334,8 @@ Current number of checkpoints: ![](https://img.shields.io/endpoint?url=https://h
 1. **[GPT NeoX Japanese](https://huggingface.co/docs/transformers/model_doc/gpt_neox_japanese)** (from ABEJA) released by Shinya Otani, Takayoshi Makabe, Anuj Arora, and Kyo Hattori.
 1. **[GPT-2](https://huggingface.co/docs/transformers/model_doc/gpt2)** (from OpenAI) released with the paper [Language Models are Unsupervised Multitask Learners](https://blog.openai.com/better-language-models/) by Alec Radford*, Jeffrey Wu*, Rewon Child, David Luan, Dario Amodei** and Ilya Sutskever**.
 1. **[GPT-J](https://huggingface.co/docs/transformers/model_doc/gptj)** (from EleutherAI) released in the repository [kingoflolz/mesh-transformer-jax](https://github.com/kingoflolz/mesh-transformer-jax/) by Ben Wang and Aran Komatsuzaki.
-1. **[GPT-Sw3](https://huggingface.co/docs/transformers/main/model_doc/gpt-sw3)** (from AI-Sweden) released with the paper [Lessons Learned from GPT-SW3: Building the First Large-Scale Generative Language Model for Swedish](http://www.lrec-conf.org/proceedings/lrec2022/pdf/2022.lrec-1.376.pdf) by Ariel Ekgren, Amaru Cuba Gyllensten, Evangelia Gogoulou, Alice Heiman, Severine Verlinden, Joey Öhman, Fredrik Carlsson, Magnus Sahlgren.
-1. **[Graphormer](https://huggingface.co/docs/transformers/main/model_doc/graphormer)** (from Microsoft) released with the paper [Do Transformers Really Perform Bad for Graph Representation?](https://arxiv.org/abs/2106.05234) by Chengxuan Ying, Tianle Cai, Shengjie Luo, Shuxin Zheng, Guolin Ke, Di He, Yanming Shen, Tie-Yan Liu.
+1. **[GPT-Sw3](https://huggingface.co/docs/transformers/model_doc/gpt-sw3)** (from AI-Sweden) released with the paper [Lessons Learned from GPT-SW3: Building the First Large-Scale Generative Language Model for Swedish](http://www.lrec-conf.org/proceedings/lrec2022/pdf/2022.lrec-1.376.pdf) by Ariel Ekgren, Amaru Cuba Gyllensten, Evangelia Gogoulou, Alice Heiman, Severine Verlinden, Joey Öhman, Fredrik Carlsson, Magnus Sahlgren.
+1. **[Graphormer](https://huggingface.co/docs/transformers/model_doc/graphormer)** (from Microsoft) released with the paper [Do Transformers Really Perform Bad for Graph Representation?](https://arxiv.org/abs/2106.05234) by Chengxuan Ying, Tianle Cai, Shengjie Luo, Shuxin Zheng, Guolin Ke, Di He, Yanming Shen, Tie-Yan Liu.
 1. **[GroupViT](https://huggingface.co/docs/transformers/model_doc/groupvit)** (from UCSD, NVIDIA) released with the paper [GroupViT: Semantic Segmentation Emerges from Text Supervision](https://arxiv.org/abs/2202.11094) by Jiarui Xu, Shalini De Mello, Sifei Liu, Wonmin Byeon, Thomas Breuel, Jan Kautz, Xiaolong Wang.
 1. **[Hubert](https://huggingface.co/docs/transformers/model_doc/hubert)** (from Facebook) released with the paper [HuBERT: Self-Supervised Speech Representation Learning by Masked Prediction of Hidden Units](https://arxiv.org/abs/2106.07447) by Wei-Ning Hsu, Benjamin Bolte, Yao-Hung Hubert Tsai, Kushal Lakhotia, Ruslan Salakhutdinov, Abdelrahman Mohamed.
 1. **[I-BERT](https://huggingface.co/docs/transformers/model_doc/ibert)** (from Berkeley) released with the paper [I-BERT: Integer-only BERT Quantization](https://arxiv.org/abs/2101.01321) by Sehoon Kim, Amir Gholami, Zhewei Yao, Michael W. Mahoney, Kurt Keutzer.
@@ -356,7 +356,7 @@ Current number of checkpoints: ![](https://img.shields.io/endpoint?url=https://h
 1. **[M2M100](https://huggingface.co/docs/transformers/model_doc/m2m_100)** (from Facebook) released with the paper [Beyond English-Centric Multilingual Machine Translation](https://arxiv.org/abs/2010.11125) by Angela Fan, Shruti Bhosale, Holger Schwenk, Zhiyi Ma, Ahmed El-Kishky, Siddharth Goyal, Mandeep Baines, Onur Celebi, Guillaume Wenzek, Vishrav Chaudhary, Naman Goyal, Tom Birch, Vitaliy Liptchinsky, Sergey Edunov, Edouard Grave, Michael Auli, Armand Joulin.
 1. **[MarianMT](https://huggingface.co/docs/transformers/model_doc/marian)** Machine translation models trained using [OPUS](http://opus.nlpl.eu/) data by Jörg Tiedemann. The [Marian Framework](https://marian-nmt.github.io/) is being developed by the Microsoft Translator Team.
 1. **[MarkupLM](https://huggingface.co/docs/transformers/model_doc/markuplm)** (from Microsoft Research Asia) released with the paper [MarkupLM: Pre-training of Text and Markup Language for Visually-rich Document Understanding](https://arxiv.org/abs/2110.08518) by Junlong Li, Yiheng Xu, Lei Cui, Furu Wei.
-1. **[Mask2Former](https://huggingface.co/docs/transformers/main/model_doc/mask2former)** (from FAIR and UIUC) released with the paper [Masked-attention Mask Transformer for Universal Image Segmentation](https://arxiv.org/abs/2112.01527) by Bowen Cheng, Ishan Misra, Alexander G. Schwing, Alexander Kirillov, Rohit Girdhar.
+1. **[Mask2Former](https://huggingface.co/docs/transformers/model_doc/mask2former)** (from FAIR and UIUC) released with the paper [Masked-attention Mask Transformer for Universal Image Segmentation](https://arxiv.org/abs/2112.01527) by Bowen Cheng, Ishan Misra, Alexander G. Schwing, Alexander Kirillov, Rohit Girdhar.
 1. **[MaskFormer](https://huggingface.co/docs/transformers/model_doc/maskformer)** (from Meta and UIUC) released with the paper [Per-Pixel Classification is Not All You Need for Semantic Segmentation](https://arxiv.org/abs/2107.06278) by Bowen Cheng, Alexander G. Schwing, Alexander Kirillov.
 1. **[mBART](https://huggingface.co/docs/transformers/model_doc/mbart)** (from Facebook) released with the paper [Multilingual Denoising Pre-training for Neural Machine Translation](https://arxiv.org/abs/2001.08210) by Yinhan Liu, Jiatao Gu, Naman Goyal, Xian Li, Sergey Edunov, Marjan Ghazvininejad, Mike Lewis, Luke Zettlemoyer.
 1. **[mBART-50](https://huggingface.co/docs/transformers/model_doc/mbart)** (from Facebook) released with the paper [Multilingual Translation with Extensible Multilingual Pretraining and Finetuning](https://arxiv.org/abs/2008.00401) by Yuqing Tang, Chau Tran, Xian Li, Peng-Jen Chen, Naman Goyal, Vishrav Chaudhary, Jiatao Gu, Angela Fan.
@@ -374,7 +374,7 @@ Current number of checkpoints: ![](https://img.shields.io/endpoint?url=https://h
 1. **[Nezha](https://huggingface.co/docs/transformers/model_doc/nezha)** (from Huawei Noah’s Ark Lab) released with the paper [NEZHA: Neural Contextualized Representation for Chinese Language Understanding](https://arxiv.org/abs/1909.00204) by Junqiu Wei, Xiaozhe Ren, Xiaoguang Li, Wenyong Huang, Yi Liao, Yasheng Wang, Jiashu Lin, Xin Jiang, Xiao Chen and Qun Liu.
 1. **[NLLB](https://huggingface.co/docs/transformers/model_doc/nllb)** (from Meta) released with the paper [No Language Left Behind: Scaling Human-Centered Machine Translation](https://arxiv.org/abs/2207.04672) by the NLLB team.
 1. **[Nyströmformer](https://huggingface.co/docs/transformers/model_doc/nystromformer)** (from the University of Wisconsin - Madison) released with the paper [Nyströmformer: A Nyström-Based Algorithm for Approximating Self-Attention](https://arxiv.org/abs/2102.03902) by Yunyang Xiong, Zhanpeng Zeng, Rudrasis Chakraborty, Mingxing Tan, Glenn Fung, Yin Li, Vikas Singh.
-1. **[OneFormer](https://huggingface.co/docs/transformers/main/model_doc/oneformer)** (from SHI Labs) released with the paper [OneFormer: One Transformer to Rule Universal Image Segmentation](https://arxiv.org/abs/2211.06220) by Jitesh Jain, Jiachen Li, MangTik Chiu, Ali Hassani, Nikita Orlov, Humphrey Shi.
+1. **[OneFormer](https://huggingface.co/docs/transformers/model_doc/oneformer)** (from SHI Labs) released with the paper [OneFormer: One Transformer to Rule Universal Image Segmentation](https://arxiv.org/abs/2211.06220) by Jitesh Jain, Jiachen Li, MangTik Chiu, Ali Hassani, Nikita Orlov, Humphrey Shi.
 1. **[OPT](https://huggingface.co/docs/transformers/master/model_doc/opt)** (from Meta AI) released with the paper [OPT: Open Pre-trained Transformer Language Models](https://arxiv.org/abs/2205.01068) by Susan Zhang, Stephen Roller, Naman Goyal, Mikel Artetxe, Moya Chen, Shuohui Chen et al.
 1. **[OWL-ViT](https://huggingface.co/docs/transformers/model_doc/owlvit)** (from Google AI) released with the paper [Simple Open-Vocabulary Object Detection with Vision Transformers](https://arxiv.org/abs/2205.06230) by Matthias Minderer, Alexey Gritsenko, Austin Stone, Maxim Neumann, Dirk Weissenborn, Alexey Dosovitskiy, Aravindh Mahendran, Anurag Arnab, Mostafa Dehghani, Zhuoran Shen, Xiao Wang, Xiaohua Zhai, Thomas Kipf, and Neil Houlsby.
 1. **[Pegasus](https://huggingface.co/docs/transformers/model_doc/pegasus)** (from Google) released with the paper [PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization](https://arxiv.org/abs/1912.08777) by Jingqing Zhang, Yao Zhao, Mohammad Saleh and Peter J. Liu.
@@ -392,8 +392,8 @@ Current number of checkpoints: ![](https://img.shields.io/endpoint?url=https://h
 1. **[RemBERT](https://huggingface.co/docs/transformers/model_doc/rembert)** (from Google Research) released with the paper [Rethinking embedding coupling in pre-trained language models](https://arxiv.org/abs/2010.12821) by Hyung Won Chung, Thibault Févry, Henry Tsai, M. Johnson, Sebastian Ruder.
 1. **[ResNet](https://huggingface.co/docs/transformers/model_doc/resnet)** (from Microsoft Research) released with the paper [Deep Residual Learning for Image Recognition](https://arxiv.org/abs/1512.03385) by Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun.
 1. **[RoBERTa](https://huggingface.co/docs/transformers/model_doc/roberta)** (from Facebook), released together with the paper [RoBERTa: A Robustly Optimized BERT Pretraining Approach](https://arxiv.org/abs/1907.11692) by Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, Veselin Stoyanov.
-1. **[RoBERTa-PreLayerNorm](https://huggingface.co/docs/transformers/main/model_doc/roberta-prelayernorm)** (from Facebook) released with the paper [fairseq: A Fast, Extensible Toolkit for Sequence Modeling](https://arxiv.org/abs/1904.01038) by Myle Ott, Sergey Edunov, Alexei Baevski, Angela Fan, Sam Gross, Nathan Ng, David Grangier, Michael Auli.
-1. **[RoCBert](https://huggingface.co/docs/transformers/main/model_doc/roc_bert)** (from WeChatAI) released with the paper [RoCBert: Robust Chinese Bert with Multimodal Contrastive Pretraining](https://aclanthology.org/2022.acl-long.65.pdf) by HuiSu, WeiweiShi, XiaoyuShen, XiaoZhou, TuoJi, JiaruiFang, JieZhou.
+1. **[RoBERTa-PreLayerNorm](https://huggingface.co/docs/transformers/model_doc/roberta-prelayernorm)** (from Facebook) released with the paper [fairseq: A Fast, Extensible Toolkit for Sequence Modeling](https://arxiv.org/abs/1904.01038) by Myle Ott, Sergey Edunov, Alexei Baevski, Angela Fan, Sam Gross, Nathan Ng, David Grangier, Michael Auli.
+1. **[RoCBert](https://huggingface.co/docs/transformers/model_doc/roc_bert)** (from WeChatAI) released with the paper [RoCBert: Robust Chinese Bert with Multimodal Contrastive Pretraining](https://aclanthology.org/2022.acl-long.65.pdf) by HuiSu, WeiweiShi, XiaoyuShen, XiaoZhou, TuoJi, JiaruiFang, JieZhou.
 1. **[RoFormer](https://huggingface.co/docs/transformers/model_doc/roformer)** (from ZhuiyiTechnology), released together with the paper [RoFormer: Enhanced Transformer with Rotary Position Embedding](https://arxiv.org/abs/2104.09864) by Jianlin Su and Yu Lu and Shengfeng Pan and Bo Wen and Yunfeng Liu.
 1. **[SegFormer](https://huggingface.co/docs/transformers/model_doc/segformer)** (from NVIDIA) released with the paper [SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers](https://arxiv.org/abs/2105.15203) by Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M. Alvarez, Ping Luo.
 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.
@@ -404,28 +404,28 @@ Current number of checkpoints: ![](https://img.shields.io/endpoint?url=https://h
 1. **[SqueezeBERT](https://huggingface.co/docs/transformers/model_doc/squeezebert)** (from Berkeley) released with the paper [SqueezeBERT: What can computer vision teach NLP about efficient neural networks?](https://arxiv.org/abs/2006.11316) by Forrest N. Iandola, Albert E. Shaw, Ravi Krishna, and Kurt W. Keutzer.
 1. **[Swin Transformer](https://huggingface.co/docs/transformers/model_doc/swin)** (from Microsoft) released with the paper [Swin Transformer: Hierarchical Vision Transformer using Shifted Windows](https://arxiv.org/abs/2103.14030) by Ze Liu, Yutong Lin, Yue Cao, Han Hu, Yixuan Wei, Zheng Zhang, Stephen Lin, Baining Guo.
 1. **[Swin Transformer V2](https://huggingface.co/docs/transformers/model_doc/swinv2)** (from Microsoft) released with the paper [Swin Transformer V2: Scaling Up Capacity and Resolution](https://arxiv.org/abs/2111.09883) by Ze Liu, Han Hu, Yutong Lin, Zhuliang Yao, Zhenda Xie, Yixuan Wei, Jia Ning, Yue Cao, Zheng Zhang, Li Dong, Furu Wei, Baining Guo.
-1. **[Swin2SR](https://huggingface.co/docs/transformers/main/model_doc/swin2sr)** (from University of Würzburg) released with the paper [Swin2SR: SwinV2 Transformer for Compressed Image Super-Resolution and Restoration](https://arxiv.org/abs/2209.11345) by Marcos V. Conde, Ui-Jin Choi, Maxime Burchi, Radu Timofte.
-1. **[SwitchTransformers](https://huggingface.co/docs/transformers/main/model_doc/switch_transformers)** (from Google) released with the paper [Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity](https://arxiv.org/abs/2101.03961) by William Fedus, Barret Zoph, Noam Shazeer.
+1. **[Swin2SR](https://huggingface.co/docs/transformers/model_doc/swin2sr)** (from University of Würzburg) released with the paper [Swin2SR: SwinV2 Transformer for Compressed Image Super-Resolution and Restoration](https://arxiv.org/abs/2209.11345) by Marcos V. Conde, Ui-Jin Choi, Maxime Burchi, Radu Timofte.
+1. **[SwitchTransformers](https://huggingface.co/docs/transformers/model_doc/switch_transformers)** (from Google) released with the paper [Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity](https://arxiv.org/abs/2101.03961) by William Fedus, Barret Zoph, Noam Shazeer.
 1. **[T5](https://huggingface.co/docs/transformers/model_doc/t5)** (from Google AI) released with the paper [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://arxiv.org/abs/1910.10683) by Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu.
 1. **[T5v1.1](https://huggingface.co/docs/transformers/model_doc/t5v1.1)** (from Google AI) released in the repository [google-research/text-to-text-transfer-transformer](https://github.com/google-research/text-to-text-transfer-transformer/blob/main/released_checkpoints.md#t511) by Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu.
 1. **[Table Transformer](https://huggingface.co/docs/transformers/model_doc/table-transformer)** (from Microsoft Research) released with the paper [PubTables-1M: Towards Comprehensive Table Extraction From Unstructured Documents](https://arxiv.org/abs/2110.00061) by Brandon Smock, Rohith Pesala, Robin Abraham.
 1. **[TAPAS](https://huggingface.co/docs/transformers/model_doc/tapas)** (from Google AI) released with the paper [TAPAS: Weakly Supervised Table Parsing via Pre-training](https://arxiv.org/abs/2004.02349) by Jonathan Herzig, Paweł Krzysztof Nowak, Thomas Müller, Francesco Piccinno and Julian Martin Eisenschlos.
 1. **[TAPEX](https://huggingface.co/docs/transformers/model_doc/tapex)** (from Microsoft Research) released with the paper [TAPEX: Table Pre-training via Learning a Neural SQL Executor](https://arxiv.org/abs/2107.07653) by Qian Liu, Bei Chen, Jiaqi Guo, Morteza Ziyadi, Zeqi Lin, Weizhu Chen, Jian-Guang Lou.
 1. **[Time Series Transformer](https://huggingface.co/docs/transformers/model_doc/time_series_transformer)** (from HuggingFace).
-1. **[TimeSformer](https://huggingface.co/docs/transformers/main/model_doc/timesformer)** (from Facebook) released with the paper [Is Space-Time Attention All You Need for Video Understanding?](https://arxiv.org/abs/2102.05095) by Gedas Bertasius, Heng Wang, Lorenzo Torresani.
+1. **[TimeSformer](https://huggingface.co/docs/transformers/model_doc/timesformer)** (from Facebook) released with the paper [Is Space-Time Attention All You Need for Video Understanding?](https://arxiv.org/abs/2102.05095) by Gedas Bertasius, Heng Wang, Lorenzo Torresani.
 1. **[Trajectory Transformer](https://huggingface.co/docs/transformers/model_doc/trajectory_transformers)** (from the University of California at Berkeley) released with the paper [Offline Reinforcement Learning as One Big Sequence Modeling Problem](https://arxiv.org/abs/2106.02039) by Michael Janner, Qiyang Li, Sergey Levine
 1. **[Transformer-XL](https://huggingface.co/docs/transformers/model_doc/transfo-xl)** (from Google/CMU) released with the paper [Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context](https://arxiv.org/abs/1901.02860) by Zihang Dai*, Zhilin Yang*, Yiming Yang, Jaime Carbonell, Quoc V. Le, Ruslan Salakhutdinov.
 1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (from Microsoft), released together with the paper [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) by Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei.
 1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (from Google Research) released with the paper [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) by Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
 1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (from Microsoft Research) released with the paper [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597) by Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang.
 1. **[UniSpeechSat](https://huggingface.co/docs/transformers/model_doc/unispeech-sat)** (from Microsoft Research) released with the paper [UNISPEECH-SAT: UNIVERSAL SPEECH REPRESENTATION LEARNING WITH SPEAKER AWARE PRE-TRAINING](https://arxiv.org/abs/2110.05752) by Sanyuan Chen, Yu Wu, Chengyi Wang, Zhengyang Chen, Zhuo Chen, Shujie Liu, Jian Wu, Yao Qian, Furu Wei, Jinyu Li, Xiangzhan Yu.
-1. **[UPerNet](https://huggingface.co/docs/transformers/main/model_doc/upernet)** (from Peking University) released with the paper [Unified Perceptual Parsing for Scene Understanding](https://arxiv.org/abs/1807.10221) by Tete Xiao, Yingcheng Liu, Bolei Zhou, Yuning Jiang, Jian Sun.
+1. **[UPerNet](https://huggingface.co/docs/transformers/model_doc/upernet)** (from Peking University) released with the paper [Unified Perceptual Parsing for Scene Understanding](https://arxiv.org/abs/1807.10221) by Tete Xiao, Yingcheng Liu, Bolei Zhou, Yuning Jiang, Jian Sun.
 1. **[VAN](https://huggingface.co/docs/transformers/model_doc/van)** (from Tsinghua University and Nankai University) released with the paper [Visual Attention Network](https://arxiv.org/abs/2202.09741) by Meng-Hao Guo, Cheng-Ze Lu, Zheng-Ning Liu, Ming-Ming Cheng, Shi-Min Hu.
 1. **[VideoMAE](https://huggingface.co/docs/transformers/model_doc/videomae)** (from Multimedia Computing Group, Nanjing University) released with the paper [VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training](https://arxiv.org/abs/2203.12602) by Zhan Tong, Yibing Song, Jue Wang, Limin Wang.
 1. **[ViLT](https://huggingface.co/docs/transformers/model_doc/vilt)** (from NAVER AI Lab/Kakao Enterprise/Kakao Brain) released with the paper [ViLT: Vision-and-Language Transformer Without Convolution or Region Supervision](https://arxiv.org/abs/2102.03334) by Wonjae Kim, Bokyung Son, Ildoo Kim.
 1. **[Vision Transformer (ViT)](https://huggingface.co/docs/transformers/model_doc/vit)** (from Google AI) released with the paper [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) by Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby.
 1. **[VisualBERT](https://huggingface.co/docs/transformers/model_doc/visual_bert)** (from UCLA NLP) released with the paper [VisualBERT: A Simple and Performant Baseline for Vision and Language](https://arxiv.org/pdf/1908.03557) by Liunian Harold Li, Mark Yatskar, Da Yin, Cho-Jui Hsieh, Kai-Wei Chang.
-1. **[ViT Hybrid](https://huggingface.co/docs/transformers/main/model_doc/vit_hybrid)** (from Google AI) released with the paper [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) by Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby.
+1. **[ViT Hybrid](https://huggingface.co/docs/transformers/model_doc/vit_hybrid)** (from Google AI) released with the paper [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) by Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby.
 1. **[ViTMAE](https://huggingface.co/docs/transformers/model_doc/vit_mae)** (from Meta AI) released with the paper [Masked Autoencoders Are Scalable Vision Learners](https://arxiv.org/abs/2111.06377) by Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Piotr Dollár, Ross Girshick.
 1. **[ViTMSN](https://huggingface.co/docs/transformers/model_doc/vit_msn)** (from Meta AI) released with the paper [Masked Siamese Networks for Label-Efficient Learning](https://arxiv.org/abs/2204.07141) by Mahmoud Assran, Mathilde Caron, Ishan Misra, Piotr Bojanowski, Florian Bordes, Pascal Vincent, Armand Joulin, Michael Rabbat, Nicolas Ballas.
 1. **[Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/wav2vec2)** (from Facebook AI) released with the paper [wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations](https://arxiv.org/abs/2006.11477) by Alexei Baevski, Henry Zhou, Abdelrahman Mohamed, Michael Auli.

docs/source/en/model_doc/mask2former.mdx

@@ -22,8 +22,8 @@ The abstract from the paper is the following:
 of semantics defines a task. While only the semantics of each task differ, current research focuses on designing specialized architectures for each task. We present Masked-attention Mask Transformer (Mask2Former), a new architecture capable of addressing any image segmentation task (panoptic, instance or semantic). Its key components include masked attention, which extracts localized features by constraining cross-attention within predicted mask regions. In addition to reducing the research effort by at least three times, it outperforms the best specialized architectures by a significant margin on four popular datasets. Most notably, Mask2Former sets a new state-of-the-art for panoptic segmentation (57.8 PQ on COCO), instance segmentation (50.1 AP on COCO) and semantic segmentation (57.7 mIoU on ADE20K).*
 
 Tips:
-- Mask2Former uses the same preprocessing and postprocessing steps as [MaskFormer](maskformer). Use [`MaskFormerImageProcessor`] or [`AutoImageProcessor`] to prepare images and optional targets for the model.
-- To get the final segmentation, depending on the task, you can call [`~MaskFormerImageProcessor.post_process_semantic_segmentation`] or [`~MaskFormerImageProcessor.post_process_instance_segmentation`] or [`~MaskFormerImageProcessor.post_process_panoptic_segmentation`]. All three tasks can be solved using [`Mask2FormerForUniversalSegmentation`] output, panoptic segmentation accepts an optional `label_ids_to_fuse` argument to fuse instances of the target object/s (e.g. sky) together.
+- Mask2Former uses the same preprocessing and postprocessing steps as [MaskFormer](maskformer). Use [`Mask2FormerImageProcessor`] or [`AutoImageProcessor`] to prepare images and optional targets for the model.
+- To get the final segmentation, depending on the task, you can call [`~Mask2FormerImageProcessor.post_process_semantic_segmentation`] or [`~Mask2FormerImageProcessor.post_process_instance_segmentation`] or [`~Mask2FormerImageProcessor.post_process_panoptic_segmentation`]. All three tasks can be solved using [`Mask2FormerForUniversalSegmentation`] output, panoptic segmentation accepts an optional `label_ids_to_fuse` argument to fuse instances of the target object/s (e.g. sky) together.
 
 This model was contributed by [Shivalika Singh](https://huggingface.co/shivi) and [Alara Dirik](https://huggingface.co/adirik). The original code can be found [here](https://github.com/facebookresearch/Mask2Former).
 
@@ -55,3 +55,12 @@ The resource should ideally demonstrate something new instead of duplicating an
 
 [[autodoc]] Mask2FormerForUniversalSegmentation
     - forward
+
+## Mask2FormerImageProcessor
+
+[[autodoc]] Mask2FormerImageProcessor
+    - preprocess
+    - encode_inputs
+    - post_process_semantic_segmentation
+    - post_process_instance_segmentation
+    - post_process_panoptic_segmentation

examples/flax/question-answering/run_qa.py

@@ -61,7 +61,7 @@ from utils_qa import postprocess_qa_predictions
 logger = logging.getLogger(__name__)
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 Array = Any
 Dataset = datasets.arrow_dataset.Dataset

examples/flax/text-classification/run_flax_glue.py

@@ -54,7 +54,7 @@ from transformers.utils import check_min_version, get_full_repo_name, send_examp
 
 logger = logging.getLogger(__name__)
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 Array = Any
 Dataset = datasets.arrow_dataset.Dataset

examples/flax/token-classification/run_flax_ner.py

@@ -55,7 +55,7 @@ from transformers.utils.versions import require_version
 
 logger = logging.getLogger(__name__)
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 require_version("datasets>=1.8.0", "To fix: pip install -r examples/pytorch/token-classification/requirements.txt")
 

examples/pytorch/audio-classification/run_audio_classification.py

@@ -45,7 +45,7 @@ from transformers.utils.versions import require_version
 logger = logging.getLogger(__name__)
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 require_version("datasets>=1.14.0", "To fix: pip install -r examples/pytorch/audio-classification/requirements.txt")
 

examples/pytorch/contrastive-image-text/run_clip.py

@@ -54,7 +54,7 @@ from transformers.utils.versions import require_version
 logger = logging.getLogger(__name__)
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 require_version("datasets>=1.8.0", "To fix: pip install -r examples/pytorch/contrastive-image-text/requirements.txt")
 

examples/pytorch/image-classification/run_image_classification.py

@@ -55,7 +55,7 @@ from transformers.utils.versions import require_version
 logger = logging.getLogger(__name__)
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 require_version("datasets>=1.8.0", "To fix: pip install -r examples/pytorch/image-classification/requirements.txt")
 

examples/pytorch/image-classification/run_image_classification_no_trainer.py

@@ -47,7 +47,7 @@ from transformers.utils.versions import require_version
 
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 logger = get_logger(__name__)
 

examples/pytorch/image-pretraining/run_mae.py

@@ -43,7 +43,7 @@ from transformers.utils.versions import require_version
 logger = logging.getLogger(__name__)
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 require_version("datasets>=1.8.0", "To fix: pip install -r examples/pytorch/image-pretraining/requirements.txt")
 

examples/pytorch/image-pretraining/run_mim.py

@@ -48,7 +48,7 @@ Any model supported by the AutoModelForMaskedImageModeling API can be used.
 logger = logging.getLogger(__name__)
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 require_version("datasets>=1.8.0", "To fix: pip install -r examples/pytorch/image-pretraining/requirements.txt")
 

examples/pytorch/language-modeling/run_clm.py

@@ -55,7 +55,7 @@ from transformers.utils.versions import require_version
 
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 require_version("datasets>=1.8.0", "To fix: pip install -r examples/pytorch/language-modeling/requirements.txt")
 

examples/pytorch/language-modeling/run_clm_no_trainer.py

@@ -57,7 +57,7 @@ from transformers.utils.versions import require_version
 
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 logger = get_logger(__name__)
 

examples/pytorch/language-modeling/run_mlm.py

@@ -53,7 +53,7 @@ from transformers.utils.versions import require_version
 
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 require_version("datasets>=1.8.0", "To fix: pip install -r examples/pytorch/language-modeling/requirements.txt")
 

examples/pytorch/language-modeling/run_mlm_no_trainer.py

@@ -57,7 +57,7 @@ from transformers.utils.versions import require_version
 
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 logger = get_logger(__name__)
 require_version("datasets>=1.8.0", "To fix: pip install -r examples/pytorch/language-modeling/requirements.txt")

examples/pytorch/language-modeling/run_plm.py

@@ -47,7 +47,7 @@ from transformers.utils.versions import require_version
 
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 require_version("datasets>=1.8.0", "To fix: pip install -r examples/pytorch/language-modeling/requirements.txt")
 

examples/pytorch/multiple-choice/run_swag.py

@@ -47,7 +47,7 @@ from transformers.utils import PaddingStrategy, check_min_version, send_example_
 
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 logger = logging.getLogger(__name__)
 

examples/pytorch/multiple-choice/run_swag_no_trainer.py

@@ -56,7 +56,7 @@ from transformers.utils import PaddingStrategy, check_min_version, get_full_repo
 
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 logger = get_logger(__name__)
 # You should update this to your particular problem to have better documentation of `model_type`

examples/pytorch/question-answering/run_qa.py

@@ -49,7 +49,7 @@ from utils_qa import postprocess_qa_predictions
 
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 require_version("datasets>=1.8.0", "To fix: pip install -r examples/pytorch/question-answering/requirements.txt")
 

examples/pytorch/question-answering/run_qa_beam_search.py

@@ -48,7 +48,7 @@ from utils_qa import postprocess_qa_predictions_with_beam_search
 
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 require_version("datasets>=1.8.0", "To fix: pip install -r examples/pytorch/question-answering/requirements.txt")
 

examples/pytorch/question-answering/run_qa_beam_search_no_trainer.py

@@ -56,7 +56,7 @@ from utils_qa import postprocess_qa_predictions_with_beam_search
 
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 require_version("datasets>=1.8.0", "To fix: pip install -r examples/pytorch/question-answering/requirements.txt")
 

examples/pytorch/question-answering/run_qa_no_trainer.py

@@ -57,7 +57,7 @@ from utils_qa import postprocess_qa_predictions
 
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 require_version("datasets>=1.8.0", "To fix: pip install -r examples/pytorch/question-answering/requirements.txt")
 

examples/pytorch/question-answering/run_seq2seq_qa.py

@@ -45,7 +45,7 @@ from transformers.utils.versions import require_version
 
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 require_version("datasets>=1.8.0", "To fix: pip install -r examples/pytorch/question-answering/requirements.txt")
 

examples/pytorch/semantic-segmentation/run_semantic_segmentation.py

@@ -51,7 +51,7 @@ from transformers.utils.versions import require_version
 logger = logging.getLogger(__name__)
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 require_version("datasets>=2.0.0", "To fix: pip install -r examples/pytorch/semantic-segmentation/requirements.txt")
 

examples/pytorch/semantic-segmentation/run_semantic_segmentation_no_trainer.py

@@ -50,7 +50,7 @@ from transformers.utils.versions import require_version
 
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 logger = get_logger(__name__)
 

examples/pytorch/speech-recognition/run_speech_recognition_ctc.py

@@ -50,7 +50,7 @@ from transformers.utils.versions import require_version
 
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 require_version("datasets>=1.18.0", "To fix: pip install -r examples/pytorch/speech-recognition/requirements.txt")
 

examples/pytorch/speech-recognition/run_speech_recognition_seq2seq.py

@@ -48,7 +48,7 @@ from transformers.utils.versions import require_version
 
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 require_version("datasets>=1.18.0", "To fix: pip install -r examples/pytorch/speech-recognition/requirements.txt")
 

examples/pytorch/summarization/run_summarization.py

@@ -52,7 +52,7 @@ from transformers.utils.versions import require_version
 
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 require_version("datasets>=1.8.0", "To fix: pip install -r examples/pytorch/summarization/requirements.txt")
 

examples/pytorch/summarization/run_summarization_no_trainer.py

@@ -56,7 +56,7 @@ from transformers.utils.versions import require_version
 
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 logger = get_logger(__name__)
 require_version("datasets>=1.8.0", "To fix: pip install -r examples/pytorch/summarization/requirements.txt")

examples/pytorch/text-classification/run_glue.py

@@ -48,7 +48,7 @@ from transformers.utils.versions import require_version
 
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 require_version("datasets>=1.8.0", "To fix: pip install -r examples/pytorch/text-classification/requirements.txt")
 

examples/pytorch/text-classification/run_glue_no_trainer.py

@@ -48,7 +48,7 @@ from transformers.utils.versions import require_version
 
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 logger = get_logger(__name__)
 

examples/pytorch/text-classification/run_xnli.py

@@ -48,7 +48,7 @@ from transformers.utils.versions import require_version
 
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 require_version("datasets>=1.8.0", "To fix: pip install -r examples/pytorch/text-classification/requirements.txt")
 

examples/pytorch/token-classification/run_ner.py

@@ -49,7 +49,7 @@ from transformers.utils.versions import require_version
 
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 require_version("datasets>=1.8.0", "To fix: pip install -r examples/pytorch/token-classification/requirements.txt")
 

examples/pytorch/token-classification/run_ner_no_trainer.py

@@ -55,7 +55,7 @@ from transformers.utils.versions import require_version
 
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 logger = get_logger(__name__)
 require_version("datasets>=1.8.0", "To fix: pip install -r examples/pytorch/token-classification/requirements.txt")

examples/pytorch/translation/run_translation.py

@@ -52,7 +52,7 @@ from transformers.utils.versions import require_version
 
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 require_version("datasets>=1.8.0", "To fix: pip install -r examples/pytorch/translation/requirements.txt")
 

examples/pytorch/translation/run_translation_no_trainer.py

@@ -57,7 +57,7 @@ from transformers.utils.versions import require_version
 
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 logger = get_logger(__name__)
 require_version("datasets>=1.8.0", "To fix: pip install -r examples/pytorch/translation/requirements.txt")

examples/tensorflow/multiple-choice/run_swag.py

@@ -50,7 +50,7 @@ from transformers.utils import PaddingStrategy, check_min_version, send_example_
 
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 logger = logging.getLogger(__name__)
 

examples/tensorflow/question-answering/run_qa.py

@@ -48,7 +48,7 @@ from utils_qa import postprocess_qa_predictions
 
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 logger = logging.getLogger(__name__)
 

examples/tensorflow/summarization/run_summarization.py

@@ -53,7 +53,7 @@ from transformers.utils.versions import require_version
 
 # region Checking dependencies
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 require_version("datasets>=1.8.0", "To fix: pip install -r examples/pytorch/summarization/requirements.txt")
 

examples/tensorflow/text-classification/run_glue.py

@@ -47,7 +47,7 @@ from transformers.utils import check_min_version, send_example_telemetry
 
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 task_to_keys = {
     "cola": ("sentence", None),

examples/tensorflow/translation/run_translation.py

@@ -56,7 +56,7 @@ from transformers.utils.versions import require_version
 
 # region Dependencies and constants
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.26.0.dev0")
+check_min_version("4.26.0")
 
 require_version("datasets>=1.8.0", "To fix: pip install -r examples/pytorch/summarization/requirements.txt")
 

setup.py

@@ -413,7 +413,7 @@ install_requires = [
 
 setup(
     name="transformers",
-    version="4.26.0.dev0",  # expected format is one of x.y.z.dev0, or x.y.z.rc1 or x.y.z (no to dashes, yes to dots)
+    version="4.26.1",  # expected format is one of x.y.z.dev0, or x.y.z.rc1 or x.y.z (no to dashes, yes to dots)
     author="The Hugging Face team (past and future) with the help of all our contributors (https://github.com/huggingface/transformers/graphs/contributors)",
     author_email="transformers@huggingface.co",
     description="State-of-the-art Machine Learning for JAX, PyTorch and TensorFlow",
@@ -424,7 +424,8 @@ setup(
     url="https://github.com/huggingface/transformers",
     package_dir={"": "src"},
     packages=find_packages("src"),
-    package_data={"transformers": ["py.typed", "*.cu", "*.cpp", "*.cuh", "*.h"]},
+    include_package_data=True,
+    package_data={"transformers": ["*.cu", "*.cpp", "*.cuh", "*.h", "*.pyx"]},
     zip_safe=False,
     extras_require=extras,
     entry_points={"console_scripts": ["transformers-cli=transformers.commands.transformers_cli:main"]},

src/transformers/__init__.py

@@ -22,7 +22,7 @@
 # to defer the actual importing for when the objects are requested. This way `import transformers` provides the names
 # in the namespace without actually importing anything (and especially none of the backends).
 
-__version__ = "4.26.0.dev0"
+__version__ = "4.26.1"
 
 from typing import TYPE_CHECKING
 
@@ -799,6 +799,7 @@ else:
     _import_structure["models.layoutlmv2"].extend(["LayoutLMv2FeatureExtractor", "LayoutLMv2ImageProcessor"])
     _import_structure["models.layoutlmv3"].extend(["LayoutLMv3FeatureExtractor", "LayoutLMv3ImageProcessor"])
     _import_structure["models.levit"].extend(["LevitFeatureExtractor", "LevitImageProcessor"])
+    _import_structure["models.mask2former"].append("Mask2FormerImageProcessor")
     _import_structure["models.maskformer"].extend(["MaskFormerFeatureExtractor", "MaskFormerImageProcessor"])
     _import_structure["models.mobilenet_v1"].extend(["MobileNetV1FeatureExtractor", "MobileNetV1ImageProcessor"])
     _import_structure["models.mobilenet_v2"].extend(["MobileNetV2FeatureExtractor", "MobileNetV2ImageProcessor"])
@@ -4152,6 +4153,7 @@ if TYPE_CHECKING:
         from .models.layoutlmv2 import LayoutLMv2FeatureExtractor, LayoutLMv2ImageProcessor
         from .models.layoutlmv3 import LayoutLMv3FeatureExtractor, LayoutLMv3ImageProcessor
         from .models.levit import LevitFeatureExtractor, LevitImageProcessor
+        from .models.mask2former import Mask2FormerImageProcessor
         from .models.maskformer import MaskFormerFeatureExtractor, MaskFormerImageProcessor
         from .models.mobilenet_v1 import MobileNetV1FeatureExtractor, MobileNetV1ImageProcessor
         from .models.mobilenet_v2 import MobileNetV2FeatureExtractor, MobileNetV2ImageProcessor

src/transformers/models/auto/image_processing_auto.py

@@ -55,14 +55,14 @@ IMAGE_PROCESSOR_MAPPING_NAMES = OrderedDict(
         ("dpt", "DPTImageProcessor"),
         ("efficientformer", "EfficientFormerImageProcessor"),
         ("flava", "FlavaImageProcessor"),
-        ("git", ("CLIPImageProcessor", "VideoMAEImageProcessor")),
+        ("git", "CLIPImageProcessor"),
         ("glpn", "GLPNImageProcessor"),
         ("groupvit", "CLIPImageProcessor"),
         ("imagegpt", "ImageGPTImageProcessor"),
         ("layoutlmv2", "LayoutLMv2ImageProcessor"),
         ("layoutlmv3", "LayoutLMv3ImageProcessor"),
         ("levit", "LevitImageProcessor"),
-        ("mask2former", "MaskFormerImageProcessor"),
+        ("mask2former", "Mask2FormerImageProcessor"),
         ("maskformer", "MaskFormerImageProcessor"),
         ("mobilenet_v1", "MobileNetV1ImageProcessor"),
         ("mobilenet_v2", "MobileNetV2ImageProcessor"),

src/transformers/models/longt5/modeling_longt5.py

@@ -276,6 +276,8 @@ class LongT5DenseActDense(nn.Module):
         hidden_states = self.wi(hidden_states)
         hidden_states = self.act(hidden_states)
         hidden_states = self.dropout(hidden_states)
+        if hidden_states.dtype != self.wo.weight.dtype and self.wo.weight.dtype != torch.int8:
+            hidden_states = hidden_states.to(self.wo.weight.dtype)
         hidden_states = self.wo(hidden_states)
         return hidden_states
 

src/transformers/models/mask2former/__init__.py

@@ -27,6 +27,13 @@ _import_structure = {
     ],
 }
 
+try:
+    if not is_vision_available():
+        raise OptionalDependencyNotAvailable()
+except OptionalDependencyNotAvailable:
+    pass
+else:
+    _import_structure["image_processing_mask2former"] = ["Mask2FormerImageProcessor"]
 
 try:
     if not is_torch_available():
@@ -44,6 +51,14 @@ else:
 if TYPE_CHECKING:
     from .configuration_mask2former import MASK2FORMER_PRETRAINED_CONFIG_ARCHIVE_MAP, Mask2FormerConfig
 
+    try:
+        if not is_vision_available():
+            raise OptionalDependencyNotAvailable()
+    except OptionalDependencyNotAvailable:
+        pass
+    else:
+        from .image_processing_mask2former import Mask2FormerImageProcessor
+
     try:
         if not is_torch_available():
             raise OptionalDependencyNotAvailable()

src/transformers/models/mask2former/convert_mask2former_original_pytorch_checkpoint_to_pytorch.py

@@ -33,8 +33,8 @@ from huggingface_hub import hf_hub_download
 from transformers import (
     Mask2FormerConfig,
     Mask2FormerForUniversalSegmentation,
+    Mask2FormerImageProcessor,
     Mask2FormerModel,
-    MaskFormerImageProcessor,
     SwinConfig,
 )
 from transformers.models.mask2former.modeling_mask2former import (
@@ -193,11 +193,11 @@ class OriginalMask2FormerConfigToOursConverter:
 
 
 class OriginalMask2FormerConfigToFeatureExtractorConverter:
-    def __call__(self, original_config: object) -> MaskFormerImageProcessor:
+    def __call__(self, original_config: object) -> Mask2FormerImageProcessor:
         model = original_config.MODEL
         model_input = original_config.INPUT
 
-        return MaskFormerImageProcessor(
+        return Mask2FormerImageProcessor(
             image_mean=(torch.tensor(model.PIXEL_MEAN) / 255).tolist(),
             image_std=(torch.tensor(model.PIXEL_STD) / 255).tolist(),
             size=model_input.MIN_SIZE_TEST,
@@ -847,7 +847,7 @@ class OriginalMask2FormerCheckpointToOursConverter:
 def test(
     original_model,
     our_model: Mask2FormerForUniversalSegmentation,
-    feature_extractor: MaskFormerImageProcessor,
+    feature_extractor: Mask2FormerImageProcessor,
     tolerance: float,
 ):
     with torch.no_grad():

src/transformers/models/mask2former/modeling_mask2former.py

@@ -49,6 +49,7 @@ logger = logging.get_logger(__name__)
 
 _CONFIG_FOR_DOC = "Mask2FormerConfig"
 _CHECKPOINT_FOR_DOC = "facebook/mask2former-swin-small-coco-instance"
+_IMAGE_PROCESSOR_FOR_DOC = "Mask2FormerImageProcessor"
 
 MASK2FORMER_PRETRAINED_MODEL_ARCHIVE_LIST = [
     "facebook/mask2former-swin-small-coco-instance",
@@ -194,10 +195,10 @@ class Mask2FormerForUniversalSegmentationOutput(ModelOutput):
     """
     Class for outputs of [`Mask2FormerForUniversalSegmentationOutput`].
 
-    This output can be directly passed to [`~MaskFormerImageProcessor.post_process_semantic_segmentation`] or
-    [`~MaskFormerImageProcessor.post_process_instance_segmentation`] or
-    [`~MaskFormerImageProcessor.post_process_panoptic_segmentation`] to compute final segmentation maps. Please, see
-    [`~MaskFormerImageProcessor] for details regarding usage.
+    This output can be directly passed to [`~Mask2FormerImageProcessor.post_process_semantic_segmentation`] or
+    [`~Mask2FormerImageProcessor.post_process_instance_segmentation`] or
+    [`~Mask2FormerImageProcessor.post_process_panoptic_segmentation`] to compute final segmentation maps. Please, see
+    [`~Mask2FormerImageProcessor] for details regarding usage.
 
     Args:
         loss (`torch.Tensor`, *optional*):

src/transformers/models/maskformer/image_processing_maskformer.py

@@ -1016,6 +1016,7 @@ class MaskFormerImageProcessor(BaseImageProcessor):
         overlap_mask_area_threshold: float = 0.8,
         target_sizes: Optional[List[Tuple[int, int]]] = None,
         return_coco_annotation: Optional[bool] = False,
+        return_binary_maps: Optional[bool] = False,
     ) -> List[Dict]:
         """
         Converts the output of [`MaskFormerForInstanceSegmentationOutput`] into instance segmentation predictions. Only
@@ -1034,9 +1035,11 @@ class MaskFormerImageProcessor(BaseImageProcessor):
             target_sizes (`List[Tuple]`, *optional*):
                 List of length (batch_size), where each list item (`Tuple[int, int]]`) corresponds to the requested
                 final size (height, width) of each prediction. If left to None, predictions will not be resized.
-            return_coco_annotation (`bool`, *optional*):
-                Defaults to `False`. If set to `True`, segmentation maps are returned in COCO run-length encoding (RLE)
-                format.
+            return_coco_annotation (`bool`, *optional*, defaults to `False`):
+                If set to `True`, segmentation maps are returned in COCO run-length encoding (RLE) format.
+            return_binary_maps (`bool`, *optional*, defaults to `False`):
+                If set to `True`, segmentation maps are returned as a concatenated tensor of binary segmentation maps
+                (one per detected instance).
         Returns:
             `List[Dict]`: A list of dictionaries, one per image, each dictionary containing two keys:
             - **segmentation** -- A tensor of shape `(height, width)` where each pixel represents a `segment_id` or
@@ -1047,47 +1050,73 @@ class MaskFormerImageProcessor(BaseImageProcessor):
                 - **label_id** -- An integer representing the label / semantic class id corresponding to `segment_id`.
                 - **score** -- Prediction score of segment with `segment_id`.
         """
-        class_queries_logits = outputs.class_queries_logits  # [batch_size, num_queries, num_classes+1]
-        masks_queries_logits = outputs.masks_queries_logits  # [batch_size, num_queries, height, width]
+        if return_coco_annotation and return_binary_maps:
+            raise ValueError("return_coco_annotation and return_binary_maps can not be both set to True.")
 
-        batch_size = class_queries_logits.shape[0]
-        num_labels = class_queries_logits.shape[-1] - 1
+        # [batch_size, num_queries, num_classes+1]
+        class_queries_logits = outputs.class_queries_logits
+        # [batch_size, num_queries, height, width]
+        masks_queries_logits = outputs.masks_queries_logits
 
-        mask_probs = masks_queries_logits.sigmoid()  # [batch_size, num_queries, height, width]
-
-        # Predicted label and score of each query (batch_size, num_queries)
-        pred_scores, pred_labels = nn.functional.softmax(class_queries_logits, dim=-1).max(-1)
+        device = masks_queries_logits.device
+        num_classes = class_queries_logits.shape[-1] - 1
+        num_queries = class_queries_logits.shape[-2]
 
         # Loop over items in batch size
         results: List[Dict[str, TensorType]] = []
 
-        for i in range(batch_size):
-            mask_probs_item, pred_scores_item, pred_labels_item = remove_low_and_no_objects(
-                mask_probs[i], pred_scores[i], pred_labels[i], threshold, num_labels
+        for i in range(class_queries_logits.shape[0]):
+            mask_pred = masks_queries_logits[i]
+            mask_cls = class_queries_logits[i]
+
+            scores = torch.nn.functional.softmax(mask_cls, dim=-1)[:, :-1]
+            labels = torch.arange(num_classes, device=device).unsqueeze(0).repeat(num_queries, 1).flatten(0, 1)
+
+            scores_per_image, topk_indices = scores.flatten(0, 1).topk(num_queries, sorted=False)
+            labels_per_image = labels[topk_indices]
+
+            topk_indices = topk_indices // num_classes
+            mask_pred = mask_pred[topk_indices]
+            pred_masks = (mask_pred > 0).float()
+
+            # Calculate average mask prob
+            mask_scores_per_image = (mask_pred.sigmoid().flatten(1) * pred_masks.flatten(1)).sum(1) / (
+                pred_masks.flatten(1).sum(1) + 1e-6
             )
+            pred_scores = scores_per_image * mask_scores_per_image
+            pred_classes = labels_per_image
 
-            # No mask found
-            if mask_probs_item.shape[0] <= 0:
-                height, width = target_sizes[i] if target_sizes is not None else mask_probs_item.shape[1:]
-                segmentation = torch.zeros((height, width)) - 1
-                results.append({"segmentation": segmentation, "segments_info": []})
-                continue
+            segmentation = torch.zeros(masks_queries_logits.shape[2:]) - 1
+            if target_sizes is not None:
+                segmentation = torch.zeros(target_sizes[i]) - 1
+                pred_masks = torch.nn.functional.interpolate(
+                    pred_masks.unsqueeze(0), size=target_sizes[i], mode="nearest"
+                )[0]
 
-            # Get segmentation map and segment information of batch item
-            target_size = target_sizes[i] if target_sizes is not None else None
-            segmentation, segments = compute_segments(
-                mask_probs=mask_probs_item,
-                pred_scores=pred_scores_item,
-                pred_labels=pred_labels_item,
-                mask_threshold=mask_threshold,
-                overlap_mask_area_threshold=overlap_mask_area_threshold,
-                label_ids_to_fuse=[],
-                target_size=target_size,
-            )
+            instance_maps, segments = [], []
+            current_segment_id = 0
+            for j in range(num_queries):
+                score = pred_scores[j].item()
 
-            # Return segmentation map in run-length encoding (RLE) format
-            if return_coco_annotation:
-                segmentation = convert_segmentation_to_rle(segmentation)
+                if not torch.all(pred_masks[j] == 0) and score >= threshold:
+                    segmentation[pred_masks[j] == 1] = current_segment_id
+                    segments.append(
+                        {
+                            "id": current_segment_id,
+                            "label_id": pred_classes[j].item(),
+                            "was_fused": False,
+                            "score": round(score, 6),
+                        }
+                    )
+                    current_segment_id += 1
+                    instance_maps.append(pred_masks[j])
+                    # Return segmentation map in run-length encoding (RLE) format
+                    if return_coco_annotation:
+                        segmentation = convert_segmentation_to_rle(segmentation)
+
+            # Return a concatenated tensor of binary instance maps
+            if return_binary_maps and len(instance_maps) != 0:
+                segmentation = torch.stack(instance_maps, dim=0)
 
             results.append({"segmentation": segmentation, "segments_info": segments})
         return results

src/transformers/models/mt5/modeling_mt5.py

@@ -146,6 +146,8 @@ class MT5DenseActDense(nn.Module):
         hidden_states = self.wi(hidden_states)
         hidden_states = self.act(hidden_states)
         hidden_states = self.dropout(hidden_states)
+        if hidden_states.dtype != self.wo.weight.dtype and self.wo.weight.dtype != torch.int8:
+            hidden_states = hidden_states.to(self.wo.weight.dtype)
         hidden_states = self.wo(hidden_states)
         return hidden_states
 
@@ -168,7 +170,8 @@ class MT5DenseGatedActDense(nn.Module):
 
         # To make 8bit quantization work for google/flan-t5-xxl, self.wo is kept in float32.
         # See https://github.com/huggingface/transformers/issues/20287
-        if hidden_states.dtype != self.wo.weight.dtype:
+        # we also make sure the weights are not in `int8` in case users will force `_keep_in_fp32_modules` to be `None``
+        if hidden_states.dtype != self.wo.weight.dtype and self.wo.weight.dtype != torch.int8:
             hidden_states = hidden_states.to(self.wo.weight.dtype)
 
         hidden_states = self.wo(hidden_states)

src/transformers/models/switch_transformers/modeling_switch_transformers.py

@@ -273,6 +273,8 @@ class SwitchTransformersDenseActDense(nn.Module):
         hidden_states = self.wi(hidden_states)
         hidden_states = self.act(hidden_states)
         hidden_states = self.dropout(hidden_states)
+        if hidden_states.dtype != self.wo.weight.dtype and self.wo.weight.dtype != torch.int8:
+            hidden_states = hidden_states.to(self.wo.weight.dtype)
         hidden_states = self.wo(hidden_states)
         return hidden_states
 

src/transformers/models/t5/modeling_t5.py

@@ -289,6 +289,8 @@ class T5DenseActDense(nn.Module):
         hidden_states = self.wi(hidden_states)
         hidden_states = self.act(hidden_states)
         hidden_states = self.dropout(hidden_states)
+        if hidden_states.dtype != self.wo.weight.dtype and self.wo.weight.dtype != torch.int8:
+            hidden_states = hidden_states.to(self.wo.weight.dtype)
         hidden_states = self.wo(hidden_states)
         return hidden_states
 
@@ -310,7 +312,8 @@ class T5DenseGatedActDense(nn.Module):
 
         # To make 8bit quantization work for google/flan-t5-xxl, self.wo is kept in float32.
         # See https://github.com/huggingface/transformers/issues/20287
-        if hidden_states.dtype != self.wo.weight.dtype:
+        # we also make sure the weights are not in `int8` in case users will force `_keep_in_fp32_modules` to be `None``
+        if hidden_states.dtype != self.wo.weight.dtype and self.wo.weight.dtype != torch.int8:
             hidden_states = hidden_states.to(self.wo.weight.dtype)
 
         hidden_states = self.wo(hidden_states)

src/transformers/trainer_utils.py

@@ -646,9 +646,9 @@ def find_executable_batch_size(
 
     if auto_find_batch_size:
         requires_backends(find_executable_batch_size, "accelerate")
-        import accelerate.memory_utils as mem_utils
+        from accelerate.utils import find_executable_batch_size as accelerate_find_executable_batch_size
 
-        return mem_utils.find_executable_batch_size(function=function, starting_batch_size=starting_batch_size)
+        return accelerate_find_executable_batch_size(function=function, starting_batch_size=starting_batch_size)
 
     return functools.partial(function, batch_size=starting_batch_size)
 

src/transformers/utils/dummy_vision_objects.py

@@ -269,6 +269,13 @@ class LevitImageProcessor(metaclass=DummyObject):
         requires_backends(self, ["vision"])
 
 
+class Mask2FormerImageProcessor(metaclass=DummyObject):
+    _backends = ["vision"]
+
+    def __init__(self, *args, **kwargs):
+        requires_backends(self, ["vision"])
+
+
 class MaskFormerFeatureExtractor(metaclass=DummyObject):
     _backends = ["vision"]
 

tests/mixed_int8/test_mixed_int8.py

@@ -163,6 +163,70 @@ class MixedInt8Test(BaseMixedInt8Test):
         self.assertTrue(model.decoder.block[0].layer[2].DenseReluDense.wo.weight.dtype == torch.float32)
 
 
+@require_bitsandbytes
+@require_accelerate
+@require_torch
+@require_torch_gpu
+@slow
+class MixedInt8T5Test(unittest.TestCase):
+    @classmethod
+    def setUpClass(cls):
+        cls.model_name = "t5-small"
+        cls.dense_act_model_name = "google/flan-t5-small"  # flan-t5 uses dense-act instead of dense-relu-dense
+        cls.tokenizer = AutoTokenizer.from_pretrained(cls.model_name)
+        cls.input_text = "Translate in German: Hello, my dog is cute"
+
+    def tearDown(self):
+        r"""
+        TearDown function needs to be called at the end of each test to free the GPU memory and cache, also to
+        avoid unexpected behaviors. Please see: https://discuss.pytorch.org/t/how-can-we-release-gpu-memory-cache/14530/27
+        """
+        gc.collect()
+        torch.cuda.empty_cache()
+
+    def test_inference_without_keep_in_fp32(self):
+        r"""
+        Test whether it is possible to mix both `int8` and `fp32` weights when using `keep_in_fp32_modules` correctly.
+        `flan-t5-small` uses `T5DenseGatedActDense` whereas `t5-small` uses `T5DenseReluDense`. We need to test
+        both cases.
+        """
+        from transformers import T5ForConditionalGeneration
+
+        T5ForConditionalGeneration._keep_in_fp32_modules = None
+
+        # test with `t5-small`
+        model = T5ForConditionalGeneration.from_pretrained(self.model_name, load_in_8bit=True, device_map="auto")
+        encoded_input = self.tokenizer(self.input_text, return_tensors="pt").to(0)
+        _ = model.generate(**encoded_input)
+
+        # test with `flan-t5-small`
+        model = T5ForConditionalGeneration.from_pretrained(
+            self.dense_act_model_name, load_in_8bit=True, device_map="auto"
+        )
+        encoded_input = self.tokenizer(self.input_text, return_tensors="pt").to(0)
+        _ = model.generate(**encoded_input)
+
+    def test_inference_with_keep_in_fp32(self):
+        r"""
+        Test whether it is possible to mix both `int8` and `fp32` weights when using `keep_in_fp32_modules` correctly.
+        `flan-t5-small` uses `T5DenseGatedActDense` whereas `t5-small` uses `T5DenseReluDense`. We need to test
+        both cases.
+        """
+        from transformers import T5ForConditionalGeneration
+
+        # test with `t5-small`
+        model = T5ForConditionalGeneration.from_pretrained(self.model_name, load_in_8bit=True, device_map="auto")
+        encoded_input = self.tokenizer(self.input_text, return_tensors="pt").to(0)
+        _ = model.generate(**encoded_input)
+
+        # test with `flan-t5-small`
+        model = T5ForConditionalGeneration.from_pretrained(
+            self.dense_act_model_name, load_in_8bit=True, device_map="auto"
+        )
+        encoded_input = self.tokenizer(self.input_text, return_tensors="pt").to(0)
+        _ = model.generate(**encoded_input)
+
+
 class MixedInt8ModelClassesTest(BaseMixedInt8Test):
     def setUp(self):
         super().setUp()

tests/models/mask2former/test_image_processing_mask2former.py

@@ -0,0 +1,611 @@
+# coding=utf-8
+# Copyright 2022 HuggingFace Inc.
+#
+# 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.
+
+
+import unittest
+
+import numpy as np
+from datasets import load_dataset
+
+from huggingface_hub import hf_hub_download
+from transformers.testing_utils import require_torch, require_vision
+from transformers.utils import is_torch_available, is_vision_available
+
+from ...test_image_processing_common import ImageProcessingSavingTestMixin, prepare_image_inputs
+
+
+if is_torch_available():
+    import torch
+
+    if is_vision_available():
+        from transformers import Mask2FormerImageProcessor
+        from transformers.models.mask2former.image_processing_mask2former import binary_mask_to_rle
+        from transformers.models.mask2former.modeling_mask2former import Mask2FormerForUniversalSegmentationOutput
+
+if is_vision_available():
+    from PIL import Image
+
+
+class Mask2FormerImageProcessingTester(unittest.TestCase):
+    def __init__(
+        self,
+        parent,
+        batch_size=7,
+        num_channels=3,
+        min_resolution=30,
+        max_resolution=400,
+        size=None,
+        do_resize=True,
+        do_normalize=True,
+        image_mean=[0.5, 0.5, 0.5],
+        image_std=[0.5, 0.5, 0.5],
+        num_labels=10,
+        do_reduce_labels=True,
+        ignore_index=255,
+    ):
+        self.parent = parent
+        self.batch_size = batch_size
+        self.num_channels = num_channels
+        self.min_resolution = min_resolution
+        self.max_resolution = max_resolution
+        self.do_resize = do_resize
+        self.size = {"shortest_edge": 32, "longest_edge": 1333} if size is None else size
+        self.do_normalize = do_normalize
+        self.image_mean = image_mean
+        self.image_std = image_std
+        self.size_divisor = 0
+        # for the post_process_functions
+        self.batch_size = 2
+        self.num_queries = 3
+        self.num_classes = 2
+        self.height = 3
+        self.width = 4
+        self.num_labels = num_labels
+        self.do_reduce_labels = do_reduce_labels
+        self.ignore_index = ignore_index
+
+    def prepare_image_processor_dict(self):
+        return {
+            "do_resize": self.do_resize,
+            "size": self.size,
+            "do_normalize": self.do_normalize,
+            "image_mean": self.image_mean,
+            "image_std": self.image_std,
+            "size_divisor": self.size_divisor,
+            "num_labels": self.num_labels,
+            "do_reduce_labels": self.do_reduce_labels,
+            "ignore_index": self.ignore_index,
+        }
+
+    def get_expected_values(self, image_inputs, batched=False):
+        """
+        This function computes the expected height and width when providing images to Mask2FormerImageProcessor,
+        assuming do_resize is set to True with a scalar size.
+        """
+        if not batched:
+            image = image_inputs[0]
+            if isinstance(image, Image.Image):
+                w, h = image.size
+            else:
+                h, w = image.shape[1], image.shape[2]
+            if w < h:
+                expected_height = int(self.size["shortest_edge"] * h / w)
+                expected_width = self.size["shortest_edge"]
+            elif w > h:
+                expected_height = self.size["shortest_edge"]
+                expected_width = int(self.size["shortest_edge"] * w / h)
+            else:
+                expected_height = self.size["shortest_edge"]
+                expected_width = self.size["shortest_edge"]
+
+        else:
+            expected_values = []
+            for image in image_inputs:
+                expected_height, expected_width = self.get_expected_values([image])
+                expected_values.append((expected_height, expected_width))
+            expected_height = max(expected_values, key=lambda item: item[0])[0]
+            expected_width = max(expected_values, key=lambda item: item[1])[1]
+
+        return expected_height, expected_width
+
+    def get_fake_mask2former_outputs(self):
+        return Mask2FormerForUniversalSegmentationOutput(
+            # +1 for null class
+            class_queries_logits=torch.randn((self.batch_size, self.num_queries, self.num_classes + 1)),
+            masks_queries_logits=torch.randn((self.batch_size, self.num_queries, self.height, self.width)),
+        )
+
+
+@require_torch
+@require_vision
+class Mask2FormerImageProcessingTest(ImageProcessingSavingTestMixin, unittest.TestCase):
+
+    image_processing_class = Mask2FormerImageProcessor if (is_vision_available() and is_torch_available()) else None
+
+    def setUp(self):
+        self.image_processor_tester = Mask2FormerImageProcessingTester(self)
+
+    @property
+    def image_processor_dict(self):
+        return self.image_processor_tester.prepare_image_processor_dict()
+
+    def test_image_processor_properties(self):
+        image_processing = self.image_processing_class(**self.image_processor_dict)
+        self.assertTrue(hasattr(image_processing, "image_mean"))
+        self.assertTrue(hasattr(image_processing, "image_std"))
+        self.assertTrue(hasattr(image_processing, "do_normalize"))
+        self.assertTrue(hasattr(image_processing, "do_resize"))
+        self.assertTrue(hasattr(image_processing, "size"))
+        self.assertTrue(hasattr(image_processing, "max_size"))
+        self.assertTrue(hasattr(image_processing, "ignore_index"))
+        self.assertTrue(hasattr(image_processing, "num_labels"))
+
+    def test_image_processor_from_dict_with_kwargs(self):
+        image_processor = self.image_processing_class.from_dict(self.image_processor_dict)
+        self.assertEqual(image_processor.size, {"shortest_edge": 32, "longest_edge": 1333})
+        self.assertEqual(image_processor.size_divisor, 0)
+
+        image_processor = self.image_processing_class.from_dict(
+            self.image_processor_dict, size=42, max_size=84, size_divisibility=8
+        )
+        self.assertEqual(image_processor.size, {"shortest_edge": 42, "longest_edge": 84})
+        self.assertEqual(image_processor.size_divisor, 8)
+
+    def test_batch_feature(self):
+        pass
+
+    def test_call_pil(self):
+        # Initialize image_processing
+        image_processing = self.image_processing_class(**self.image_processor_dict)
+        # create random PIL images
+        image_inputs = prepare_image_inputs(self.image_processor_tester, equal_resolution=False)
+        for image in image_inputs:
+            self.assertIsInstance(image, Image.Image)
+
+        # Test not batched input
+        encoded_images = image_processing(image_inputs[0], return_tensors="pt").pixel_values
+
+        expected_height, expected_width = self.image_processor_tester.get_expected_values(image_inputs)
+
+        self.assertEqual(
+            encoded_images.shape,
+            (1, self.image_processor_tester.num_channels, expected_height, expected_width),
+        )
+
+        # Test batched
+        expected_height, expected_width = self.image_processor_tester.get_expected_values(image_inputs, batched=True)
+
+        encoded_images = image_processing(image_inputs, return_tensors="pt").pixel_values
+        self.assertEqual(
+            encoded_images.shape,
+            (
+                self.image_processor_tester.batch_size,
+                self.image_processor_tester.num_channels,
+                expected_height,
+                expected_width,
+            ),
+        )
+
+    def test_call_numpy(self):
+        # Initialize image_processing
+        image_processing = self.image_processing_class(**self.image_processor_dict)
+        # create random numpy tensors
+        image_inputs = prepare_image_inputs(self.image_processor_tester, equal_resolution=False, numpify=True)
+        for image in image_inputs:
+            self.assertIsInstance(image, np.ndarray)
+
+        # Test not batched input
+        encoded_images = image_processing(image_inputs[0], return_tensors="pt").pixel_values
+
+        expected_height, expected_width = self.image_processor_tester.get_expected_values(image_inputs)
+
+        self.assertEqual(
+            encoded_images.shape,
+            (1, self.image_processor_tester.num_channels, expected_height, expected_width),
+        )
+
+        # Test batched
+        encoded_images = image_processing(image_inputs, return_tensors="pt").pixel_values
+
+        expected_height, expected_width = self.image_processor_tester.get_expected_values(image_inputs, batched=True)
+
+        self.assertEqual(
+            encoded_images.shape,
+            (
+                self.image_processor_tester.batch_size,
+                self.image_processor_tester.num_channels,
+                expected_height,
+                expected_width,
+            ),
+        )
+
+    def test_call_pytorch(self):
+        # Initialize image_processing
+        image_processing = self.image_processing_class(**self.image_processor_dict)
+        # create random PyTorch tensors
+        image_inputs = prepare_image_inputs(self.image_processor_tester, equal_resolution=False, torchify=True)
+        for image in image_inputs:
+            self.assertIsInstance(image, torch.Tensor)
+
+        # Test not batched input
+        encoded_images = image_processing(image_inputs[0], return_tensors="pt").pixel_values
+
+        expected_height, expected_width = self.image_processor_tester.get_expected_values(image_inputs)
+
+        self.assertEqual(
+            encoded_images.shape,
+            (1, self.image_processor_tester.num_channels, expected_height, expected_width),
+        )
+
+        # Test batched
+        encoded_images = image_processing(image_inputs, return_tensors="pt").pixel_values
+
+        expected_height, expected_width = self.image_processor_tester.get_expected_values(image_inputs, batched=True)
+
+        self.assertEqual(
+            encoded_images.shape,
+            (
+                self.image_processor_tester.batch_size,
+                self.image_processor_tester.num_channels,
+                expected_height,
+                expected_width,
+            ),
+        )
+
+    def test_equivalence_pad_and_create_pixel_mask(self):
+        # Initialize image_processings
+        image_processing_1 = self.image_processing_class(**self.image_processor_dict)
+        image_processing_2 = self.image_processing_class(
+            do_resize=False, do_normalize=False, do_rescale=False, num_labels=self.image_processor_tester.num_classes
+        )
+        # create random PyTorch tensors
+        image_inputs = prepare_image_inputs(self.image_processor_tester, equal_resolution=False, torchify=True)
+        for image in image_inputs:
+            self.assertIsInstance(image, torch.Tensor)
+
+        # Test whether the method "pad_and_return_pixel_mask" and calling the image processor return the same tensors
+        encoded_images_with_method = image_processing_1.encode_inputs(image_inputs, return_tensors="pt")
+        encoded_images = image_processing_2(image_inputs, return_tensors="pt")
+
+        self.assertTrue(
+            torch.allclose(encoded_images_with_method["pixel_values"], encoded_images["pixel_values"], atol=1e-4)
+        )
+        self.assertTrue(
+            torch.allclose(encoded_images_with_method["pixel_mask"], encoded_images["pixel_mask"], atol=1e-4)
+        )
+
+    def comm_get_image_processing_inputs(
+        self, with_segmentation_maps=False, is_instance_map=False, segmentation_type="np"
+    ):
+        image_processing = self.image_processing_class(**self.image_processor_dict)
+        # prepare image and target
+        num_labels = self.image_processor_tester.num_labels
+        annotations = None
+        instance_id_to_semantic_id = None
+        image_inputs = prepare_image_inputs(self.image_processor_tester, equal_resolution=False)
+        if with_segmentation_maps:
+            high = num_labels
+            if is_instance_map:
+                labels_expanded = list(range(num_labels)) * 2
+                instance_id_to_semantic_id = {
+                    instance_id: label_id for instance_id, label_id in enumerate(labels_expanded)
+                }
+            annotations = [
+                np.random.randint(0, high * 2, (img.size[1], img.size[0])).astype(np.uint8) for img in image_inputs
+            ]
+            if segmentation_type == "pil":
+                annotations = [Image.fromarray(annotation) for annotation in annotations]
+
+        inputs = image_processing(
+            image_inputs,
+            annotations,
+            return_tensors="pt",
+            instance_id_to_semantic_id=instance_id_to_semantic_id,
+            pad_and_return_pixel_mask=True,
+        )
+
+        return inputs
+
+    def test_init_without_params(self):
+        pass
+
+    def test_with_size_divisor(self):
+        size_divisors = [8, 16, 32]
+        weird_input_sizes = [(407, 802), (582, 1094)]
+        for size_divisor in size_divisors:
+            image_processor_dict = {**self.image_processor_dict, **{"size_divisor": size_divisor}}
+            image_processing = self.image_processing_class(**image_processor_dict)
+            for weird_input_size in weird_input_sizes:
+                inputs = image_processing([np.ones((3, *weird_input_size))], return_tensors="pt")
+                pixel_values = inputs["pixel_values"]
+                # check if divisible
+                self.assertTrue((pixel_values.shape[-1] % size_divisor) == 0)
+                self.assertTrue((pixel_values.shape[-2] % size_divisor) == 0)
+
+    def test_call_with_segmentation_maps(self):
+        def common(is_instance_map=False, segmentation_type=None):
+            inputs = self.comm_get_image_processing_inputs(
+                with_segmentation_maps=True, is_instance_map=is_instance_map, segmentation_type=segmentation_type
+            )
+
+            mask_labels = inputs["mask_labels"]
+            class_labels = inputs["class_labels"]
+            pixel_values = inputs["pixel_values"]
+
+            # check the batch_size
+            for mask_label, class_label in zip(mask_labels, class_labels):
+                self.assertEqual(mask_label.shape[0], class_label.shape[0])
+                # this ensure padding has happened
+                self.assertEqual(mask_label.shape[1:], pixel_values.shape[2:])
+
+        common()
+        common(is_instance_map=True)
+        common(is_instance_map=False, segmentation_type="pil")
+        common(is_instance_map=True, segmentation_type="pil")
+
+    def test_integration_instance_segmentation(self):
+        # load 2 images and corresponding annotations from the hub
+        repo_id = "nielsr/image-segmentation-toy-data"
+        image1 = Image.open(
+            hf_hub_download(repo_id=repo_id, filename="instance_segmentation_image_1.png", repo_type="dataset")
+        )
+        image2 = Image.open(
+            hf_hub_download(repo_id=repo_id, filename="instance_segmentation_image_2.png", repo_type="dataset")
+        )
+        annotation1 = Image.open(
+            hf_hub_download(repo_id=repo_id, filename="instance_segmentation_annotation_1.png", repo_type="dataset")
+        )
+        annotation2 = Image.open(
+            hf_hub_download(repo_id=repo_id, filename="instance_segmentation_annotation_2.png", repo_type="dataset")
+        )
+
+        # get instance segmentations and instance-to-segmentation mappings
+        def get_instance_segmentation_and_mapping(annotation):
+            instance_seg = np.array(annotation)[:, :, 1]
+            class_id_map = np.array(annotation)[:, :, 0]
+            class_labels = np.unique(class_id_map)
+
+            # create mapping between instance IDs and semantic category IDs
+            inst2class = {}
+            for label in class_labels:
+                instance_ids = np.unique(instance_seg[class_id_map == label])
+                inst2class.update({i: label for i in instance_ids})
+
+            return instance_seg, inst2class
+
+        instance_seg1, inst2class1 = get_instance_segmentation_and_mapping(annotation1)
+        instance_seg2, inst2class2 = get_instance_segmentation_and_mapping(annotation2)
+
+        # create a image processor
+        image_processing = Mask2FormerImageProcessor(reduce_labels=True, ignore_index=255, size=(512, 512))
+
+        # prepare the images and annotations
+        inputs = image_processing(
+            [image1, image2],
+            [instance_seg1, instance_seg2],
+            instance_id_to_semantic_id=[inst2class1, inst2class2],
+            return_tensors="pt",
+        )
+
+        # verify the pixel values and pixel mask
+        self.assertEqual(inputs["pixel_values"].shape, (2, 3, 512, 512))
+        self.assertEqual(inputs["pixel_mask"].shape, (2, 512, 512))
+
+        # verify the class labels
+        self.assertEqual(len(inputs["class_labels"]), 2)
+        self.assertTrue(torch.allclose(inputs["class_labels"][0], torch.tensor([30, 55])))
+        self.assertTrue(torch.allclose(inputs["class_labels"][1], torch.tensor([4, 4, 23, 55])))
+
+        # verify the mask labels
+        self.assertEqual(len(inputs["mask_labels"]), 2)
+        self.assertEqual(inputs["mask_labels"][0].shape, (2, 512, 512))
+        self.assertEqual(inputs["mask_labels"][1].shape, (4, 512, 512))
+        self.assertEquals(inputs["mask_labels"][0].sum().item(), 41527.0)
+        self.assertEquals(inputs["mask_labels"][1].sum().item(), 26259.0)
+
+    def test_integration_semantic_segmentation(self):
+        # load 2 images and corresponding semantic annotations from the hub
+        repo_id = "nielsr/image-segmentation-toy-data"
+        image1 = Image.open(
+            hf_hub_download(repo_id=repo_id, filename="semantic_segmentation_image_1.png", repo_type="dataset")
+        )
+        image2 = Image.open(
+            hf_hub_download(repo_id=repo_id, filename="semantic_segmentation_image_2.png", repo_type="dataset")
+        )
+        annotation1 = Image.open(
+            hf_hub_download(repo_id=repo_id, filename="semantic_segmentation_annotation_1.png", repo_type="dataset")
+        )
+        annotation2 = Image.open(
+            hf_hub_download(repo_id=repo_id, filename="semantic_segmentation_annotation_2.png", repo_type="dataset")
+        )
+
+        # create a image processor
+        image_processing = Mask2FormerImageProcessor(reduce_labels=True, ignore_index=255, size=(512, 512))
+
+        # prepare the images and annotations
+        inputs = image_processing(
+            [image1, image2],
+            [annotation1, annotation2],
+            return_tensors="pt",
+        )
+
+        # verify the pixel values and pixel mask
+        self.assertEqual(inputs["pixel_values"].shape, (2, 3, 512, 512))
+        self.assertEqual(inputs["pixel_mask"].shape, (2, 512, 512))
+
+        # verify the class labels
+        self.assertEqual(len(inputs["class_labels"]), 2)
+        self.assertTrue(torch.allclose(inputs["class_labels"][0], torch.tensor([2, 4, 60])))
+        self.assertTrue(torch.allclose(inputs["class_labels"][1], torch.tensor([0, 3, 7, 8, 15, 28, 30, 143])))
+
+        # verify the mask labels
+        self.assertEqual(len(inputs["mask_labels"]), 2)
+        self.assertEqual(inputs["mask_labels"][0].shape, (3, 512, 512))
+        self.assertEqual(inputs["mask_labels"][1].shape, (8, 512, 512))
+        self.assertEquals(inputs["mask_labels"][0].sum().item(), 170200.0)
+        self.assertEquals(inputs["mask_labels"][1].sum().item(), 257036.0)
+
+    def test_integration_panoptic_segmentation(self):
+        # load 2 images and corresponding panoptic annotations from the hub
+        dataset = load_dataset("nielsr/ade20k-panoptic-demo")
+        image1 = dataset["train"][0]["image"]
+        image2 = dataset["train"][1]["image"]
+        segments_info1 = dataset["train"][0]["segments_info"]
+        segments_info2 = dataset["train"][1]["segments_info"]
+        annotation1 = dataset["train"][0]["label"]
+        annotation2 = dataset["train"][1]["label"]
+
+        def rgb_to_id(color):
+            if isinstance(color, np.ndarray) and len(color.shape) == 3:
+                if color.dtype == np.uint8:
+                    color = color.astype(np.int32)
+                return color[:, :, 0] + 256 * color[:, :, 1] + 256 * 256 * color[:, :, 2]
+            return int(color[0] + 256 * color[1] + 256 * 256 * color[2])
+
+        def create_panoptic_map(annotation, segments_info):
+            annotation = np.array(annotation)
+            # convert RGB to segment IDs per pixel
+            # 0 is the "ignore" label, for which we don't need to make binary masks
+            panoptic_map = rgb_to_id(annotation)
+
+            # create mapping between segment IDs and semantic classes
+            inst2class = {segment["id"]: segment["category_id"] for segment in segments_info}
+
+            return panoptic_map, inst2class
+
+        panoptic_map1, inst2class1 = create_panoptic_map(annotation1, segments_info1)
+        panoptic_map2, inst2class2 = create_panoptic_map(annotation2, segments_info2)
+
+        # create a image processor
+        image_processing = Mask2FormerImageProcessor(ignore_index=0, do_resize=False)
+
+        # prepare the images and annotations
+        pixel_values_list = [np.moveaxis(np.array(image1), -1, 0), np.moveaxis(np.array(image2), -1, 0)]
+        inputs = image_processing.encode_inputs(
+            pixel_values_list,
+            [panoptic_map1, panoptic_map2],
+            instance_id_to_semantic_id=[inst2class1, inst2class2],
+            return_tensors="pt",
+        )
+
+        # verify the pixel values and pixel mask
+        self.assertEqual(inputs["pixel_values"].shape, (2, 3, 512, 711))
+        self.assertEqual(inputs["pixel_mask"].shape, (2, 512, 711))
+
+        # verify the class labels
+        self.assertEqual(len(inputs["class_labels"]), 2)
+        # fmt: off
+        expected_class_labels = torch.tensor([4, 17, 32, 42, 42, 42, 42, 42, 42, 42, 32, 12, 12, 12, 12, 12, 42, 42, 12, 12, 12, 42, 12, 12, 12, 12, 12, 3, 12, 12, 12, 12, 42, 42, 42, 12, 42, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 5, 12, 12, 12, 12, 12, 12, 12, 0, 43, 43, 43, 96, 43, 104, 43, 31, 125, 31, 125, 138, 87, 125, 149, 138, 125, 87, 87])  # noqa: E231
+        # fmt: on
+        self.assertTrue(torch.allclose(inputs["class_labels"][0], torch.tensor(expected_class_labels)))
+        # fmt: off
+        expected_class_labels = torch.tensor([19, 19, 19, 19, 19, 19, 19, 19, 19, 19, 19, 67, 82, 19, 19, 17, 19, 19, 19, 19, 19, 19, 19, 19, 19, 12, 12, 42, 12, 12, 12, 12, 3, 14, 12, 12, 12, 12, 12, 12, 12, 12, 14, 5, 12, 12, 0, 115, 43, 43, 115, 43, 43, 43, 8, 8, 8, 138, 138, 125, 143])  # noqa: E231
+        # fmt: on
+        self.assertTrue(torch.allclose(inputs["class_labels"][1], expected_class_labels))
+
+        # verify the mask labels
+        self.assertEqual(len(inputs["mask_labels"]), 2)
+        self.assertEqual(inputs["mask_labels"][0].shape, (79, 512, 711))
+        self.assertEqual(inputs["mask_labels"][1].shape, (61, 512, 711))
+        self.assertEquals(inputs["mask_labels"][0].sum().item(), 315193.0)
+        self.assertEquals(inputs["mask_labels"][1].sum().item(), 350747.0)
+
+    def test_binary_mask_to_rle(self):
+        fake_binary_mask = np.zeros((20, 50))
+        fake_binary_mask[0, 20:] = 1
+        fake_binary_mask[1, :15] = 1
+        fake_binary_mask[5, :10] = 1
+
+        rle = binary_mask_to_rle(fake_binary_mask)
+        self.assertEqual(len(rle), 4)
+        self.assertEqual(rle[0], 21)
+        self.assertEqual(rle[1], 45)
+
+    def test_post_process_semantic_segmentation(self):
+        fature_extractor = self.image_processing_class(num_labels=self.image_processor_tester.num_classes)
+        outputs = self.image_processor_tester.get_fake_mask2former_outputs()
+
+        segmentation = fature_extractor.post_process_semantic_segmentation(outputs)
+
+        self.assertEqual(len(segmentation), self.image_processor_tester.batch_size)
+        self.assertEqual(segmentation[0].shape, (384, 384))
+
+        target_sizes = [(1, 4) for i in range(self.image_processor_tester.batch_size)]
+        segmentation = fature_extractor.post_process_semantic_segmentation(outputs, target_sizes=target_sizes)
+
+        self.assertEqual(segmentation[0].shape, target_sizes[0])
+
+    def test_post_process_instance_segmentation(self):
+        feature_extractor = self.image_processing_class(num_labels=self.image_processor_tester.num_classes)
+        outputs = self.image_processor_tester.get_fake_mask2former_outputs()
+        segmentation = feature_extractor.post_process_instance_segmentation(outputs, threshold=0)
+
+        self.assertTrue(len(segmentation) == self.image_processor_tester.batch_size)
+        for el in segmentation:
+            self.assertTrue("segmentation" in el)
+            self.assertTrue("segments_info" in el)
+            self.assertEqual(type(el["segments_info"]), list)
+            self.assertEqual(el["segmentation"].shape, (384, 384))
+
+        segmentation = feature_extractor.post_process_instance_segmentation(
+            outputs, threshold=0, return_binary_maps=True
+        )
+
+        self.assertTrue(len(segmentation) == self.image_processor_tester.batch_size)
+        for el in segmentation:
+            self.assertTrue("segmentation" in el)
+            self.assertTrue("segments_info" in el)
+            self.assertEqual(type(el["segments_info"]), list)
+            self.assertEqual(len(el["segmentation"].shape), 3)
+            self.assertEqual(el["segmentation"].shape[1:], (384, 384))
+
+    def test_post_process_panoptic_segmentation(self):
+        image_processing = self.image_processing_class(num_labels=self.image_processor_tester.num_classes)
+        outputs = self.image_processor_tester.get_fake_mask2former_outputs()
+        segmentation = image_processing.post_process_panoptic_segmentation(outputs, threshold=0)
+
+        self.assertTrue(len(segmentation) == self.image_processor_tester.batch_size)
+        for el in segmentation:
+            self.assertTrue("segmentation" in el)
+            self.assertTrue("segments_info" in el)
+            self.assertEqual(type(el["segments_info"]), list)
+            self.assertEqual(el["segmentation"].shape, (384, 384))
+
+    def test_post_process_label_fusing(self):
+        image_processor = self.image_processing_class(num_labels=self.image_processor_tester.num_classes)
+        outputs = self.image_processor_tester.get_fake_mask2former_outputs()
+
+        segmentation = image_processor.post_process_panoptic_segmentation(
+            outputs, threshold=0, mask_threshold=0, overlap_mask_area_threshold=0
+        )
+        unfused_segments = [el["segments_info"] for el in segmentation]
+
+        fused_segmentation = image_processor.post_process_panoptic_segmentation(
+            outputs, threshold=0, mask_threshold=0, overlap_mask_area_threshold=0, label_ids_to_fuse={1}
+        )
+        fused_segments = [el["segments_info"] for el in fused_segmentation]
+
+        for el_unfused, el_fused in zip(unfused_segments, fused_segments):
+            if len(el_unfused) == 0:
+                self.assertEqual(len(el_unfused), len(el_fused))
+                continue
+
+            # Get number of segments to be fused
+            fuse_targets = [1 for el in el_unfused if el["label_id"] in {1}]
+            num_to_fuse = 0 if len(fuse_targets) == 0 else sum(fuse_targets) - 1
+            # Expected number of segments after fusing
+            expected_num_segments = max([el["id"] for el in el_unfused]) - num_to_fuse
+            num_segments_fused = max([el["id"] for el in el_fused])
+            self.assertEqual(num_segments_fused, expected_num_segments)

tests/models/mask2former/test_modeling_mask2former.py

@@ -34,7 +34,7 @@ if is_torch_available():
     from transformers import Mask2FormerForUniversalSegmentation, Mask2FormerModel
 
     if is_vision_available():
-        from transformers import MaskFormerImageProcessor
+        from transformers import Mask2FormerImageProcessor
 
 if is_vision_available():
     from PIL import Image
@@ -325,7 +325,7 @@ class Mask2FormerModelIntegrationTest(unittest.TestCase):
 
     @cached_property
     def default_feature_extractor(self):
-        return MaskFormerImageProcessor.from_pretrained(self.model_checkpoints) if is_vision_available() else None
+        return Mask2FormerImageProcessor.from_pretrained(self.model_checkpoints) if is_vision_available() else None
 
     def test_inference_no_head(self):
         model = Mask2FormerModel.from_pretrained(self.model_checkpoints).to(torch_device)

tests/models/maskformer/test_image_processing_maskformer.py

@@ -576,6 +576,34 @@ class MaskFormerImageProcessingTest(ImageProcessingSavingTestMixin, unittest.Tes
 
         self.assertEqual(segmentation[0].shape, target_sizes[0])
 
+    def test_post_process_instance_segmentation(self):
+        feature_extractor = self.image_processing_class(num_labels=self.image_processor_tester.num_classes)
+        outputs = self.image_processor_tester.get_fake_maskformer_outputs()
+        segmentation = feature_extractor.post_process_instance_segmentation(outputs, threshold=0)
+
+        self.assertTrue(len(segmentation) == self.image_processor_tester.batch_size)
+        for el in segmentation:
+            self.assertTrue("segmentation" in el)
+            self.assertTrue("segments_info" in el)
+            self.assertEqual(type(el["segments_info"]), list)
+            self.assertEqual(
+                el["segmentation"].shape, (self.image_processor_tester.height, self.image_processor_tester.width)
+            )
+
+        segmentation = feature_extractor.post_process_instance_segmentation(
+            outputs, threshold=0, return_binary_maps=True
+        )
+
+        self.assertTrue(len(segmentation) == self.image_processor_tester.batch_size)
+        for el in segmentation:
+            self.assertTrue("segmentation" in el)
+            self.assertTrue("segments_info" in el)
+            self.assertEqual(type(el["segments_info"]), list)
+            self.assertEqual(len(el["segmentation"].shape), 3)
+            self.assertEqual(
+                el["segmentation"].shape[1:], (self.image_processor_tester.height, self.image_processor_tester.width)
+            )
+
     def test_post_process_panoptic_segmentation(self):
         image_processing = self.image_processing_class(num_labels=self.image_processor_tester.num_classes)
         outputs = self.image_processor_tester.get_fake_maskformer_outputs()