[GPTNeoX] Fix BC issue with 4.36 (#28602)

* fix dtype issue

* add a test

* update copied from mentions

* nits

* fixup

* fix copies

* Apply suggestions from code review

.circleci/TROUBLESHOOT.md

@@ -1,6 +1,6 @@
 # Troubleshooting
 
-This is a document explaining how to deal with various issues on Circle-CI. The entries may include actually solutions or pointers to Issues that cover those.
+This is a document explaining how to deal with various issues on Circle-CI. The entries may include actual solutions or pointers to Issues that cover those.
 
 ## Circle CI
 

.circleci/config.yml

@@ -209,6 +209,7 @@ jobs:
             - run: python utils/update_metadata.py --check-only
             - run: python utils/check_task_guides.py
             - run: python utils/check_docstrings.py
+            - run: python utils/check_support_list.py
 
 workflows:
     version: 2

.circleci/create_circleci_config.py

@@ -470,7 +470,7 @@ exotic_models_job = CircleCIJob(
         "pip install -U --upgrade-strategy eager 'git+https://github.com/facebookresearch/detectron2.git'",
         "sudo apt install tesseract-ocr",
         "pip install -U --upgrade-strategy eager pytesseract",
-        "pip install -U --upgrade-strategy eager natten",
+        "pip install -U --upgrade-strategy eager 'natten<0.15.0'",
         "pip install -U --upgrade-strategy eager python-Levenshtein",
         "pip install -U --upgrade-strategy eager opencv-python",
         "pip install -U --upgrade-strategy eager nltk",
@@ -514,7 +514,8 @@ doc_test_job = CircleCIJob(
         "pip install -U --upgrade-strategy eager -e .[dev]",
         "pip install -U --upgrade-strategy eager -e git+https://github.com/huggingface/accelerate@main#egg=accelerate",
         "pip install --upgrade --upgrade-strategy eager pytest pytest-sugar",
-        "pip install -U --upgrade-strategy eager natten",
+        "pip install -U --upgrade-strategy eager 'natten<0.15.0'",
+        "pip install -U --upgrade-strategy eager g2p-en",
         "find -name __pycache__ -delete",
         "find . -name \*.pyc -delete",
         # Add an empty file to keep the test step running correctly even no file is selected to be tested.

.github/workflows/TROUBLESHOOT.md

@@ -1,6 +1,6 @@
 # Troubleshooting
 
-This is a document explaining how to deal with various issues on github-actions self-hosted CI. The entries may include actually solutions or pointers to Issues that cover those.
+This is a document explaining how to deal with various issues on github-actions self-hosted CI. The entries may include actual solutions or pointers to Issues that cover those.
 
 ## GitHub Actions (self-hosted CI)
 

.github/workflows/build-past-ci-docker-images.yml

@@ -15,7 +15,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        version: ["1.13", "1.12", "1.11", "1.10"]
+        version: ["1.13", "1.12", "1.11"]
     runs-on: ubuntu-22.04
     steps:
       -

.github/workflows/check_tiny_models.yml

@@ -36,7 +36,7 @@ jobs:
           pip install --upgrade pip
           python -m pip install -U .[sklearn,torch,testing,sentencepiece,torch-speech,vision,timm,video,tf-cpu]
           pip install tensorflow_probability
-          python -m pip install -U natten
+          python -m pip install -U 'natten<0.15.0'
 
       - name: Create all tiny models (locally)
         run: |
@@ -62,7 +62,7 @@ jobs:
           path: reports/tests_pipelines
 
       - name: Create + Upload tiny models for new model architecture(s)
-        run: | 
+        run: |
           python utils/update_tiny_models.py --num_workers 2
 
       - name: Full report

.github/workflows/self-nightly-past-ci-caller.yml

@@ -56,21 +56,10 @@ jobs:
       sha: ${{ github.sha }}
     secrets: inherit
 
-  run_past_ci_pytorch_1-10:
-    name: PyTorch 1.10
-    if: (cancelled() != true) && ((github.event_name == 'schedule') || ((github.event_name == 'push') && startsWith(github.ref_name, 'run_past_ci')))
-    needs: [run_past_ci_pytorch_1-11]
-    uses: ./.github/workflows/self-past.yml
-    with:
-      framework: pytorch
-      version: "1.10"
-      sha: ${{ github.sha }}
-    secrets: inherit
-
   run_past_ci_tensorflow_2-11:
     name: TensorFlow 2.11
     if: (cancelled() != true) && ((github.event_name == 'push') && startsWith(github.ref_name, 'run_past_ci'))
-    needs: [run_past_ci_pytorch_1-10]
+    needs: [run_past_ci_pytorch_1-11]
     uses: ./.github/workflows/self-past.yml
     with:
       framework: tensorflow

.github/workflows/self-nightly-scheduled.yml

@@ -16,6 +16,7 @@ env:
   OMP_NUM_THREADS: 8
   MKL_NUM_THREADS: 8
   RUN_SLOW: yes
+  HF_HUB_READ_TOKEN: ${{ secrets.HF_HUB_READ_TOKEN }}
   SIGOPT_API_TOKEN: ${{ secrets.SIGOPT_API_TOKEN }}
   TF_FORCE_GPU_ALLOW_GROWTH: true
   RUN_PT_TF_CROSS_TESTS: 1

.github/workflows/self-past.yml

@@ -27,6 +27,7 @@ env:
   OMP_NUM_THREADS: 8
   MKL_NUM_THREADS: 8
   RUN_SLOW: yes
+  HF_HUB_READ_TOKEN: ${{ secrets.HF_HUB_READ_TOKEN }}
   SIGOPT_API_TOKEN: ${{ secrets.SIGOPT_API_TOKEN }}
   TF_FORCE_GPU_ALLOW_GROWTH: true
   RUN_PT_TF_CROSS_TESTS: 1

.github/workflows/self-push-amd.yml

@@ -15,6 +15,7 @@ env:
   PYTEST_TIMEOUT: 60
   TF_FORCE_GPU_ALLOW_GROWTH: true
   RUN_PT_TF_CROSS_TESTS: 1
+  HF_HUB_READ_TOKEN: ${{ secrets.HF_HUB_READ_TOKEN }}
 
 jobs:
   check_runner_status:

.github/workflows/self-push-caller.yml

@@ -25,7 +25,7 @@ jobs:
         
         - name: Get changed files
           id: changed-files
-          uses: tj-actions/changed-files@v22.2
+          uses: tj-actions/changed-files@v41
         
         - name: Was setup changed 
           id: was_changed

.github/workflows/self-scheduled-amd.yml

@@ -16,6 +16,7 @@ env:
   OMP_NUM_THREADS: 8
   MKL_NUM_THREADS: 8
   RUN_SLOW: yes
+  HF_HUB_READ_TOKEN: ${{ secrets.HF_HUB_READ_TOKEN }}
   SIGOPT_API_TOKEN: ${{ secrets.SIGOPT_API_TOKEN }}
 
 

.github/workflows/self-scheduled.yml

@@ -20,6 +20,9 @@ env:
   OMP_NUM_THREADS: 8
   MKL_NUM_THREADS: 8
   RUN_SLOW: yes
+  # For gated repositories, we still need to agree to share information on the Hub repo. page in order to get access.
+  # This token is created under the bot `hf-transformers-bot`.
+  HF_HUB_READ_TOKEN: ${{ secrets.HF_HUB_READ_TOKEN }}
   SIGOPT_API_TOKEN: ${{ secrets.SIGOPT_API_TOKEN }}
   TF_FORCE_GPU_ALLOW_GROWTH: true
   RUN_PT_TF_CROSS_TESTS: 1

Makefile

@@ -5,12 +5,14 @@ export PYTHONPATH = src
 
 check_dirs := examples tests src utils
 
+exclude_folders := examples/research_projects
+
 modified_only_fixup:
 	$(eval modified_py_files := $(shell python utils/get_modified_files.py $(check_dirs)))
 	@if test -n "$(modified_py_files)"; then \
 		echo "Checking/fixing $(modified_py_files)"; \
-		ruff check $(modified_py_files) --fix; \
-		ruff format $(modified_py_files);\
+		ruff check $(modified_py_files) --fix --exclude $(exclude_folders); \
+		ruff format $(modified_py_files) --exclude $(exclude_folders);\
 	else \
 		echo "No library .py files were modified"; \
 	fi
@@ -44,6 +46,7 @@ repo-consistency:
 	python utils/update_metadata.py --check-only
 	python utils/check_task_guides.py
 	python utils/check_docstrings.py
+	python utils/check_support_list.py
 
 # this target runs checks on all files
 
@@ -64,8 +67,8 @@ extra_style_checks:
 # this target runs checks on all files and potentially modifies some of them
 
 style:
-	ruff check $(check_dirs) setup.py conftest.py --fix
-	ruff format $(check_dirs) setup.py conftest.py
+	ruff check $(check_dirs) setup.py conftest.py --fix --exclude $(exclude_folders)
+	ruff format $(check_dirs) setup.py conftest.py --exclude $(exclude_folders)
 	${MAKE} autogenerate_code
 	${MAKE} extra_style_checks
 

README.md

@@ -228,7 +228,7 @@ The model itself is a regular [Pytorch `nn.Module`](https://pytorch.org/docs/sta
 1. Lower compute costs, smaller carbon footprint:
     - Researchers can share trained models instead of always retraining.
     - Practitioners can reduce compute time and production costs.
-    - Dozens of architectures with over 60,000 pretrained models across all modalities.
+    - Dozens of architectures with over 400,000 pretrained models across all modalities.
 
 1. Choose the right framework for every part of a model's lifetime:
     - Train state-of-the-art models in 3 lines of code.
@@ -250,7 +250,7 @@ The model itself is a regular [Pytorch `nn.Module`](https://pytorch.org/docs/sta
 
 ### With pip
 
-This repository is tested on Python 3.8+, Flax 0.4.1+, PyTorch 1.10+, and TensorFlow 2.6+.
+This repository is tested on Python 3.8+, Flax 0.4.1+, PyTorch 1.11+, and TensorFlow 2.6+.
 
 You should install 🤗 Transformers in a [virtual environment](https://docs.python.org/3/library/venv.html). If you're unfamiliar with Python virtual environments, check out the [user guide](https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/).
 
@@ -269,14 +269,14 @@ If you'd like to play with the examples or need the bleeding edge of the code an
 
 ### With conda
 
-Since Transformers version v4.0.0, we now have a conda channel: `huggingface`.
-
 🤗 Transformers can be installed using conda as follows:
 
 ```shell script
-conda install -c huggingface transformers
+conda install conda-forge::transformers
 ```
 
+> **_NOTE:_** Installing `transformers` from the `huggingface` channel is deprecated.
+
 Follow the installation pages of Flax, PyTorch or TensorFlow to see how to install them with conda.
 
 > **_NOTE:_**  On Windows, you may be prompted to activate Developer Mode in order to benefit from caching. If this is not an option for you, please let us know in [this issue](https://github.com/huggingface/huggingface_hub/issues/1062).
@@ -358,6 +358,7 @@ Current number of checkpoints: ![](https://img.shields.io/endpoint?url=https://h
 1. **[ErnieM](https://huggingface.co/docs/transformers/model_doc/ernie_m)** (from Baidu) released with the paper [ERNIE-M: Enhanced Multilingual Representation by Aligning Cross-lingual Semantics with Monolingual Corpora](https://arxiv.org/abs/2012.15674) by Xuan Ouyang, Shuohuan Wang, Chao Pang, Yu Sun, Hao Tian, Hua Wu, Haifeng Wang.
 1. **[ESM](https://huggingface.co/docs/transformers/model_doc/esm)** (from Meta AI) are transformer protein language models.  **ESM-1b** was released with the paper [Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences](https://www.pnas.org/content/118/15/e2016239118) by Alexander Rives, Joshua Meier, Tom Sercu, Siddharth Goyal, Zeming Lin, Jason Liu, Demi Guo, Myle Ott, C. Lawrence Zitnick, Jerry Ma, and Rob Fergus. **ESM-1v** was released with the paper [Language models enable zero-shot prediction of the effects of mutations on protein function](https://doi.org/10.1101/2021.07.09.450648) by Joshua Meier, Roshan Rao, Robert Verkuil, Jason Liu, Tom Sercu and Alexander Rives. **ESM-2 and ESMFold** were released with the paper [Language models of protein sequences at the scale of evolution enable accurate structure prediction](https://doi.org/10.1101/2022.07.20.500902) by Zeming Lin, Halil Akin, Roshan Rao, Brian Hie, Zhongkai Zhu, Wenting Lu, Allan dos Santos Costa, Maryam Fazel-Zarandi, Tom Sercu, Sal Candido, Alexander Rives.
 1. **[Falcon](https://huggingface.co/docs/transformers/model_doc/falcon)** (from Technology Innovation Institute) by Almazrouei, Ebtesam and Alobeidli, Hamza and Alshamsi, Abdulaziz and Cappelli, Alessandro and Cojocaru, Ruxandra and Debbah, Merouane and Goffinet, Etienne and Heslow, Daniel and Launay, Julien and Malartic, Quentin and Noune, Badreddine and Pannier, Baptiste and Penedo, Guilherme.
+1. **[FastSpeech2Conformer](model_doc/fastspeech2_conformer)** (from ESPnet) released with the paper [Recent Developments On Espnet Toolkit Boosted By Conformer](https://arxiv.org/abs/2010.13956) by Pengcheng Guo, Florian Boyer, Xuankai Chang, Tomoki Hayashi, Yosuke Higuchi, Hirofumi Inaguma, Naoyuki Kamo, Chenda Li, Daniel Garcia-Romero, Jiatong Shi, Jing Shi, Shinji Watanabe, Kun Wei, Wangyou Zhang, and Yuekai Zhang.
 1. **[FLAN-T5](https://huggingface.co/docs/transformers/model_doc/flan-t5)** (from Google AI) released in the repository [google-research/t5x](https://github.com/google-research/t5x/blob/main/docs/models.md#flan-t5-checkpoints) by Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Eric Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Albert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdhery, Sharan Narang, Gaurav Mishra, Adams Yu, Vincent Zhao, Yanping Huang, Andrew Dai, Hongkun Yu, Slav Petrov, Ed H. Chi, Jeff Dean, Jacob Devlin, Adam Roberts, Denny Zhou, Quoc V. Le, and Jason Wei
 1. **[FLAN-UL2](https://huggingface.co/docs/transformers/model_doc/flan-ul2)** (from Google AI) released in the repository [google-research/t5x](https://github.com/google-research/t5x/blob/main/docs/models.md#flan-ul2-checkpoints) by Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Eric Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Albert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdhery, Sharan Narang, Gaurav Mishra, Adams Yu, Vincent Zhao, Yanping Huang, Andrew Dai, Hongkun Yu, Slav Petrov, Ed H. Chi, Jeff Dean, Jacob Devlin, Adam Roberts, Denny Zhou, Quoc V. Le, and Jason Wei
 1. **[FlauBERT](https://huggingface.co/docs/transformers/model_doc/flaubert)** (from CNRS) released with the paper [FlauBERT: Unsupervised Language Model Pre-training for French](https://arxiv.org/abs/1912.05372) by Hang Le, Loïc Vial, Jibril Frej, Vincent Segonne, Maximin Coavoux, Benjamin Lecouteux, Alexandre Allauzen, Benoît Crabbé, Laurent Besacier, Didier Schwab.
@@ -457,6 +458,7 @@ Current number of checkpoints: ![](https://img.shields.io/endpoint?url=https://h
 1. **[ProphetNet](https://huggingface.co/docs/transformers/model_doc/prophetnet)** (from Microsoft Research) released with the paper [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) by Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou.
 1. **[PVT](https://huggingface.co/docs/transformers/model_doc/pvt)** (from Nanjing University, The University of Hong Kong etc.) released with the paper [Pyramid Vision Transformer: A Versatile Backbone for Dense Prediction without Convolutions](https://arxiv.org/pdf/2102.12122.pdf) by Wenhai Wang, Enze Xie, Xiang Li, Deng-Ping Fan, Kaitao Song, Ding Liang, Tong Lu, Ping Luo, Ling Shao.
 1. **[QDQBert](https://huggingface.co/docs/transformers/model_doc/qdqbert)** (from NVIDIA) released with the paper [Integer Quantization for Deep Learning Inference: Principles and Empirical Evaluation](https://arxiv.org/abs/2004.09602) by Hao Wu, Patrick Judd, Xiaojie Zhang, Mikhail Isaev and Paulius Micikevicius.
+1. **[Qwen2](https://huggingface.co/docs/transformers/model_doc/qwen2)** (from the Qwen team, Alibaba Group) released with the paper [Qwen Technical Report](https://arxiv.org/abs/2309.16609) by Jinze Bai, Shuai Bai, Yunfei Chu, Zeyu Cui, Kai Dang, Xiaodong Deng, Yang Fan, Wenbin Ge, Yu Han, Fei Huang, Binyuan Hui, Luo Ji, Mei Li, Junyang Lin, Runji Lin, Dayiheng Liu, Gao Liu, Chengqiang Lu, Keming Lu, Jianxin Ma, Rui Men, Xingzhang Ren, Xuancheng Ren, Chuanqi Tan, Sinan Tan, Jianhong Tu, Peng Wang, Shijie Wang, Wei Wang, Shengguang Wu, Benfeng Xu, Jin Xu, An Yang, Hao Yang, Jian Yang, Shusheng Yang, Yang Yao, Bowen Yu, Hongyi Yuan, Zheng Yuan, Jianwei Zhang, Xingxuan Zhang, Yichang Zhang, Zhenru Zhang, Chang Zhou, Jingren Zhou, Xiaohuan Zhou and Tianhang Zhu.
 1. **[RAG](https://huggingface.co/docs/transformers/model_doc/rag)** (from Facebook) released with the paper [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://arxiv.org/abs/2005.11401) by Patrick Lewis, Ethan Perez, Aleksandara Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, Sebastian Riedel, Douwe Kiela.
 1. **[REALM](https://huggingface.co/docs/transformers/model_doc/realm.html)** (from Google Research) released with the paper [REALM: Retrieval-Augmented Language Model Pre-Training](https://arxiv.org/abs/2002.08909) by Kelvin Guu, Kenton Lee, Zora Tung, Panupong Pasupat and Ming-Wei Chang.
 1. **[Reformer](https://huggingface.co/docs/transformers/model_doc/reformer)** (from Google Research) released with the paper [Reformer: The Efficient Transformer](https://arxiv.org/abs/2001.04451) by Nikita Kitaev, Łukasz Kaiser, Anselm Levskaya.
@@ -474,6 +476,7 @@ Current number of checkpoints: ![](https://img.shields.io/endpoint?url=https://h
 1. **[Segment Anything](https://huggingface.co/docs/transformers/model_doc/sam)** (from Meta AI) released with the paper [Segment Anything](https://arxiv.org/pdf/2304.02643v1.pdf) by Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alex Berg, Wan-Yen Lo, Piotr Dollar, Ross Girshick.
 1. **[SEW](https://huggingface.co/docs/transformers/model_doc/sew)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
 1. **[SEW-D](https://huggingface.co/docs/transformers/model_doc/sew_d)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
+1. **[SigLIP](https://huggingface.co/docs/transformers/model_doc/siglip)** (from Google AI) released with the paper [Sigmoid Loss for Language Image Pre-Training](https://arxiv.org/abs/2303.15343) by Xiaohua Zhai, Basil Mustafa, Alexander Kolesnikov, Lucas Beyer.
 1. **[SpeechT5](https://huggingface.co/docs/transformers/model_doc/speecht5)** (from Microsoft Research) released with the paper [SpeechT5: Unified-Modal Encoder-Decoder Pre-Training for Spoken Language Processing](https://arxiv.org/abs/2110.07205) by Junyi Ao, Rui Wang, Long Zhou, Chengyi Wang, Shuo Ren, Yu Wu, Shujie Liu, Tom Ko, Qing Li, Yu Zhang, Zhihua Wei, Yao Qian, Jinyu Li, Furu Wei.
 1. **[SpeechToTextTransformer](https://huggingface.co/docs/transformers/model_doc/speech_to_text)** (from Facebook), released together with the paper [fairseq S2T: Fast Speech-to-Text Modeling with fairseq](https://arxiv.org/abs/2010.05171) by Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Dmytro Okhonko, Juan Pino.
 1. **[SpeechToTextTransformer2](https://huggingface.co/docs/transformers/model_doc/speech_to_text_2)** (from Facebook), released together with the paper [Large-Scale Self- and Semi-Supervised Learning for Speech Translation](https://arxiv.org/abs/2104.06678) by Changhan Wang, Anne Wu, Juan Pino, Alexei Baevski, Michael Auli, Alexis Conneau.
@@ -516,6 +519,7 @@ Current number of checkpoints: ![](https://img.shields.io/endpoint?url=https://h
 1. **[VITS](https://huggingface.co/docs/transformers/model_doc/vits)** (from Kakao Enterprise) released with the paper [Conditional Variational Autoencoder with Adversarial Learning for End-to-End Text-to-Speech](https://arxiv.org/abs/2106.06103) by Jaehyeon Kim, Jungil Kong, Juhee Son.
 1. **[ViViT](https://huggingface.co/docs/transformers/model_doc/vivit)** (from Google Research) released with the paper [ViViT: A Video Vision Transformer](https://arxiv.org/abs/2103.15691) by Anurag Arnab, Mostafa Dehghani, Georg Heigold, Chen Sun, Mario Lučić, Cordelia Schmid.
 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.
+1. **[Wav2Vec2-BERT](https://huggingface.co/docs/transformers/model_doc/wav2vec2-bert)** (from Meta AI) released with the paper [Seamless: Multilingual Expressive and Streaming Speech Translation](https://ai.meta.com/research/publications/seamless-multilingual-expressive-and-streaming-speech-translation/) by the Seamless Communication team.
 1. **[Wav2Vec2-Conformer](https://huggingface.co/docs/transformers/model_doc/wav2vec2-conformer)** (from Facebook AI) released with the paper [FAIRSEQ S2T: Fast Speech-to-Text Modeling with FAIRSEQ](https://arxiv.org/abs/2010.05171) by Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Sravya Popuri, Dmytro Okhonko, Juan Pino.
 1. **[Wav2Vec2Phoneme](https://huggingface.co/docs/transformers/model_doc/wav2vec2_phoneme)** (from Facebook AI) released with the paper [Simple and Effective Zero-shot Cross-lingual Phoneme Recognition](https://arxiv.org/abs/2109.11680) by Qiantong Xu, Alexei Baevski, Michael Auli.
 1. **[WavLM](https://huggingface.co/docs/transformers/model_doc/wavlm)** (from Microsoft Research) released with the paper [WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing](https://arxiv.org/abs/2110.13900) by Sanyuan Chen, Chengyi Wang, Zhengyang Chen, Yu Wu, Shujie Liu, Zhuo Chen, Jinyu Li, Naoyuki Kanda, Takuya Yoshioka, Xiong Xiao, Jian Wu, Long Zhou, Shuo Ren, Yanmin Qian, Yao Qian, Jian Wu, Michael Zeng, Furu Wei.

README_es.md

@@ -225,7 +225,7 @@ El modelo en si es un [Pytorch `nn.Module`](https://pytorch.org/docs/stable/nn.h
 
 ### Con pip
 
-Este repositorio está probado en Python 3.8+, Flax 0.4.1+, PyTorch 1.10+ y TensorFlow 2.6+.
+Este repositorio está probado en Python 3.8+, Flax 0.4.1+, PyTorch 1.11+ y TensorFlow 2.6+.
 
 Deberías instalar 🤗 Transformers en un [ambiente virtual](https://docs.python.org/3/library/venv.html). Si no estas familiarizado con los entornos virtuales de Python, consulta la [guía de usuario](https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/).
 
@@ -244,14 +244,14 @@ Si deseas jugar con los ejemplos o necesitas la última versión del código y n
 
 ### Con conda
 
-Desde la versión v4.0.0 de Transformers, ahora tenemos un canal conda: `huggingface`.
-
 🤗 Transformers se puede instalar usando conda de la siguiente manera:
 
 ```shell script
-conda install -c huggingface transformers
+conda install conda-forge::transformers
 ```
 
+> **_NOTA:_** Instalar `transformers` desde el canal `huggingface` está obsoleto.
+
 Sigue las páginas de instalación de Flax, PyTorch o TensorFlow para ver cómo instalarlos con conda.
 
 > **_NOTA:_**  En Windows, es posible que se le pida que active el modo de desarrollador para beneficiarse del almacenamiento en caché. Si esta no es una opción para usted, háganoslo saber en [esta issue](https://github.com/huggingface/huggingface_hub/issues/1062).
@@ -296,7 +296,7 @@ Número actual de puntos de control: ![](https://img.shields.io/endpoint?url=htt
 1. **[CLAP](https://huggingface.co/docs/transformers/model_doc/clap)** (from LAION-AI) released with the paper [Large-scale Contrastive Language-Audio Pretraining with Feature Fusion and Keyword-to-Caption Augmentation](https://arxiv.org/abs/2211.06687) by Yusong Wu, Ke Chen, Tianyu Zhang, Yuchen Hui, Taylor Berg-Kirkpatrick, Shlomo Dubnov.
 1. **[CLIP](https://huggingface.co/docs/transformers/model_doc/clip)** (from OpenAI) released with the paper [Learning Transferable Visual Models From Natural Language Supervision](https://arxiv.org/abs/2103.00020) by Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, Ilya Sutskever.
 1. **[CLIPSeg](https://huggingface.co/docs/transformers/model_doc/clipseg)** (from University of Göttingen) released with the paper [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003) by Timo Lüddecke and Alexander Ecker.
-1. **[CLVP](https://huggingface.co/docs/transformers/model_doc/clvp)** released with the paper [Better speech synthesis through scaling](https://arxiv.org/abs/2305.07243) by James Betker. 
+1. **[CLVP](https://huggingface.co/docs/transformers/model_doc/clvp)** released with the paper [Better speech synthesis through scaling](https://arxiv.org/abs/2305.07243) by James Betker.
 1. **[CodeGen](https://huggingface.co/docs/transformers/model_doc/codegen)** (from Salesforce) released with the paper [A Conversational Paradigm for Program Synthesis](https://arxiv.org/abs/2203.13474) by Erik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu, Huan Wang, Yingbo Zhou, Silvio Savarese, Caiming Xiong.
 1. **[CodeLlama](https://huggingface.co/docs/transformers/model_doc/llama_code)** (from MetaAI) released with the paper [Code Llama: Open Foundation Models for Code](https://ai.meta.com/research/publications/code-llama-open-foundation-models-for-code/) by Baptiste Rozière, Jonas Gehring, Fabian Gloeckle, Sten Sootla, Itai Gat, Xiaoqing Ellen Tan, Yossi Adi, Jingyu Liu, Tal Remez, Jérémy Rapin, Artyom Kozhevnikov, Ivan Evtimov, Joanna Bitton, Manish Bhatt, Cristian Canton Ferrer, Aaron Grattafiori, Wenhan Xiong, Alexandre Défossez, Jade Copet, Faisal Azhar, Hugo Touvron, Louis Martin, Nicolas Usunier, Thomas Scialom, Gabriel Synnaeve.
 1. **[Conditional DETR](https://huggingface.co/docs/transformers/model_doc/conditional_detr)** (from Microsoft Research Asia) released with the paper [Conditional DETR for Fast Training Convergence](https://arxiv.org/abs/2108.06152) by Depu Meng, Xiaokang Chen, Zejia Fan, Gang Zeng, Houqiang Li, Yuhui Yuan, Lei Sun, Jingdong Wang.
@@ -333,6 +333,7 @@ Número actual de puntos de control: ![](https://img.shields.io/endpoint?url=htt
 1. **[ErnieM](https://huggingface.co/docs/transformers/model_doc/ernie_m)** (from Baidu) released with the paper [ERNIE-M: Enhanced Multilingual Representation by Aligning Cross-lingual Semantics with Monolingual Corpora](https://arxiv.org/abs/2012.15674) by Xuan Ouyang, Shuohuan Wang, Chao Pang, Yu Sun, Hao Tian, Hua Wu, Haifeng Wang.
 1. **[ESM](https://huggingface.co/docs/transformers/model_doc/esm)** (from Meta AI) are transformer protein language models.  **ESM-1b** was released with the paper [Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences](https://www.pnas.org/content/118/15/e2016239118) by Alexander Rives, Joshua Meier, Tom Sercu, Siddharth Goyal, Zeming Lin, Jason Liu, Demi Guo, Myle Ott, C. Lawrence Zitnick, Jerry Ma, and Rob Fergus. **ESM-1v** was released with the paper [Language models enable zero-shot prediction of the effects of mutations on protein function](https://doi.org/10.1101/2021.07.09.450648) by Joshua Meier, Roshan Rao, Robert Verkuil, Jason Liu, Tom Sercu and Alexander Rives. **ESM-2** was released with the paper [Language models of protein sequences at the scale of evolution enable accurate structure prediction](https://doi.org/10.1101/2022.07.20.500902) by Zeming Lin, Halil Akin, Roshan Rao, Brian Hie, Zhongkai Zhu, Wenting Lu, Allan dos Santos Costa, Maryam Fazel-Zarandi, Tom Sercu, Sal Candido, Alexander Rives.
 1. **[Falcon](https://huggingface.co/docs/transformers/model_doc/falcon)** (from Technology Innovation Institute) by Almazrouei, Ebtesam and Alobeidli, Hamza and Alshamsi, Abdulaziz and Cappelli, Alessandro and Cojocaru, Ruxandra and Debbah, Merouane and Goffinet, Etienne and Heslow, Daniel and Launay, Julien and Malartic, Quentin and Noune, Badreddine and Pannier, Baptiste and Penedo, Guilherme.
+1. **[FastSpeech2Conformer](model_doc/fastspeech2_conformer)** (from ESPnet) released with the paper [Recent Developments On Espnet Toolkit Boosted By Conformer](https://arxiv.org/abs/2010.13956) by Pengcheng Guo, Florian Boyer, Xuankai Chang, Tomoki Hayashi, Yosuke Higuchi, Hirofumi Inaguma, Naoyuki Kamo, Chenda Li, Daniel Garcia-Romero, Jiatong Shi, Jing Shi, Shinji Watanabe, Kun Wei, Wangyou Zhang, and Yuekai Zhang.
 1. **[FLAN-T5](https://huggingface.co/docs/transformers/model_doc/flan-t5)** (from Google AI) released in the repository [google-research/t5x](https://github.com/google-research/t5x/blob/main/docs/models.md#flan-t5-checkpoints) by Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Eric Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Albert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdhery, Sharan Narang, Gaurav Mishra, Adams Yu, Vincent Zhao, Yanping Huang, Andrew Dai, Hongkun Yu, Slav Petrov, Ed H. Chi, Jeff Dean, Jacob Devlin, Adam Roberts, Denny Zhou, Quoc V. Le, and Jason Wei
 1. **[FLAN-UL2](https://huggingface.co/docs/transformers/model_doc/flan-ul2)** (from Google AI) released in the repository [google-research/t5x](https://github.com/google-research/t5x/blob/main/docs/models.md#flan-ul2-checkpoints) by Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Eric Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Albert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdhery, Sharan Narang, Gaurav Mishra, Adams Yu, Vincent Zhao, Yanping Huang, Andrew Dai, Hongkun Yu, Slav Petrov, Ed H. Chi, Jeff Dean, Jacob Devlin, Adam Roberts, Denny Zhou, Quoc V. Le, and Jason Wei
 1. **[FlauBERT](https://huggingface.co/docs/transformers/model_doc/flaubert)** (from CNRS) released with the paper [FlauBERT: Unsupervised Language Model Pre-training for French](https://arxiv.org/abs/1912.05372) by Hang Le, Loïc Vial, Jibril Frej, Vincent Segonne, Maximin Coavoux, Benjamin Lecouteux, Alexandre Allauzen, Benoît Crabbé, Laurent Besacier, Didier Schwab.
@@ -357,7 +358,7 @@ Número actual de puntos de control: ![](https://img.shields.io/endpoint?url=htt
 1. **[HerBERT](https://huggingface.co/docs/transformers/model_doc/herbert)** (from Allegro.pl, AGH University of Science and Technology) released with the paper [KLEJ: Comprehensive Benchmark for Polish Language Understanding](https://www.aclweb.org/anthology/2020.acl-main.111.pdf) by Piotr Rybak, Robert Mroczkowski, Janusz Tracz, Ireneusz Gawlik.
 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.
-1. **[IDEFICS](https://huggingface.co/docs/transformers/model_doc/idefics)** (from HuggingFace) released with the paper [OBELICS: An Open Web-Scale Filtered Dataset of Interleaved Image-Text Documents](https://huggingface.co/papers/2306.16527) by Hugo Laurençon, Lucile Saulnier, Léo Tronchon, Stas Bekman, Amanpreet Singh, Anton Lozhkov, Thomas Wang, Siddharth Karamcheti, Alexander M. Rush, Douwe Kiela, Matthieu Cord, Victor Sanh. 
+1. **[IDEFICS](https://huggingface.co/docs/transformers/model_doc/idefics)** (from HuggingFace) released with the paper [OBELICS: An Open Web-Scale Filtered Dataset of Interleaved Image-Text Documents](https://huggingface.co/papers/2306.16527) by Hugo Laurençon, Lucile Saulnier, Léo Tronchon, Stas Bekman, Amanpreet Singh, Anton Lozhkov, Thomas Wang, Siddharth Karamcheti, Alexander M. Rush, Douwe Kiela, Matthieu Cord, Victor Sanh.
 1. **[ImageGPT](https://huggingface.co/docs/transformers/model_doc/imagegpt)** (from OpenAI) released with the paper [Generative Pretraining from Pixels](https://openai.com/blog/image-gpt/) by Mark Chen, Alec Radford, Rewon Child, Jeffrey Wu, Heewoo Jun, David Luan, Ilya Sutskever.
 1. **[Informer](https://huggingface.co/docs/transformers/model_doc/informer)** (from Beihang University, UC Berkeley, Rutgers University, SEDD Company) released with the paper [Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecasting](https://arxiv.org/abs/2012.07436) by Haoyi Zhou, Shanghang Zhang, Jieqi Peng, Shuai Zhang, Jianxin Li, Hui Xiong, and Wancai Zhang.
 1. **[InstructBLIP](https://huggingface.co/docs/transformers/model_doc/instructblip)** (from Salesforce) released with the paper [InstructBLIP: Towards General-purpose Vision-Language Models with Instruction Tuning](https://arxiv.org/abs/2305.06500) by Wenliang Dai, Junnan Li, Dongxu Li, Anthony Meng Huat Tiong, Junqi Zhao, Weisheng Wang, Boyang Li, Pascale Fung, Steven Hoi.
@@ -391,8 +392,8 @@ Número actual de puntos de control: ![](https://img.shields.io/endpoint?url=htt
 1. **[Megatron-BERT](https://huggingface.co/docs/transformers/model_doc/megatron-bert)** (from NVIDIA) released with the paper [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) by Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro.
 1. **[Megatron-GPT2](https://huggingface.co/docs/transformers/model_doc/megatron_gpt2)** (from NVIDIA) released with the paper [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) by Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro.
 1. **[MGP-STR](https://huggingface.co/docs/transformers/model_doc/mgp-str)** (from Alibaba Research) released with the paper [Multi-Granularity Prediction for Scene Text Recognition](https://arxiv.org/abs/2209.03592) by Peng Wang, Cheng Da, and Cong Yao.
-1. **[Mistral](https://huggingface.co/docs/transformers/model_doc/mistral)** (from Mistral AI) by The Mistral AI team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed.. 
-1. **[Mixtral](https://huggingface.co/docs/transformers/model_doc/mixtral)** (from Mistral AI) by The [Mistral AI](https://mistral.ai) team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed. 
+1. **[Mistral](https://huggingface.co/docs/transformers/model_doc/mistral)** (from Mistral AI) by The Mistral AI team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed..
+1. **[Mixtral](https://huggingface.co/docs/transformers/model_doc/mixtral)** (from Mistral AI) by The [Mistral AI](https://mistral.ai) team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed.
 1. **[mLUKE](https://huggingface.co/docs/transformers/model_doc/mluke)** (from Studio Ousia) released with the paper [mLUKE: The Power of Entity Representations in Multilingual Pretrained Language Models](https://arxiv.org/abs/2110.08151) by Ryokan Ri, Ikuya Yamada, and Yoshimasa Tsuruoka.
 1. **[MMS](https://huggingface.co/docs/transformers/model_doc/mms)** (from Facebook) released with the paper [Scaling Speech Technology to 1,000+ Languages](https://arxiv.org/abs/2305.13516) by Vineel Pratap, Andros Tjandra, Bowen Shi, Paden Tomasello, Arun Babu, Sayani Kundu, Ali Elkahky, Zhaoheng Ni, Apoorv Vyas, Maryam Fazel-Zarandi, Alexei Baevski, Yossi Adi, Xiaohui Zhang, Wei-Ning Hsu, Alexis Conneau, Michael Auli.
 1. **[MobileBERT](https://huggingface.co/docs/transformers/model_doc/mobilebert)** (from CMU/Google Brain) released with the paper [MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices](https://arxiv.org/abs/2004.02984) by Zhiqing Sun, Hongkun Yu, Xiaodan Song, Renjie Liu, Yiming Yang, and Denny Zhou.
@@ -428,10 +429,11 @@ Número actual de puntos de control: ![](https://img.shields.io/endpoint?url=htt
 1. **[Pix2Struct](https://huggingface.co/docs/transformers/model_doc/pix2struct)** (from Google) released with the paper [Pix2Struct: Screenshot Parsing as Pretraining for Visual Language Understanding](https://arxiv.org/abs/2210.03347) by Kenton Lee, Mandar Joshi, Iulia Turc, Hexiang Hu, Fangyu Liu, Julian Eisenschlos, Urvashi Khandelwal, Peter Shaw, Ming-Wei Chang, Kristina Toutanova.
 1. **[PLBart](https://huggingface.co/docs/transformers/model_doc/plbart)** (from UCLA NLP) released with the paper [Unified Pre-training for Program Understanding and Generation](https://arxiv.org/abs/2103.06333) by Wasi Uddin Ahmad, Saikat Chakraborty, Baishakhi Ray, Kai-Wei Chang.
 1. **[PoolFormer](https://huggingface.co/docs/transformers/model_doc/poolformer)** (from Sea AI Labs) released with the paper [MetaFormer is Actually What You Need for Vision](https://arxiv.org/abs/2111.11418) by Yu, Weihao and Luo, Mi and Zhou, Pan and Si, Chenyang and Zhou, Yichen and Wang, Xinchao and Feng, Jiashi and Yan, Shuicheng.
-1. **[Pop2Piano](https://huggingface.co/docs/transformers/model_doc/pop2piano)** released with the paper [Pop2Piano : Pop Audio-based Piano Cover Generation](https://arxiv.org/abs/2211.00895) by Jongho Choi, Kyogu Lee. 
+1. **[Pop2Piano](https://huggingface.co/docs/transformers/model_doc/pop2piano)** released with the paper [Pop2Piano : Pop Audio-based Piano Cover Generation](https://arxiv.org/abs/2211.00895) by Jongho Choi, Kyogu Lee.
 1. **[ProphetNet](https://huggingface.co/docs/transformers/model_doc/prophetnet)** (from Microsoft Research) released with the paper [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) by Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou.
 1. **[PVT](https://huggingface.co/docs/transformers/model_doc/pvt)** (from Nanjing University, The University of Hong Kong etc.) released with the paper [Pyramid Vision Transformer: A Versatile Backbone for Dense Prediction without Convolutions](https://arxiv.org/pdf/2102.12122.pdf) by Wenhai Wang, Enze Xie, Xiang Li, Deng-Ping Fan, Kaitao Song, Ding Liang, Tong Lu, Ping Luo, Ling Shao.
 1. **[QDQBert](https://huggingface.co/docs/transformers/model_doc/qdqbert)** (from NVIDIA) released with the paper [Integer Quantization for Deep Learning Inference: Principles and Empirical Evaluation](https://arxiv.org/abs/2004.09602) by Hao Wu, Patrick Judd, Xiaojie Zhang, Mikhail Isaev and Paulius Micikevicius.
+1. **[Qwen2](https://huggingface.co/docs/transformers/model_doc/qwen2)** (from the Qwen team, Alibaba Group) released with the paper [Qwen Technical Report](https://arxiv.org/abs/2309.16609) by Jinze Bai, Shuai Bai, Yunfei Chu, Zeyu Cui, Kai Dang, Xiaodong Deng, Yang Fan, Wenbin Ge, Yu Han, Fei Huang, Binyuan Hui, Luo Ji, Mei Li, Junyang Lin, Runji Lin, Dayiheng Liu, Gao Liu, Chengqiang Lu, Keming Lu, Jianxin Ma, Rui Men, Xingzhang Ren, Xuancheng Ren, Chuanqi Tan, Sinan Tan, Jianhong Tu, Peng Wang, Shijie Wang, Wei Wang, Shengguang Wu, Benfeng Xu, Jin Xu, An Yang, Hao Yang, Jian Yang, Shusheng Yang, Yang Yao, Bowen Yu, Hongyi Yuan, Zheng Yuan, Jianwei Zhang, Xingxuan Zhang, Yichang Zhang, Zhenru Zhang, Chang Zhou, Jingren Zhou, Xiaohuan Zhou and Tianhang Zhu.
 1. **[RAG](https://huggingface.co/docs/transformers/model_doc/rag)** (from Facebook) released with the paper [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://arxiv.org/abs/2005.11401) by Patrick Lewis, Ethan Perez, Aleksandara Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, Sebastian Riedel, Douwe Kiela.
 1. **[REALM](https://huggingface.co/docs/transformers/model_doc/realm.html)** (from Google Research) released with the paper [REALM: Retrieval-Augmented Language Model Pre-Training](https://arxiv.org/abs/2002.08909) by Kelvin Guu, Kenton Lee, Zora Tung, Panupong Pasupat and Ming-Wei Chang.
 1. **[Reformer](https://huggingface.co/docs/transformers/model_doc/reformer)** (from Google Research) released with the paper [Reformer: The Efficient Transformer](https://arxiv.org/abs/2001.04451) by Nikita Kitaev, Łukasz Kaiser, Anselm Levskaya.
@@ -449,6 +451,7 @@ Número actual de puntos de control: ![](https://img.shields.io/endpoint?url=htt
 1. **[Segment Anything](https://huggingface.co/docs/transformers/model_doc/sam)** (from Meta AI) released with the paper [Segment Anything](https://arxiv.org/pdf/2304.02643v1.pdf) by Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alex Berg, Wan-Yen Lo, Piotr Dollar, Ross Girshick.
 1. **[SEW](https://huggingface.co/docs/transformers/model_doc/sew)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
 1. **[SEW-D](https://huggingface.co/docs/transformers/model_doc/sew_d)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
+1. **[SigLIP](https://huggingface.co/docs/transformers/model_doc/siglip)** (from Google AI) released with the paper [Sigmoid Loss for Language Image Pre-Training](https://arxiv.org/abs/2303.15343) by Xiaohua Zhai, Basil Mustafa, Alexander Kolesnikov, Lucas Beyer.
 1. **[SpeechT5](https://huggingface.co/docs/transformers/model_doc/speecht5)** (from Microsoft Research) released with the paper [SpeechT5: Unified-Modal Encoder-Decoder Pre-Training for Spoken Language Processing](https://arxiv.org/abs/2110.07205) by Junyi Ao, Rui Wang, Long Zhou, Chengyi Wang, Shuo Ren, Yu Wu, Shujie Liu, Tom Ko, Qing Li, Yu Zhang, Zhihua Wei, Yao Qian, Jinyu Li, Furu Wei.
 1. **[SpeechToTextTransformer](https://huggingface.co/docs/transformers/model_doc/speech_to_text)** (from Facebook), released together with the paper [fairseq S2T: Fast Speech-to-Text Modeling with fairseq](https://arxiv.org/abs/2010.05171) by Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Dmytro Okhonko, Juan Pino.
 1. **[SpeechToTextTransformer2](https://huggingface.co/docs/transformers/model_doc/speech_to_text_2)** (from Facebook), released together with the paper [Large-Scale Self- and Semi-Supervised Learning for Speech Translation](https://arxiv.org/abs/2104.06678) by Changhan Wang, Anne Wu, Juan Pino, Alexei Baevski, Michael Auli, Alexis Conneau.
@@ -475,7 +478,7 @@ Número actual de puntos de control: ![](https://img.shields.io/endpoint?url=htt
 1. **[UMT5](https://huggingface.co/docs/transformers/model_doc/umt5)** (from Google Research) released with the paper [UniMax: Fairer and More Effective Language Sampling for Large-Scale Multilingual Pretraining](https://openreview.net/forum?id=kXwdL1cWOAi) by Hyung Won Chung, Xavier Garcia, Adam Roberts, Yi Tay, Orhan Firat, Sharan Narang, Noah Constant.
 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. **[UnivNet](https://huggingface.co/docs/transformers/model_doc/univnet)** (from Kakao Corporation) released with the paper [UnivNet: A Neural Vocoder with Multi-Resolution Spectrogram Discriminators for High-Fidelity Waveform Generation](https://arxiv.org/abs/2106.07889) by Won Jang, Dan Lim, Jaesam Yoon, Bongwan Kim, and Juntae Kim. 
+1. **[UnivNet](https://huggingface.co/docs/transformers/model_doc/univnet)** (from Kakao Corporation) released with the paper [UnivNet: A Neural Vocoder with Multi-Resolution Spectrogram Discriminators for High-Fidelity Waveform Generation](https://arxiv.org/abs/2106.07889) by Won Jang, Dan Lim, Jaesam Yoon, Bongwan Kim, and Juntae Kim.
 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.
@@ -491,6 +494,7 @@ Número actual de puntos de control: ![](https://img.shields.io/endpoint?url=htt
 1. **[VITS](https://huggingface.co/docs/transformers/model_doc/vits)** (from Kakao Enterprise) released with the paper [Conditional Variational Autoencoder with Adversarial Learning for End-to-End Text-to-Speech](https://arxiv.org/abs/2106.06103) by Jaehyeon Kim, Jungil Kong, Juhee Son.
 1. **[ViViT](https://huggingface.co/docs/transformers/model_doc/vivit)** (from Google Research) released with the paper [ViViT: A Video Vision Transformer](https://arxiv.org/abs/2103.15691) by Anurag Arnab, Mostafa Dehghani, Georg Heigold, Chen Sun, Mario Lučić, Cordelia Schmid.
 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.
+1. **[Wav2Vec2-BERT](https://huggingface.co/docs/transformers/model_doc/wav2vec2-bert)** (from Meta AI) released with the paper [Seamless: Multilingual Expressive and Streaming Speech Translation](https://ai.meta.com/research/publications/seamless-multilingual-expressive-and-streaming-speech-translation/) by the Seamless Communication team.
 1. **[Wav2Vec2-Conformer](https://huggingface.co/docs/transformers/model_doc/wav2vec2-conformer)** (from Facebook AI) released with the paper [FAIRSEQ S2T: Fast Speech-to-Text Modeling with FAIRSEQ](https://arxiv.org/abs/2010.05171) by Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Sravya Popuri, Dmytro Okhonko, Juan Pino.
 1. **[Wav2Vec2Phoneme](https://huggingface.co/docs/transformers/model_doc/wav2vec2_phoneme)** (from Facebook AI) released with the paper [Simple and Effective Zero-shot Cross-lingual Phoneme Recognition](https://arxiv.org/abs/2109.11680) by Qiantong Xu, Alexei Baevski, Michael Auli.
 1. **[WavLM](https://huggingface.co/docs/transformers/model_doc/wavlm)** (from Microsoft Research) released with the paper [WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing](https://arxiv.org/abs/2110.13900) by Sanyuan Chen, Chengyi Wang, Zhengyang Chen, Yu Wu, Shujie Liu, Zhuo Chen, Jinyu Li, Naoyuki Kanda, Takuya Yoshioka, Xiong Xiao, Jian Wu, Long Zhou, Shuo Ren, Yanmin Qian, Yao Qian, Jian Wu, Michael Zeng, Furu Wei.

README_hd.md

@@ -201,13 +201,13 @@ checkpoint: जाँच बिंदु
 
 ### पिप का उपयोग करना
 
-इस रिपॉजिटरी का परीक्षण Python 3.8+, Flax 0.4.1+, PyTorch 1.10+ और TensorFlow 2.6+ के तहत किया गया है।
+इस रिपॉजिटरी का परीक्षण Python 3.8+, Flax 0.4.1+, PyTorch 1.11+ और TensorFlow 2.6+ के तहत किया गया है।
 
 आप [वर्चुअल एनवायरनमेंट](https://docs.python.org/3/library/venv.html) में 🤗 ट्रांसफॉर्मर इंस्टॉल कर सकते हैं। यदि आप अभी तक पायथन के वर्चुअल एनवायरनमेंट से परिचित नहीं हैं, तो कृपया इसे [उपयोगकर्ता निर्देश](https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/) पढ़ें।
 
 सबसे पहले, पायथन के उस संस्करण के साथ एक आभासी वातावरण बनाएं जिसका आप उपयोग करने और उसे सक्रिय करने की योजना बना रहे हैं।
 
-फिर, आपको Flax, PyTorch या TensorFlow में से किसी एक को स्थापित करने की आवश्यकता है। अपने प्लेटफ़ॉर्म पर इन फ़्रेमवर्क को स्थापित करने के लिए, [TensorFlow स्थापना पृष्ठ](https://www.tensorflow.org/install/), [PyTorch स्थापना पृष्ठ](https://pytorch.org/get-started/locally) 
+फिर, आपको Flax, PyTorch या TensorFlow में से किसी एक को स्थापित करने की आवश्यकता है। अपने प्लेटफ़ॉर्म पर इन फ़्रेमवर्क को स्थापित करने के लिए, [TensorFlow स्थापना पृष्ठ](https://www.tensorflow.org/install/), [PyTorch स्थापना पृष्ठ](https://pytorch.org/get-started/locally)
 
 देखें start-locally या [Flax स्थापना पृष्ठ](https://github.com/google/flax#quick-install).
 
@@ -221,14 +221,14 @@ pip install transformers
 
 ### कोंडा का उपयोग करना
 
-ट्रांसफॉर्मर संस्करण 4.0.0 के बाद से, हमारे पास एक कोंडा चैनल है: `हगिंगफेस`।
-
 ट्रांसफॉर्मर कोंडा के माध्यम से निम्नानुसार स्थापित किया जा सकता है:
 
 ```shell script
-conda install -c huggingface transformers
+conda install conda-forge::transformers
 ```
 
+> **_नोट:_** `huggingface` चैनल से `transformers` इंस्टॉल करना पुराना पड़ चुका है।
+
 कोंडा के माध्यम से Flax, PyTorch, या TensorFlow में से किसी एक को स्थापित करने के लिए, निर्देशों के लिए उनके संबंधित स्थापना पृष्ठ देखें।
 
 ## मॉडल आर्किटेक्चर
@@ -270,7 +270,7 @@ conda install -c huggingface transformers
 1. **[CLAP](https://huggingface.co/docs/transformers/model_doc/clap)** (LAION-AI से) Yusong Wu, Ke Chen, Tianyu Zhang, Yuchen Hui, Taylor Berg-Kirkpatrick, Shlomo Dubnov. द्वाराअनुसंधान पत्र [Large-scale Contrastive Language-Audio Pretraining with Feature Fusion and Keyword-to-Caption Augmentation](https://arxiv.org/abs/2211.06687) के साथ जारी किया गया
 1. **[CLIP](https://huggingface.co/docs/transformers/model_doc/clip)** (OpenAI से) साथ वाला पेपर [लर्निंग ट्रांसफरेबल विजुअल मॉडल फ्रॉम नेचुरल लैंग्वेज सुपरविजन](https://arxiv.org /abs/2103.00020) एलेक रैडफोर्ड, जोंग वूक किम, क्रिस हैलासी, आदित्य रमेश, गेब्रियल गोह, संध्या अग्रवाल, गिरीश शास्त्री, अमांडा एस्केल, पामेला मिश्किन, जैक क्लार्क, ग्रेचेन क्रुएगर, इल्या सुत्स्केवर द्वारा।
 1. **[CLIPSeg](https://huggingface.co/docs/transformers/model_doc/clipseg)** (from University of Göttingen) released with the paper [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003) by Timo Lüddecke and Alexander Ecker.
-1. **[CLVP](https://huggingface.co/docs/transformers/model_doc/clvp)** released with the paper [Better speech synthesis through scaling](https://arxiv.org/abs/2305.07243) by James Betker. 
+1. **[CLVP](https://huggingface.co/docs/transformers/model_doc/clvp)** released with the paper [Better speech synthesis through scaling](https://arxiv.org/abs/2305.07243) by James Betker.
 1. **[CodeGen](https://huggingface.co/docs/transformers/model_doc/codegen)** (सेल्सफोर्स से) साथ में पेपर [प्रोग्राम सिंथेसिस के लिए एक संवादात्मक प्रतिमान](https://arxiv.org/abs/2203.13474) एरिक निजकैंप, बो पैंग, हिरोआकी हयाशी, लिफू तू, हुआन वांग, यिंगबो झोउ, सिल्वियो सावरेस, कैमिंग जिओंग रिलीज।
 1. **[CodeLlama](https://huggingface.co/docs/transformers/model_doc/llama_code)** (MetaAI से) Baptiste Rozière, Jonas Gehring, Fabian Gloeckle, Sten Sootla, Itai Gat, Xiaoqing Ellen Tan, Yossi Adi, Jingyu Liu, Tal Remez, Jérémy Rapin, Artyom Kozhevnikov, Ivan Evtimov, Joanna Bitton, Manish Bhatt, Cristian Canton Ferrer, Aaron Grattafiori, Wenhan Xiong, Alexandre Défossez, Jade Copet, Faisal Azhar, Hugo Touvron, Louis Martin, Nicolas Usunier, Thomas Scialom, Gabriel Synnaeve. द्वाराअनुसंधान पत्र [Code Llama: Open Foundation Models for Code](https://ai.meta.com/research/publications/code-llama-open-foundation-models-for-code/) के साथ जारी किया गया
 1. **[Conditional DETR](https://huggingface.co/docs/transformers/model_doc/conditional_detr)** (माइक्रोसॉफ्ट रिसर्च एशिया से) कागज के साथ [फास्ट ट्रेनिंग कन्वर्जेंस के लिए सशर्त डीईटीआर](https://arxiv. org/abs/2108.06152) डेपू मेंग, ज़ियाओकांग चेन, ज़ेजिया फैन, गैंग ज़ेंग, होउकियांग ली, युहुई युआन, लेई सन, जिंगडोंग वांग द्वारा।
@@ -307,6 +307,7 @@ conda install -c huggingface transformers
 1. **[ErnieM](https://huggingface.co/docs/transformers/model_doc/ernie_m)** (Baidu से) Xuan Ouyang, Shuohuan Wang, Chao Pang, Yu Sun, Hao Tian, Hua Wu, Haifeng Wang. द्वाराअनुसंधान पत्र [ERNIE-M: Enhanced Multilingual Representation by Aligning Cross-lingual Semantics with Monolingual Corpora](https://arxiv.org/abs/2012.15674) के साथ जारी किया गया
 1. **[ESM](https://huggingface.co/docs/transformers/model_doc/esm)** (मेटा AI से) ट्रांसफॉर्मर प्रोटीन भाषा मॉडल हैं। **ESM-1b** पेपर के साथ जारी किया गया था [ अलेक्जेंडर राइव्स, जोशुआ मेयर, टॉम सर्कु, सिद्धार्थ गोयल, ज़ेमिंग लिन द्वारा जैविक संरचना और कार्य असुरक्षित सीखने को 250 मिलियन प्रोटीन अनुक्रमों तक स्केल करने से उभरता है] (https://www.pnas.org/content/118/15/e2016239118) जेसन लियू, डेमी गुओ, मायल ओट, सी. लॉरेंस ज़िटनिक, जेरी मा और रॉब फर्गस। **ESM-1v** को पेपर के साथ जारी किया गया था [भाषा मॉडल प्रोटीन फ़ंक्शन पर उत्परिवर्तन के प्रभावों की शून्य-शॉट भविष्यवाणी को सक्षम करते हैं] (https://doi.org/10.1101/2021.07.09.450648) जोशुआ मेयर, रोशन राव, रॉबर्ट वेरकुइल, जेसन लियू, टॉम सर्कु और अलेक्जेंडर राइव्स द्वारा। **ESM-2** को पेपर के साथ जारी किया गया था [भाषा मॉडल विकास के पैमाने पर प्रोटीन अनुक्रम सटीक संरचना भविष्यवाणी को सक्षम करते हैं](https://doi.org/10.1101/2022.07.20.500902) ज़ेमिंग लिन, हलील अकिन, रोशन राव, ब्रायन ही, झोंगकाई झू, वेंटिंग लू, ए द्वारा लान डॉस सैंटोस कोस्टा, मरियम फ़ज़ल-ज़रंडी, टॉम सर्कू, साल कैंडिडो, अलेक्जेंडर राइव्स।
 1. **[Falcon](https://huggingface.co/docs/transformers/model_doc/falcon)** (from Technology Innovation Institute) by Almazrouei, Ebtesam and Alobeidli, Hamza and Alshamsi, Abdulaziz and Cappelli, Alessandro and Cojocaru, Ruxandra and Debbah, Merouane and Goffinet, Etienne and Heslow, Daniel and Launay, Julien and Malartic, Quentin and Noune, Badreddine and Pannier, Baptiste and Penedo, Guilherme.
+1. **[FastSpeech2Conformer](model_doc/fastspeech2_conformer)** (ESPnet and Microsoft Research से) Pengcheng Guo, Florian Boyer, Xuankai Chang, Tomoki Hayashi, Yosuke Higuchi, Hirofumi Inaguma, Naoyuki Kamo, Chenda Li, Daniel Garcia-Romero, Jiatong Shi, Jing Shi, Shinji Watanabe, Kun Wei, Wangyou Zhang, and Yuekai Zhang. द्वाराअनुसंधान पत्र [Fastspeech 2: Fast And High-quality End-to-End Text To Speech](https://arxiv.org/pdf/2006.04558.pdf) के साथ जारी किया गया
 1. **[FLAN-T5](https://huggingface.co/docs/transformers/model_doc/flan-t5)** (from Google AI) released in the repository [google-research/t5x](https://github.com/google-research/t5x/blob/main/docs/models.md#flan-t5-checkpoints) by Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Eric Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Albert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdhery, Sharan Narang, Gaurav Mishra, Adams Yu, Vincent Zhao, Yanping Huang, Andrew Dai, Hongkun Yu, Slav Petrov, Ed H. Chi, Jeff Dean, Jacob Devlin, Adam Roberts, Denny Zhou, Quoc V. Le, and Jason Wei
 1. **[FLAN-UL2](https://huggingface.co/docs/transformers/model_doc/flan-ul2)** (from Google AI) released in the repository [google-research/t5x](https://github.com/google-research/t5x/blob/main/docs/models.md#flan-ul2-checkpoints) by Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Eric Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Albert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdhery, Sharan Narang, Gaurav Mishra, Adams Yu, Vincent Zhao, Yanping Huang, Andrew Dai, Hongkun Yu, Slav Petrov, Ed H. Chi, Jeff Dean, Jacob Devlin, Adam Roberts, Denny Zhou, Quoc V. Le, and Jason Wei
 1. **[FlauBERT](https://huggingface.co/docs/transformers/model_doc/flaubert)** (CNRS से) साथ वाला पेपर [FlauBERT: Unsupervised Language Model Pre-training for फ़्रेंच](https://arxiv .org/abs/1912.05372) Hang Le, Loïc Vial, Jibril Frej, Vincent Segonne, Maximin Coavoux, बेंजामिन लेकोउटेक्स, अलेक्जेंड्रे अल्लाउज़ेन, बेनोइट क्रैबे, लॉरेंट बेसेसियर, डिडिएर श्वाब द्वारा।
@@ -331,7 +332,7 @@ conda install -c huggingface transformers
 1. **[HerBERT](https://huggingface.co/docs/transformers/model_doc/herbert)** (Allegro.pl, AGH University of Science and Technology से) Piotr Rybak, Robert Mroczkowski, Janusz Tracz, Ireneusz Gawlik. द्वाराअनुसंधान पत्र [KLEJ: Comprehensive Benchmark for Polish Language Understanding](https://www.aclweb.org/anthology/2020.acl-main.111.pdf) के साथ जारी किया गया
 1. **[Hubert](https://huggingface.co/docs/transformers/model_doc/hubert)** (फेसबुक से) साथ में पेपर [ह्यूबर्ट: सेल्फ सुपरवाइज्ड स्पीच रिप्रेजेंटेशन लर्निंग बाय मास्क्ड प्रेडिक्शन ऑफ हिडन यूनिट्स](https ://arxiv.org/abs/2106.07447) वेई-निंग सू, बेंजामिन बोल्टे, याओ-हंग ह्यूबर्ट त्साई, कुशाल लखोटिया, रुस्लान सालाखुतदीनोव, अब्देलरहमान मोहम्मद द्वारा।
 1. **[I-BERT](https://huggingface.co/docs/transformers/model_doc/ibert)** (बर्कले से) साथ में कागज [I-BERT: Integer-only BERT Quantization](https:// arxiv.org/abs/2101.01321) सेहून किम, अमीर घोलमी, ज़ेवेई याओ, माइकल डब्ल्यू महोनी, कर्ट केटज़र द्वारा।
-1. **[IDEFICS](https://huggingface.co/docs/transformers/model_doc/idefics)** (from HuggingFace) released with the paper [OBELICS: An Open Web-Scale Filtered Dataset of Interleaved Image-Text Documents](https://huggingface.co/papers/2306.16527) by Hugo Laurençon, Lucile Saulnier, Léo Tronchon, Stas Bekman, Amanpreet Singh, Anton Lozhkov, Thomas Wang, Siddharth Karamcheti, Alexander M. Rush, Douwe Kiela, Matthieu Cord, Victor Sanh. 
+1. **[IDEFICS](https://huggingface.co/docs/transformers/model_doc/idefics)** (from HuggingFace) released with the paper [OBELICS: An Open Web-Scale Filtered Dataset of Interleaved Image-Text Documents](https://huggingface.co/papers/2306.16527) by Hugo Laurençon, Lucile Saulnier, Léo Tronchon, Stas Bekman, Amanpreet Singh, Anton Lozhkov, Thomas Wang, Siddharth Karamcheti, Alexander M. Rush, Douwe Kiela, Matthieu Cord, Victor Sanh.
 1. **[ImageGPT](https://huggingface.co/docs/transformers/model_doc/imagegpt)** (from OpenAI) released with the paper [Generative Pretraining from Pixels](https://openai.com/blog/image-gpt/) by Mark Chen, Alec Radford, Rewon Child, Jeffrey Wu, Heewoo Jun, David Luan, Ilya Sutskever.
 1. **[Informer](https://huggingface.co/docs/transformers/model_doc/informer)** (from Beihang University, UC Berkeley, Rutgers University, SEDD Company) released with the paper [Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecasting](https://arxiv.org/abs/2012.07436) by Haoyi Zhou, Shanghang Zhang, Jieqi Peng, Shuai Zhang, Jianxin Li, Hui Xiong, and Wancai Zhang.
 1. **[InstructBLIP](https://huggingface.co/docs/transformers/model_doc/instructblip)** (Salesforce से) Wenliang Dai, Junnan Li, Dongxu Li, Anthony Meng Huat Tiong, Junqi Zhao, Weisheng Wang, Boyang Li, Pascale Fung, Steven Hoi. द्वाराअनुसंधान पत्र [InstructBLIP: Towards General-purpose Vision-Language Models with Instruction Tuning](https://arxiv.org/abs/2305.06500) के साथ जारी किया गया
@@ -365,8 +366,8 @@ conda install -c huggingface transformers
 1. **[Megatron-BERT](https://huggingface.co/docs/transformers/model_doc/megatron-bert)** (NVIDIA से) कागज के साथ [Megatron-LM: मॉडल का उपयोग करके बहु-अरब पैरामीटर भाषा मॉडल का प्रशिक्षण Parallelism](https://arxiv.org/abs/1909.08053) मोहम्मद शोएबी, मोस्टोफा पटवारी, राउल पुरी, पैट्रिक लेग्रेस्ले, जेरेड कैस्पर और ब्रायन कैटानज़ारो द्वारा।
 1. **[Megatron-GPT2](https://huggingface.co/docs/transformers/model_doc/megatron_gpt2)** (NVIDIA से) साथ वाला पेपर [Megatron-LM: ट्रेनिंग मल्टी-बिलियन पैरामीटर लैंग्वेज मॉडल्स यूजिंग मॉडल पैरेललिज़्म] (https://arxiv.org/abs/1909.08053) मोहम्मद शोएबी, मोस्टोफा पटवारी, राउल पुरी, पैट्रिक लेग्रेस्ले, जेरेड कैस्पर और ब्रायन कैटानज़ारो द्वारा पोस्ट किया गया।
 1. **[MGP-STR](https://huggingface.co/docs/transformers/model_doc/mgp-str)** (Alibaba Research से) Peng Wang, Cheng Da, and Cong Yao. द्वाराअनुसंधान पत्र [Multi-Granularity Prediction for Scene Text Recognition](https://arxiv.org/abs/2209.03592) के साथ जारी किया गया
-1. **[Mistral](https://huggingface.co/docs/transformers/model_doc/mistral)** (from Mistral AI) by The Mistral AI team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed.. 
-1. **[Mixtral](https://huggingface.co/docs/transformers/model_doc/mixtral)** (from Mistral AI) by The [Mistral AI](https://mistral.ai) team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed. 
+1. **[Mistral](https://huggingface.co/docs/transformers/model_doc/mistral)** (from Mistral AI) by The Mistral AI team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed..
+1. **[Mixtral](https://huggingface.co/docs/transformers/model_doc/mixtral)** (from Mistral AI) by The [Mistral AI](https://mistral.ai) team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed.
 1. **[mLUKE](https://huggingface.co/docs/transformers/model_doc/mluke)** (फ्रॉम Studio Ousia) साथ में पेपर [mLUKE: द पावर ऑफ एंटिटी रिप्रेजेंटेशन इन मल्टीलिंगुअल प्रीट्रेन्ड लैंग्वेज मॉडल्स](https://arxiv.org/abs/2110.08151) रयोकन री, इकुया यामाडा, और योशिमासा त्सुरोका द्वारा।
 1. **[MMS](https://huggingface.co/docs/transformers/model_doc/mms)** (Facebook से) Vineel Pratap, Andros Tjandra, Bowen Shi, Paden Tomasello, Arun Babu, Sayani Kundu, Ali Elkahky, Zhaoheng Ni, Apoorv Vyas, Maryam Fazel-Zarandi, Alexei Baevski, Yossi Adi, Xiaohui Zhang, Wei-Ning Hsu, Alexis Conneau, Michael Auli. द्वाराअनुसंधान पत्र [Scaling Speech Technology to 1,000+ Languages](https://arxiv.org/abs/2305.13516) के साथ जारी किया गया
 1. **[MobileBERT](https://huggingface.co/docs/transformers/model_doc/mobilebert)** (सीएमयू/गूगल ब्रेन से) साथ में कागज [मोबाइलबर्ट: संसाधन-सीमित उपकरणों के लिए एक कॉम्पैक्ट टास्क-अज्ञेय बीईआरटी] (https://arxiv.org/abs/2004.02984) Zhiqing Sun, Hongkun Yu, Xiaodan Song, Renjie Liu, Yiming Yang, और Denny Zhou द्वारा पोस्ट किया गया।
@@ -402,10 +403,11 @@ conda install -c huggingface transformers
 1. **[Pix2Struct](https://huggingface.co/docs/transformers/model_doc/pix2struct)** (Google से) Kenton Lee, Mandar Joshi, Iulia Turc, Hexiang Hu, Fangyu Liu, Julian Eisenschlos, Urvashi Khandelwal, Peter Shaw, Ming-Wei Chang, Kristina Toutanova. द्वाराअनुसंधान पत्र [Pix2Struct: Screenshot Parsing as Pretraining for Visual Language Understanding](https://arxiv.org/abs/2210.03347) के साथ जारी किया गया
 1. **[PLBart](https://huggingface.co/docs/transformers/model_doc/plbart)** (UCLA NLP से) साथ वाला पेपर [प्रोग्राम अंडरस्टैंडिंग एंड जेनरेशन के लिए यूनिफाइड प्री-ट्रेनिंग](https://arxiv .org/abs/2103.06333) वसी उद्दीन अहमद, सैकत चक्रवर्ती, बैशाखी रे, काई-वेई चांग द्वारा।
 1. **[PoolFormer](https://huggingface.co/docs/transformers/model_doc/poolformer)** (from Sea AI Labs) released with the paper [MetaFormer is Actually What You Need for Vision](https://arxiv.org/abs/2111.11418) by Yu, Weihao and Luo, Mi and Zhou, Pan and Si, Chenyang and Zhou, Yichen and Wang, Xinchao and Feng, Jiashi and Yan, Shuicheng.
-1. **[Pop2Piano](https://huggingface.co/docs/transformers/model_doc/pop2piano)** released with the paper [Pop2Piano : Pop Audio-based Piano Cover Generation](https://arxiv.org/abs/2211.00895) by Jongho Choi, Kyogu Lee. 
+1. **[Pop2Piano](https://huggingface.co/docs/transformers/model_doc/pop2piano)** released with the paper [Pop2Piano : Pop Audio-based Piano Cover Generation](https://arxiv.org/abs/2211.00895) by Jongho Choi, Kyogu Lee.
 1. **[ProphetNet](https://huggingface.co/docs/transformers/model_doc/prophetnet)** (माइक्रोसॉफ्ट रिसर्च से) साथ में पेपर [ProphetNet: प्रेडिक्टिंग फ्यूचर एन-ग्राम फॉर सीक्वेंस-टू-सीक्वेंस प्री-ट्रेनिंग ](https://arxiv.org/abs/2001.04063) यू यान, वीज़ेन क्यूई, येयुन गोंग, दयाहेंग लियू, नान डुआन, जिउशेंग चेन, रुओफ़ेई झांग और मिंग झोउ द्वारा पोस्ट किया गया।
 1. **[PVT](https://huggingface.co/docs/transformers/model_doc/pvt)** (Nanjing University, The University of Hong Kong etc. से) Wenhai Wang, Enze Xie, Xiang Li, Deng-Ping Fan, Kaitao Song, Ding Liang, Tong Lu, Ping Luo, Ling Shao. द्वाराअनुसंधान पत्र [Pyramid Vision Transformer: A Versatile Backbone for Dense Prediction without Convolutions](https://arxiv.org/pdf/2102.12122.pdf) के साथ जारी किया गया
 1. **[QDQBert](https://huggingface.co/docs/transformers/model_doc/qdqbert)** (NVIDIA से) साथ वाला पेपर [डीप लर्निंग इंफ़ेक्शन के लिए इंटीजर क्वांटिज़ेशन: प्रिंसिपल्स एंड एम्पिरिकल इवैल्यूएशन](https:// arxiv.org/abs/2004.09602) हाओ वू, पैट्रिक जुड, जिआओजी झांग, मिखाइल इसेव और पॉलियस माइकेविसियस द्वारा।
+1. **[Qwen2](https://huggingface.co/docs/transformers/model_doc/qwen2)** (the Qwen team, Alibaba Group से) Jinze Bai, Shuai Bai, Yunfei Chu, Zeyu Cui, Kai Dang, Xiaodong Deng, Yang Fan, Wenbin Ge, Yu Han, Fei Huang, Binyuan Hui, Luo Ji, Mei Li, Junyang Lin, Runji Lin, Dayiheng Liu, Gao Liu, Chengqiang Lu, Keming Lu, Jianxin Ma, Rui Men, Xingzhang Ren, Xuancheng Ren, Chuanqi Tan, Sinan Tan, Jianhong Tu, Peng Wang, Shijie Wang, Wei Wang, Shengguang Wu, Benfeng Xu, Jin Xu, An Yang, Hao Yang, Jian Yang, Shusheng Yang, Yang Yao, Bowen Yu, Hongyi Yuan, Zheng Yuan, Jianwei Zhang, Xingxuan Zhang, Yichang Zhang, Zhenru Zhang, Chang Zhou, Jingren Zhou, Xiaohuan Zhou and Tianhang Zhu. द्वाराअनुसंधान पत्र [Qwen Technical Report](https://arxiv.org/abs/2309.16609) के साथ जारी किया गया
 1. **[RAG](https://huggingface.co/docs/transformers/model_doc/rag)** (फेसबुक से) साथ में कागज [रिट्रीवल-ऑगमेंटेड जेनरेशन फॉर नॉलेज-इंटेंसिव एनएलपी टास्क](https://arxiv .org/abs/2005.11401) पैट्रिक लुईस, एथन पेरेज़, अलेक्जेंड्रा पिक्टस, फैबियो पेट्रोनी, व्लादिमीर कारपुखिन, नमन गोयल, हेनरिक कुटलर, माइक लुईस, वेन-ताउ यिह, टिम रॉकटाशेल, सेबस्टियन रिडेल, डौवे कीला द्वारा।
 1. **[REALM](https://huggingface.co/docs/transformers/model_doc/realm.html)** (Google अनुसंधान से) केल्विन गु, केंटन ली, ज़ोरा तुंग, पानुपोंग पसुपत और मिंग-वेई चांग द्वारा साथ में दिया गया पेपर [REALM: रिट्रीवल-ऑगमेंटेड लैंग्वेज मॉडल प्री-ट्रेनिंग](https://arxiv.org/abs/2002.08909)।
 1. **[Reformer](https://huggingface.co/docs/transformers/model_doc/reformer)** (from Google Research) released with the paper [Reformer: The Efficient Transformer](https://arxiv.org/abs/2001.04451) by Nikita Kitaev, Łukasz Kaiser, Anselm Levskaya.
@@ -423,6 +425,7 @@ conda install -c huggingface transformers
 1. **[Segment Anything](https://huggingface.co/docs/transformers/model_doc/sam)** (Meta AI से) Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alex Berg, Wan-Yen Lo, Piotr Dollar, Ross Girshick. द्वाराअनुसंधान पत्र [Segment Anything](https://arxiv.org/pdf/2304.02643v1.pdf) के साथ जारी किया गया
 1. **[SEW](https://huggingface.co/docs/transformers/model_doc/sew)** (ASAPP से) साथ देने वाला पेपर [भाषण पहचान के लिए अनसुपरवाइज्ड प्री-ट्रेनिंग में परफॉर्मेंस-एफिशिएंसी ट्रेड-ऑफ्स](https ://arxiv.org/abs/2109.06870) फेलिक्स वू, क्वांगयुन किम, जिंग पैन, क्यू हान, किलियन क्यू. वेनबर्गर, योव आर्टज़ी द्वारा।
 1. **[SEW-D](https://huggingface.co/docs/transformers/model_doc/sew_d)** (ASAPP से) साथ में पेपर [भाषण पहचान के लिए अनसुपरवाइज्ड प्री-ट्रेनिंग में परफॉर्मेंस-एफिशिएंसी ट्रेड-ऑफ्स] (https://arxiv.org/abs/2109.06870) फेलिक्स वू, क्वांगयुन किम, जिंग पैन, क्यू हान, किलियन क्यू. वेनबर्गर, योआव आर्टज़ी द्वारा पोस्ट किया गया।
+1. **[SigLIP](https://huggingface.co/docs/transformers/model_doc/siglip)** (Google AI से) Xiaohua Zhai, Basil Mustafa, Alexander Kolesnikov, Lucas Beyer. द्वाराअनुसंधान पत्र [Sigmoid Loss for Language Image Pre-Training](https://arxiv.org/abs/2303.15343) के साथ जारी किया गया
 1. **[SpeechT5](https://huggingface.co/docs/transformers/model_doc/speecht5)** (from Microsoft Research) released with the paper [SpeechT5: Unified-Modal Encoder-Decoder Pre-Training for Spoken Language Processing](https://arxiv.org/abs/2110.07205) by Junyi Ao, Rui Wang, Long Zhou, Chengyi Wang, Shuo Ren, Yu Wu, Shujie Liu, Tom Ko, Qing Li, Yu Zhang, Zhihua Wei, Yao Qian, Jinyu Li, Furu Wei.
 1. **[SpeechToTextTransformer](https://huggingface.co/docs/transformers/model_doc/speech_to_text)** (फेसबुक से), साथ में पेपर [फेयरसेक S2T: फास्ट स्पीच-टू-टेक्स्ट मॉडलिंग विद फेयरसेक](https: //arxiv.org/abs/2010.05171) चांगहान वांग, यूं तांग, जुताई मा, ऐनी वू, दिमित्रो ओखोनको, जुआन पिनो द्वारा पोस्ट किया गया。
 1. **[SpeechToTextTransformer2](https://huggingface.co/docs/transformers/model_doc/speech_to_text_2)** (फेसबुक से) साथ में पेपर [लार्ज-स्केल सेल्फ- एंड सेमी-सुपरवाइज्ड लर्निंग फॉर स्पीच ट्रांसलेशन](https://arxiv.org/abs/2104.06678) चांगहान वांग, ऐनी वू, जुआन पिनो, एलेक्सी बेवस्की, माइकल औली, एलेक्सिस द्वारा Conneau द्वारा पोस्ट किया गया।
@@ -449,7 +452,7 @@ conda install -c huggingface transformers
 1. **[UMT5](https://huggingface.co/docs/transformers/model_doc/umt5)** (Google Research से) Hyung Won Chung, Xavier Garcia, Adam Roberts, Yi Tay, Orhan Firat, Sharan Narang, Noah Constant. द्वाराअनुसंधान पत्र [UniMax: Fairer and More Effective Language Sampling for Large-Scale Multilingual Pretraining](https://openreview.net/forum?id=kXwdL1cWOAi) के साथ जारी किया गया
 1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (माइक्रोसॉफ्ट रिसर्च से) साथ में दिया गया पेपर [UniSpeech: यूनिफाइड स्पीच रिप्रेजेंटेशन लर्निंग विद लेबलेड एंड अनलेबल्ड डेटा](https:/ /arxiv.org/abs/2101.07597) चेंगई वांग, यू वू, याओ कियान, केनिची कुमातानी, शुजी लियू, फुरु वेई, माइकल ज़ेंग, ज़ुएदोंग हुआंग द्वारा।
 1. **[UniSpeechSat](https://huggingface.co/docs/transformers/model_doc/unispeech-sat)** (माइक्रोसॉफ्ट रिसर्च से) कागज के साथ [UNISPEECH-SAT: यूनिवर्सल स्पीच रिप्रेजेंटेशन लर्निंग विद स्पीकर अवेयर प्री-ट्रेनिंग ](https://arxiv.org/abs/2110.05752) सानयुआन चेन, यू वू, चेंग्यी वांग, झेंगयांग चेन, झूओ चेन, शुजी लियू, जियान वू, याओ कियान, फुरु वेई, जिन्यु ली, जियांगज़ान यू द्वारा पोस्ट किया गया।
-1. **[UnivNet](https://huggingface.co/docs/transformers/model_doc/univnet)** (from Kakao Corporation) released with the paper [UnivNet: A Neural Vocoder with Multi-Resolution Spectrogram Discriminators for High-Fidelity Waveform Generation](https://arxiv.org/abs/2106.07889) by Won Jang, Dan Lim, Jaesam Yoon, Bongwan Kim, and Juntae Kim. 
+1. **[UnivNet](https://huggingface.co/docs/transformers/model_doc/univnet)** (from Kakao Corporation) released with the paper [UnivNet: A Neural Vocoder with Multi-Resolution Spectrogram Discriminators for High-Fidelity Waveform Generation](https://arxiv.org/abs/2106.07889) by Won Jang, Dan Lim, Jaesam Yoon, Bongwan Kim, and Juntae Kim.
 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)** (सिंघुआ यूनिवर्सिटी और ननकाई यूनिवर्सिटी से) साथ में पेपर [विजुअल अटेंशन नेटवर्क](https://arxiv.org/ pdf/2202.09741.pdf) मेंग-हाओ गुओ, चेंग-ज़े लू, झेंग-निंग लियू, मिंग-मिंग चेंग, शि-मिन हू द्वारा।
 1. **[VideoMAE](https://huggingface.co/docs/transformers/model_doc/videomae)** (मल्टीमीडिया कम्प्यूटिंग ग्रुप, नानजिंग यूनिवर्सिटी से) साथ में पेपर [वीडियोएमएई: मास्क्ड ऑटोएन्कोडर स्व-पर्यवेक्षित वीडियो प्री-ट्रेनिंग के लिए डेटा-कुशल सीखने वाले हैं] (https://arxiv.org/abs/2203.12602) ज़ान टोंग, यिबिंग सॉन्ग, जुए द्वारा वांग, लिमिन वांग द्वारा पोस्ट किया गया।
@@ -465,6 +468,7 @@ conda install -c huggingface transformers
 1. **[VITS](https://huggingface.co/docs/transformers/model_doc/vits)** (Kakao Enterprise से) Jaehyeon Kim, Jungil Kong, Juhee Son. द्वाराअनुसंधान पत्र [Conditional Variational Autoencoder with Adversarial Learning for End-to-End Text-to-Speech](https://arxiv.org/abs/2106.06103) के साथ जारी किया गया
 1. **[ViViT](https://huggingface.co/docs/transformers/model_doc/vivit)** (from Google Research) released with the paper [ViViT: A Video Vision Transformer](https://arxiv.org/abs/2103.15691) by Anurag Arnab, Mostafa Dehghani, Georg Heigold, Chen Sun, Mario Lučić, Cordelia Schmid.
 1. **[Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/wav2vec2)** (फेसबुक एआई से) साथ में पेपर [wav2vec 2.0: ए फ्रेमवर्क फॉर सेल्फ-सुपरवाइज्ड लर्निंग ऑफ स्पीच रिप्रेजेंटेशन](https://arxiv.org/abs/2006.11477) एलेक्सी बेवस्की, हेनरी झोउ, अब्देलरहमान मोहम्मद, माइकल औली द्वारा।
+1. **[Wav2Vec2-BERT](https://huggingface.co/docs/transformers/model_doc/wav2vec2-bert)** (from Meta AI) released with the paper [Seamless: Multilingual Expressive and Streaming Speech Translation](https://ai.meta.com/research/publications/seamless-multilingual-expressive-and-streaming-speech-translation/) by the Seamless Communication team.
 1. **[Wav2Vec2-Conformer](https://huggingface.co/docs/transformers/model_doc/wav2vec2-conformer)** (Facebook AI से) साथ वाला पेपर [FAIRSEQ S2T: FAIRSEQ के साथ फास्ट स्पीच-टू-टेक्स्ट मॉडलिंग ](https://arxiv.org/abs/2010.05171) चांगहान वांग, यूं तांग, जुताई मा, ऐनी वू, सरव्या पोपुरी, दिमित्रो ओखोनको, जुआन पिनो द्वारा पोस्ट किया गया।
 1. **[Wav2Vec2Phoneme](https://huggingface.co/docs/transformers/model_doc/wav2vec2_phoneme)** (Facebook AI से) साथ वाला पेपर [सरल और प्रभावी जीरो-शॉट क्रॉस-लिंगुअल फोनेम रिकॉग्निशन](https://arxiv.org/abs/2109.11680) कियानटोंग जू, एलेक्सी बाएव्स्की, माइकल औली द्वारा।
 1. **[WavLM](https://huggingface.co/docs/transformers/model_doc/wavlm)** (माइक्रोसॉफ्ट रिसर्च से) पेपर के साथ जारी किया गया [WavLM: फुल स्टैक के लिए बड़े पैमाने पर स्व-पर्यवेक्षित पूर्व-प्रशिक्षण स्पीच प्रोसेसिंग](https://arxiv.org/abs/2110.13900) सानयुआन चेन, चेंगयी वांग, झेंगयांग चेन, यू वू, शुजी लियू, ज़ुओ चेन, जिन्यु ली, नाओयुकी कांडा, ताकुया योशियोका, ज़िओंग जिओ, जियान वू, लॉन्ग झोउ, शुओ रेन, यानमिन कियान, याओ कियान, जियान वू, माइकल ज़ेंग, फुरु वेई।

README_ja.md

@@ -259,7 +259,7 @@ Hugging Faceチームによって作られた **[トランスフォーマーを
 
 ### pipにて
 
-このリポジトリは、Python 3.8+, Flax 0.4.1+, PyTorch 1.10+, TensorFlow 2.6+ でテストされています。
+このリポジトリは、Python 3.8+, Flax 0.4.1+, PyTorch 1.11+, TensorFlow 2.6+ でテストされています。
 
 🤗Transformersは[仮想環境](https://docs.python.org/3/library/venv.html)にインストールする必要があります。Pythonの仮想環境に慣れていない場合は、[ユーザーガイド](https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/)を確認してください。
 
@@ -278,14 +278,14 @@ pip install transformers
 
 ### condaにて
 
-Transformersバージョン4.0.0から、condaチャンネルを搭載しました: `huggingface`。
-
 🤗Transformersは以下のようにcondaを使って設置することができます:
 
 ```shell script
-conda install -c huggingface transformers
+conda install conda-forge::transformers
 ```
 
+> **_注意:_**  `huggingface` チャンネルから `transformers` をインストールすることは非推奨です。
+
 Flax、PyTorch、TensorFlowをcondaでインストールする方法は、それぞれのインストールページに従ってください。
 
 > **_注意:_**  Windowsでは、キャッシュの恩恵を受けるために、デベロッパーモードを有効にするよう促されることがあります。このような場合は、[このissue](https://github.com/huggingface/huggingface_hub/issues/1062)でお知らせください。
@@ -330,7 +330,7 @@ Flax、PyTorch、TensorFlowをcondaでインストールする方法は、それ
 1. **[CLAP](https://huggingface.co/docs/transformers/model_doc/clap)** (LAION-AI から) Yusong Wu, Ke Chen, Tianyu Zhang, Yuchen Hui, Taylor Berg-Kirkpatrick, Shlomo Dubnov. から公開された研究論文 [Large-scale Contrastive Language-Audio Pretraining with Feature Fusion and Keyword-to-Caption Augmentation](https://arxiv.org/abs/2211.06687)
 1. **[CLIP](https://huggingface.co/docs/transformers/model_doc/clip)** (OpenAI から) Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, Ilya Sutskever から公開された研究論文: [Learning Transferable Visual Models From Natural Language Supervision](https://arxiv.org/abs/2103.00020)
 1. **[CLIPSeg](https://huggingface.co/docs/transformers/model_doc/clipseg)** (University of Göttingen から) Timo Lüddecke and Alexander Ecker から公開された研究論文: [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003)
-1. **[CLVP](https://huggingface.co/docs/transformers/model_doc/clvp)** released with the paper [Better speech synthesis through scaling](https://arxiv.org/abs/2305.07243) by James Betker. 
+1. **[CLVP](https://huggingface.co/docs/transformers/model_doc/clvp)** released with the paper [Better speech synthesis through scaling](https://arxiv.org/abs/2305.07243) by James Betker.
 1. **[CodeGen](https://huggingface.co/docs/transformers/model_doc/codegen)** (Salesforce から) Erik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu, Huan Wang, Yingbo Zhou, Silvio Savarese, Caiming Xiong から公開された研究論文: [A Conversational Paradigm for Program Synthesis](https://arxiv.org/abs/2203.13474)
 1. **[CodeLlama](https://huggingface.co/docs/transformers/model_doc/llama_code)** (MetaAI から) Baptiste Rozière, Jonas Gehring, Fabian Gloeckle, Sten Sootla, Itai Gat, Xiaoqing Ellen Tan, Yossi Adi, Jingyu Liu, Tal Remez, Jérémy Rapin, Artyom Kozhevnikov, Ivan Evtimov, Joanna Bitton, Manish Bhatt, Cristian Canton Ferrer, Aaron Grattafiori, Wenhan Xiong, Alexandre Défossez, Jade Copet, Faisal Azhar, Hugo Touvron, Louis Martin, Nicolas Usunier, Thomas Scialom, Gabriel Synnaeve. から公開された研究論文 [Code Llama: Open Foundation Models for Code](https://ai.meta.com/research/publications/code-llama-open-foundation-models-for-code/)
 1. **[Conditional DETR](https://huggingface.co/docs/transformers/model_doc/conditional_detr)** (Microsoft Research Asia から) Depu Meng, Xiaokang Chen, Zejia Fan, Gang Zeng, Houqiang Li, Yuhui Yuan, Lei Sun, Jingdong Wang から公開された研究論文: [Conditional DETR for Fast Training Convergence](https://arxiv.org/abs/2108.06152)
@@ -367,6 +367,7 @@ Flax、PyTorch、TensorFlowをcondaでインストールする方法は、それ
 1. **[ErnieM](https://huggingface.co/docs/transformers/model_doc/ernie_m)** (Baidu から) Xuan Ouyang, Shuohuan Wang, Chao Pang, Yu Sun, Hao Tian, Hua Wu, Haifeng Wang. から公開された研究論文 [ERNIE-M: Enhanced Multilingual Representation by Aligning Cross-lingual Semantics with Monolingual Corpora](https://arxiv.org/abs/2012.15674)
 1. **[ESM](https://huggingface.co/docs/transformers/model_doc/esm)** (Meta AI から) はトランスフォーマープロテイン言語モデルです.  **ESM-1b** は Alexander Rives, Joshua Meier, Tom Sercu, Siddharth Goyal, Zeming Lin, Jason Liu, Demi Guo, Myle Ott, C. Lawrence Zitnick, Jerry Ma, and Rob Fergus から公開された研究論文: [Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences](https://www.pnas.org/content/118/15/e2016239118). **ESM-1v** は Joshua Meier, Roshan Rao, Robert Verkuil, Jason Liu, Tom Sercu and Alexander Rives　から公開された研究論文: [Language models enable zero-shot prediction of the effects of mutations on protein function](https://doi.org/10.1101/2021.07.09.450648). **ESM-2** と　**ESMFold** は Zeming Lin, Halil Akin, Roshan Rao, Brian Hie, Zhongkai Zhu, Wenting Lu, Allan dos Santos Costa, Maryam Fazel-Zarandi, Tom Sercu, Sal Candido, Alexander Rives から公開された研究論文: [Language models of protein sequences at the scale of evolution enable accurate structure prediction](https://doi.org/10.1101/2022.07.20.500902)
 1. **[Falcon](https://huggingface.co/docs/transformers/model_doc/falcon)** (from Technology Innovation Institute) by Almazrouei, Ebtesam and Alobeidli, Hamza and Alshamsi, Abdulaziz and Cappelli, Alessandro and Cojocaru, Ruxandra and Debbah, Merouane and Goffinet, Etienne and Heslow, Daniel and Launay, Julien and Malartic, Quentin and Noune, Badreddine and Pannier, Baptiste and Penedo, Guilherme.
+1. **[FastSpeech2Conformer](model_doc/fastspeech2_conformer)** (ESPnet and Microsoft Research から) Pengcheng Guo, Florian Boyer, Xuankai Chang, Tomoki Hayashi, Yosuke Higuchi, Hirofumi Inaguma, Naoyuki Kamo, Chenda Li, Daniel Garcia-Romero, Jiatong Shi, Jing Shi, Shinji Watanabe, Kun Wei, Wangyou Zhang, and Yuekai Zhang. から公開された研究論文 [Fastspeech 2: Fast And High-quality End-to-End Text To Speech](https://arxiv.org/pdf/2006.04558.pdf)
 1. **[FLAN-T5](https://huggingface.co/docs/transformers/model_doc/flan-t5)** (Google AI から) Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Eric Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Albert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdhery, Sharan Narang, Gaurav Mishra, Adams Yu, Vincent Zhao, Yanping Huang, Andrew Dai, Hongkun Yu, Slav Petrov, Ed H. Chi, Jeff Dean, Jacob Devlin, Adam Roberts, Denny Zhou, Quoc V から公開されたレポジトリー [google-research/t5x](https://github.com/google-research/t5x/blob/main/docs/models.md#flan-t5-checkpoints) Le, and Jason Wei
 1. **[FLAN-UL2](https://huggingface.co/docs/transformers/model_doc/flan-ul2)** (from Google AI) released in the repository [google-research/t5x](https://github.com/google-research/t5x/blob/main/docs/models.md#flan-ul2-checkpoints) by Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Eric Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Albert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdhery, Sharan Narang, Gaurav Mishra, Adams Yu, Vincent Zhao, Yanping Huang, Andrew Dai, Hongkun Yu, Slav Petrov, Ed H. Chi, Jeff Dean, Jacob Devlin, Adam Roberts, Denny Zhou, Quoc V. Le, and Jason Wei
 1. **[FlauBERT](https://huggingface.co/docs/transformers/model_doc/flaubert)** (CNRS から) Hang Le, Loïc Vial, Jibril Frej, Vincent Segonne, Maximin Coavoux, Benjamin Lecouteux, Alexandre Allauzen, Benoît Crabbé, Laurent Besacier, Didier Schwab から公開された研究論文: [FlauBERT: Unsupervised Language Model Pre-training for French](https://arxiv.org/abs/1912.05372)
@@ -391,7 +392,7 @@ Flax、PyTorch、TensorFlowをcondaでインストールする方法は、それ
 1. **[HerBERT](https://huggingface.co/docs/transformers/model_doc/herbert)** (Allegro.pl, AGH University of Science and Technology から) Piotr Rybak, Robert Mroczkowski, Janusz Tracz, Ireneusz Gawlik. から公開された研究論文 [KLEJ: Comprehensive Benchmark for Polish Language Understanding](https://www.aclweb.org/anthology/2020.acl-main.111.pdf)
 1. **[Hubert](https://huggingface.co/docs/transformers/model_doc/hubert)** (Facebook から) Wei-Ning Hsu, Benjamin Bolte, Yao-Hung Hubert Tsai, Kushal Lakhotia, Ruslan Salakhutdinov, Abdelrahman Mohamed から公開された研究論文: [HuBERT: Self-Supervised Speech Representation Learning by Masked Prediction of Hidden Units](https://arxiv.org/abs/2106.07447)
 1. **[I-BERT](https://huggingface.co/docs/transformers/model_doc/ibert)** (Berkeley から) Sehoon Kim, Amir Gholami, Zhewei Yao, Michael W. Mahoney, Kurt Keutzer から公開された研究論文: [I-BERT: Integer-only BERT Quantization](https://arxiv.org/abs/2101.01321)
-1. **[IDEFICS](https://huggingface.co/docs/transformers/model_doc/idefics)** (from HuggingFace) released with the paper [OBELICS: An Open Web-Scale Filtered Dataset of Interleaved Image-Text Documents](https://huggingface.co/papers/2306.16527) by Hugo Laurençon, Lucile Saulnier, Léo Tronchon, Stas Bekman, Amanpreet Singh, Anton Lozhkov, Thomas Wang, Siddharth Karamcheti, Alexander M. Rush, Douwe Kiela, Matthieu Cord, Victor Sanh. 
+1. **[IDEFICS](https://huggingface.co/docs/transformers/model_doc/idefics)** (from HuggingFace) released with the paper [OBELICS: An Open Web-Scale Filtered Dataset of Interleaved Image-Text Documents](https://huggingface.co/papers/2306.16527) by Hugo Laurençon, Lucile Saulnier, Léo Tronchon, Stas Bekman, Amanpreet Singh, Anton Lozhkov, Thomas Wang, Siddharth Karamcheti, Alexander M. Rush, Douwe Kiela, Matthieu Cord, Victor Sanh.
 1. **[ImageGPT](https://huggingface.co/docs/transformers/model_doc/imagegpt)** (OpenAI から) Mark Chen, Alec Radford, Rewon Child, Jeffrey Wu, Heewoo Jun, David Luan, Ilya Sutskever から公開された研究論文: [Generative Pretraining from Pixels](https://openai.com/blog/image-gpt/)
 1. **[Informer](https://huggingface.co/docs/transformers/model_doc/informer)** (from Beihang University, UC Berkeley, Rutgers University, SEDD Company) released with the paper [Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecasting](https://arxiv.org/abs/2012.07436) by Haoyi Zhou, Shanghang Zhang, Jieqi Peng, Shuai Zhang, Jianxin Li, Hui Xiong, and Wancai Zhang.
 1. **[InstructBLIP](https://huggingface.co/docs/transformers/model_doc/instructblip)** (Salesforce から) Wenliang Dai, Junnan Li, Dongxu Li, Anthony Meng Huat Tiong, Junqi Zhao, Weisheng Wang, Boyang Li, Pascale Fung, Steven Hoi. から公開された研究論文 [InstructBLIP: Towards General-purpose Vision-Language Models with Instruction Tuning](https://arxiv.org/abs/2305.06500)
@@ -425,8 +426,8 @@ Flax、PyTorch、TensorFlowをcondaでインストールする方法は、それ
 1. **[Megatron-BERT](https://huggingface.co/docs/transformers/model_doc/megatron-bert)** (NVIDIA から) Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro から公開された研究論文: [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053)
 1. **[Megatron-GPT2](https://huggingface.co/docs/transformers/model_doc/megatron_gpt2)** (NVIDIA から) Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro から公開された研究論文: [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053)
 1. **[MGP-STR](https://huggingface.co/docs/transformers/model_doc/mgp-str)** (Alibaba Research から) Peng Wang, Cheng Da, and Cong Yao. から公開された研究論文 [Multi-Granularity Prediction for Scene Text Recognition](https://arxiv.org/abs/2209.03592)
-1. **[Mistral](https://huggingface.co/docs/transformers/model_doc/mistral)** (from Mistral AI) by The Mistral AI team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed.. 
-1. **[Mixtral](https://huggingface.co/docs/transformers/model_doc/mixtral)** (from Mistral AI) by The [Mistral AI](https://mistral.ai) team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed. 
+1. **[Mistral](https://huggingface.co/docs/transformers/model_doc/mistral)** (from Mistral AI) by The Mistral AI team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed..
+1. **[Mixtral](https://huggingface.co/docs/transformers/model_doc/mixtral)** (from Mistral AI) by The [Mistral AI](https://mistral.ai) team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed.
 1. **[mLUKE](https://huggingface.co/docs/transformers/model_doc/mluke)** (Studio Ousia から) Ryokan Ri, Ikuya Yamada, and Yoshimasa Tsuruoka から公開された研究論文: [mLUKE: The Power of Entity Representations in Multilingual Pretrained Language Models](https://arxiv.org/abs/2110.08151)
 1. **[MMS](https://huggingface.co/docs/transformers/model_doc/mms)** (Facebook から) Vineel Pratap, Andros Tjandra, Bowen Shi, Paden Tomasello, Arun Babu, Sayani Kundu, Ali Elkahky, Zhaoheng Ni, Apoorv Vyas, Maryam Fazel-Zarandi, Alexei Baevski, Yossi Adi, Xiaohui Zhang, Wei-Ning Hsu, Alexis Conneau, Michael Auli. から公開された研究論文 [Scaling Speech Technology to 1,000+ Languages](https://arxiv.org/abs/2305.13516)
 1. **[MobileBERT](https://huggingface.co/docs/transformers/model_doc/mobilebert)** (CMU/Google Brain から) Zhiqing Sun, Hongkun Yu, Xiaodan Song, Renjie Liu, Yiming Yang, and Denny Zhou から公開された研究論文: [MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices](https://arxiv.org/abs/2004.02984)
@@ -462,10 +463,11 @@ Flax、PyTorch、TensorFlowをcondaでインストールする方法は、それ
 1. **[Pix2Struct](https://huggingface.co/docs/transformers/model_doc/pix2struct)** (Google から) Kenton Lee, Mandar Joshi, Iulia Turc, Hexiang Hu, Fangyu Liu, Julian Eisenschlos, Urvashi Khandelwal, Peter Shaw, Ming-Wei Chang, Kristina Toutanova. から公開された研究論文 [Pix2Struct: Screenshot Parsing as Pretraining for Visual Language Understanding](https://arxiv.org/abs/2210.03347)
 1. **[PLBart](https://huggingface.co/docs/transformers/model_doc/plbart)** (UCLA NLP から) Wasi Uddin Ahmad, Saikat Chakraborty, Baishakhi Ray, Kai-Wei Chang から公開された研究論文: [Unified Pre-training for Program Understanding and Generation](https://arxiv.org/abs/2103.06333)
 1. **[PoolFormer](https://huggingface.co/docs/transformers/model_doc/poolformer)** (Sea AI Labs から) Yu, Weihao and Luo, Mi and Zhou, Pan and Si, Chenyang and Zhou, Yichen and Wang, Xinchao and Feng, Jiashi and Yan, Shuicheng から公開された研究論文: [MetaFormer is Actually What You Need for Vision](https://arxiv.org/abs/2111.11418)
-1. **[Pop2Piano](https://huggingface.co/docs/transformers/model_doc/pop2piano)** released with the paper [Pop2Piano : Pop Audio-based Piano Cover Generation](https://arxiv.org/abs/2211.00895) by Jongho Choi, Kyogu Lee. 
+1. **[Pop2Piano](https://huggingface.co/docs/transformers/model_doc/pop2piano)** released with the paper [Pop2Piano : Pop Audio-based Piano Cover Generation](https://arxiv.org/abs/2211.00895) by Jongho Choi, Kyogu Lee.
 1. **[ProphetNet](https://huggingface.co/docs/transformers/model_doc/prophetnet)** (Microsoft Research から) Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou から公開された研究論文: [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063)
 1. **[PVT](https://huggingface.co/docs/transformers/model_doc/pvt)** (Nanjing University, The University of Hong Kong etc. から) Wenhai Wang, Enze Xie, Xiang Li, Deng-Ping Fan, Kaitao Song, Ding Liang, Tong Lu, Ping Luo, Ling Shao. から公開された研究論文 [Pyramid Vision Transformer: A Versatile Backbone for Dense Prediction without Convolutions](https://arxiv.org/pdf/2102.12122.pdf)
 1. **[QDQBert](https://huggingface.co/docs/transformers/model_doc/qdqbert)** (NVIDIA から) Hao Wu, Patrick Judd, Xiaojie Zhang, Mikhail Isaev and Paulius Micikevicius から公開された研究論文: [Integer Quantization for Deep Learning Inference: Principles and Empirical Evaluation](https://arxiv.org/abs/2004.09602)
+1. **[Qwen2](https://huggingface.co/docs/transformers/model_doc/qwen2)** (the Qwen team, Alibaba Group から) Jinze Bai, Shuai Bai, Yunfei Chu, Zeyu Cui, Kai Dang, Xiaodong Deng, Yang Fan, Wenbin Ge, Yu Han, Fei Huang, Binyuan Hui, Luo Ji, Mei Li, Junyang Lin, Runji Lin, Dayiheng Liu, Gao Liu, Chengqiang Lu, Keming Lu, Jianxin Ma, Rui Men, Xingzhang Ren, Xuancheng Ren, Chuanqi Tan, Sinan Tan, Jianhong Tu, Peng Wang, Shijie Wang, Wei Wang, Shengguang Wu, Benfeng Xu, Jin Xu, An Yang, Hao Yang, Jian Yang, Shusheng Yang, Yang Yao, Bowen Yu, Hongyi Yuan, Zheng Yuan, Jianwei Zhang, Xingxuan Zhang, Yichang Zhang, Zhenru Zhang, Chang Zhou, Jingren Zhou, Xiaohuan Zhou and Tianhang Zhu. から公開された研究論文 [Qwen Technical Report](https://arxiv.org/abs/2309.16609)
 1. **[RAG](https://huggingface.co/docs/transformers/model_doc/rag)** (Facebook から) Patrick Lewis, Ethan Perez, Aleksandara Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, Sebastian Riedel, Douwe Kiela から公開された研究論文: [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://arxiv.org/abs/2005.11401)
 1. **[REALM](https://huggingface.co/docs/transformers/model_doc/realm.html)** (Google Research から) Kelvin Guu, Kenton Lee, Zora Tung, Panupong Pasupat and Ming-Wei Chang から公開された研究論文: [REALM: Retrieval-Augmented Language Model Pre-Training](https://arxiv.org/abs/2002.08909)
 1. **[Reformer](https://huggingface.co/docs/transformers/model_doc/reformer)** (Google Research から) Nikita Kitaev, Łukasz Kaiser, Anselm Levskaya から公開された研究論文: [Reformer: The Efficient Transformer](https://arxiv.org/abs/2001.04451)
@@ -483,6 +485,7 @@ Flax、PyTorch、TensorFlowをcondaでインストールする方法は、それ
 1. **[Segment Anything](https://huggingface.co/docs/transformers/model_doc/sam)** (Meta AI から) Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alex Berg, Wan-Yen Lo, Piotr Dollar, Ross Girshick. から公開された研究論文 [Segment Anything](https://arxiv.org/pdf/2304.02643v1.pdf)
 1. **[SEW](https://huggingface.co/docs/transformers/model_doc/sew)** (ASAPP から) Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi から公開された研究論文: [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870)
 1. **[SEW-D](https://huggingface.co/docs/transformers/model_doc/sew_d)** (ASAPP から) Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi から公開された研究論文: [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870)
+1. **[SigLIP](https://huggingface.co/docs/transformers/model_doc/siglip)** (Google AI から) Xiaohua Zhai, Basil Mustafa, Alexander Kolesnikov, Lucas Beyer. から公開された研究論文 [Sigmoid Loss for Language Image Pre-Training](https://arxiv.org/abs/2303.15343)
 1. **[SpeechT5](https://huggingface.co/docs/transformers/model_doc/speecht5)** (Microsoft Research から) Junyi Ao, Rui Wang, Long Zhou, Chengyi Wang, Shuo Ren, Yu Wu, Shujie Liu, Tom Ko, Qing Li, Yu Zhang, Zhihua Wei, Yao Qian, Jinyu Li, Furu Wei. から公開された研究論文 [SpeechT5: Unified-Modal Encoder-Decoder Pre-Training for Spoken Language Processing](https://arxiv.org/abs/2110.07205)
 1. **[SpeechToTextTransformer](https://huggingface.co/docs/transformers/model_doc/speech_to_text)** (Facebook から), Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Dmytro Okhonko, Juan Pino から公開された研究論文: [fairseq S2T: Fast Speech-to-Text Modeling with fairseq](https://arxiv.org/abs/2010.05171)
 1. **[SpeechToTextTransformer2](https://huggingface.co/docs/transformers/model_doc/speech_to_text_2)** (Facebook から), Changhan Wang, Anne Wu, Juan Pino, Alexei Baevski, Michael Auli, Alexis Conneau から公開された研究論文: [Large-Scale Self- and Semi-Supervised Learning for Speech Translation](https://arxiv.org/abs/2104.06678)
@@ -509,7 +512,7 @@ Flax、PyTorch、TensorFlowをcondaでインストールする方法は、それ
 1. **[UMT5](https://huggingface.co/docs/transformers/model_doc/umt5)** (Google Research から) Hyung Won Chung, Xavier Garcia, Adam Roberts, Yi Tay, Orhan Firat, Sharan Narang, Noah Constant. から公開された研究論文 [UniMax: Fairer and More Effective Language Sampling for Large-Scale Multilingual Pretraining](https://openreview.net/forum?id=kXwdL1cWOAi)
 1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (Microsoft Research から) Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang から公開された研究論文: [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597)
 1. **[UniSpeechSat](https://huggingface.co/docs/transformers/model_doc/unispeech-sat)** (Microsoft Research から) Sanyuan Chen, Yu Wu, Chengyi Wang, Zhengyang Chen, Zhuo Chen, Shujie Liu, Jian Wu, Yao Qian, Furu Wei, Jinyu Li, Xiangzhan Yu から公開された研究論文: [UNISPEECH-SAT: UNIVERSAL SPEECH REPRESENTATION LEARNING WITH SPEAKER AWARE PRE-TRAINING](https://arxiv.org/abs/2110.05752)
-1. **[UnivNet](https://huggingface.co/docs/transformers/model_doc/univnet)** (from Kakao Corporation) released with the paper [UnivNet: A Neural Vocoder with Multi-Resolution Spectrogram Discriminators for High-Fidelity Waveform Generation](https://arxiv.org/abs/2106.07889) by Won Jang, Dan Lim, Jaesam Yoon, Bongwan Kim, and Juntae Kim. 
+1. **[UnivNet](https://huggingface.co/docs/transformers/model_doc/univnet)** (from Kakao Corporation) released with the paper [UnivNet: A Neural Vocoder with Multi-Resolution Spectrogram Discriminators for High-Fidelity Waveform Generation](https://arxiv.org/abs/2106.07889) by Won Jang, Dan Lim, Jaesam Yoon, Bongwan Kim, and Juntae Kim.
 1. **[UPerNet](https://huggingface.co/docs/transformers/model_doc/upernet)** (Peking University から) Tete Xiao, Yingcheng Liu, Bolei Zhou, Yuning Jiang, Jian Sun. から公開された研究論文 [Unified Perceptual Parsing for Scene Understanding](https://arxiv.org/abs/1807.10221)
 1. **[VAN](https://huggingface.co/docs/transformers/model_doc/van)** (Tsinghua University and Nankai University から) Meng-Hao Guo, Cheng-Ze Lu, Zheng-Ning Liu, Ming-Ming Cheng, Shi-Min Hu から公開された研究論文: [Visual Attention Network](https://arxiv.org/abs/2202.09741)
 1. **[VideoMAE](https://huggingface.co/docs/transformers/model_doc/videomae)** (Multimedia Computing Group, Nanjing University から) Zhan Tong, Yibing Song, Jue Wang, Limin Wang から公開された研究論文: [VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training](https://arxiv.org/abs/2203.12602)
@@ -525,6 +528,7 @@ Flax、PyTorch、TensorFlowをcondaでインストールする方法は、それ
 1. **[VITS](https://huggingface.co/docs/transformers/model_doc/vits)** (Kakao Enterprise から) Jaehyeon Kim, Jungil Kong, Juhee Son. から公開された研究論文 [Conditional Variational Autoencoder with Adversarial Learning for End-to-End Text-to-Speech](https://arxiv.org/abs/2106.06103)
 1. **[ViViT](https://huggingface.co/docs/transformers/model_doc/vivit)** (from Google Research) released with the paper [ViViT: A Video Vision Transformer](https://arxiv.org/abs/2103.15691) by Anurag Arnab, Mostafa Dehghani, Georg Heigold, Chen Sun, Mario Lučić, Cordelia Schmid.
 1. **[Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/wav2vec2)** (Facebook AI から) Alexei Baevski, Henry Zhou, Abdelrahman Mohamed, Michael Auli から公開された研究論文: [wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations](https://arxiv.org/abs/2006.11477)
+1. **[Wav2Vec2-BERT](https://huggingface.co/docs/transformers/model_doc/wav2vec2-bert)** (from Meta AI) released with the paper [Seamless: Multilingual Expressive and Streaming Speech Translation](https://ai.meta.com/research/publications/seamless-multilingual-expressive-and-streaming-speech-translation/) by the Seamless Communication team.
 1. **[Wav2Vec2-Conformer](https://huggingface.co/docs/transformers/model_doc/wav2vec2-conformer)** (Facebook AI から) Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Sravya Popuri, Dmytro Okhonko, Juan Pino から公開された研究論文: [FAIRSEQ S2T: Fast Speech-to-Text Modeling with FAIRSEQ](https://arxiv.org/abs/2010.05171)
 1. **[Wav2Vec2Phoneme](https://huggingface.co/docs/transformers/model_doc/wav2vec2_phoneme)** (Facebook AI から) Qiantong Xu, Alexei Baevski, Michael Auli から公開された研究論文: [Simple and Effective Zero-shot Cross-lingual Phoneme Recognition](https://arxiv.org/abs/2109.11680)
 1. **[WavLM](https://huggingface.co/docs/transformers/model_doc/wavlm)** (Microsoft Research から) Sanyuan Chen, Chengyi Wang, Zhengyang Chen, Yu Wu, Shujie Liu, Zhuo Chen, Jinyu Li, Naoyuki Kanda, Takuya Yoshioka, Xiong Xiao, Jian Wu, Long Zhou, Shuo Ren, Yanmin Qian, Yao Qian, Jian Wu, Michael Zeng, Furu Wei から公開された研究論文: [WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing](https://arxiv.org/abs/2110.13900)

README_ko.md

@@ -176,7 +176,7 @@ limitations under the License.
 
 ### pip로 설치하기
 
-이 저장소는 Python 3.8+, Flax 0.4.1+, PyTorch 1.10+, TensorFlow 2.6+에서 테스트 되었습니다.
+이 저장소는 Python 3.8+, Flax 0.4.1+, PyTorch 1.11+, TensorFlow 2.6+에서 테스트 되었습니다.
 
 [가상 환경](https://docs.python.org/3/library/venv.html)에 🤗 Transformers를 설치하세요. Python 가상 환경에 익숙하지 않다면, [사용자 가이드](https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/)를 확인하세요.
 
@@ -195,14 +195,14 @@ pip install transformers
 
 ### conda로 설치하기
 
-Transformers 버전 v4.0.0부터, conda 채널이 생겼습니다: `huggingface`.
-
 🤗 Transformers는 다음과 같이 conda로 설치할 수 있습니다:
 
 ```shell script
-conda install -c huggingface transformers
+conda install conda-forge::transformers
 ```
 
+> **_노트:_** `huggingface` 채널에서 `transformers`를 설치하는 것은 사용이 중단되었습니다.
+
 Flax, PyTorch, TensorFlow 설치 페이지에서 이들을 conda로 설치하는 방법을 확인하세요.
 
 ## 모델 구조
@@ -245,7 +245,7 @@ Flax, PyTorch, TensorFlow 설치 페이지에서 이들을 conda로 설치하는
 1. **[CLAP](https://huggingface.co/docs/transformers/model_doc/clap)** (LAION-AI 에서 제공)은 Yusong Wu, Ke Chen, Tianyu Zhang, Yuchen Hui, Taylor Berg-Kirkpatrick, Shlomo Dubnov.의 [Large-scale Contrastive Language-Audio Pretraining with Feature Fusion and Keyword-to-Caption Augmentation](https://arxiv.org/abs/2211.06687)논문과 함께 발표했습니다.
 1. **[CLIP](https://huggingface.co/docs/transformers/model_doc/clip)** (OpenAI 에서) Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, Ilya Sutskever 의 [Learning Transferable Visual Models From Natural Language Supervision](https://arxiv.org/abs/2103.00020) 논문과 함께 발표했습니다.
 1. **[CLIPSeg](https://huggingface.co/docs/transformers/model_doc/clipseg)** (University of Göttingen 에서) Timo Lüddecke and Alexander Ecker 의 [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003) 논문과 함께 발표했습니다.
-1. **[CLVP](https://huggingface.co/docs/transformers/model_doc/clvp)** released with the paper [Better speech synthesis through scaling](https://arxiv.org/abs/2305.07243) by James Betker. 
+1. **[CLVP](https://huggingface.co/docs/transformers/model_doc/clvp)** released with the paper [Better speech synthesis through scaling](https://arxiv.org/abs/2305.07243) by James Betker.
 1. **[CodeGen](https://huggingface.co/docs/transformers/model_doc/codegen)** (Salesforce 에서) Erik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu, Huan Wang, Yingbo Zhou, Silvio Savarese, Caiming Xiong 의 [A Conversational Paradigm for Program Synthesis](https://arxiv.org/abs/2203.13474) 논문과 함께 발표했습니다.
 1. **[CodeLlama](https://huggingface.co/docs/transformers/model_doc/llama_code)** (MetaAI 에서 제공)은 Baptiste Rozière, Jonas Gehring, Fabian Gloeckle, Sten Sootla, Itai Gat, Xiaoqing Ellen Tan, Yossi Adi, Jingyu Liu, Tal Remez, Jérémy Rapin, Artyom Kozhevnikov, Ivan Evtimov, Joanna Bitton, Manish Bhatt, Cristian Canton Ferrer, Aaron Grattafiori, Wenhan Xiong, Alexandre Défossez, Jade Copet, Faisal Azhar, Hugo Touvron, Louis Martin, Nicolas Usunier, Thomas Scialom, Gabriel Synnaeve.의 [Code Llama: Open Foundation Models for Code](https://ai.meta.com/research/publications/code-llama-open-foundation-models-for-code/)논문과 함께 발표했습니다.
 1. **[Conditional DETR](https://huggingface.co/docs/transformers/model_doc/conditional_detr)** (Microsoft Research Asia 에서) Depu Meng, Xiaokang Chen, Zejia Fan, Gang Zeng, Houqiang Li, Yuhui Yuan, Lei Sun, Jingdong Wang 의 [Conditional DETR for Fast Training Convergence](https://arxiv.org/abs/2108.06152) 논문과 함께 발표했습니다.
@@ -282,6 +282,7 @@ Flax, PyTorch, TensorFlow 설치 페이지에서 이들을 conda로 설치하는
 1. **[ErnieM](https://huggingface.co/docs/transformers/model_doc/ernie_m)** (Baidu 에서 제공)은 Xuan Ouyang, Shuohuan Wang, Chao Pang, Yu Sun, Hao Tian, Hua Wu, Haifeng Wang.의 [ERNIE-M: Enhanced Multilingual Representation by Aligning Cross-lingual Semantics with Monolingual Corpora](https://arxiv.org/abs/2012.15674)논문과 함께 발표했습니다.
 1. **[ESM](https://huggingface.co/docs/transformers/model_doc/esm)** (from Meta AI) are transformer protein language models.  **ESM-1b** was released with the paper [Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences](https://www.pnas.org/content/118/15/e2016239118) by Alexander Rives, Joshua Meier, Tom Sercu, Siddharth Goyal, Zeming Lin, Jason Liu, Demi Guo, Myle Ott, C. Lawrence Zitnick, Jerry Ma, and Rob Fergus. **ESM-1v** was released with the paper [Language models enable zero-shot prediction of the effects of mutations on protein function](https://doi.org/10.1101/2021.07.09.450648) by Joshua Meier, Roshan Rao, Robert Verkuil, Jason Liu, Tom Sercu and Alexander Rives. **ESM-2** was released with the paper [Language models of protein sequences at the scale of evolution enable accurate structure prediction](https://doi.org/10.1101/2022.07.20.500902) by Zeming Lin, Halil Akin, Roshan Rao, Brian Hie, Zhongkai Zhu, Wenting Lu, Allan dos Santos Costa, Maryam Fazel-Zarandi, Tom Sercu, Sal Candido, Alexander Rives.
 1. **[Falcon](https://huggingface.co/docs/transformers/model_doc/falcon)** (from Technology Innovation Institute) by Almazrouei, Ebtesam and Alobeidli, Hamza and Alshamsi, Abdulaziz and Cappelli, Alessandro and Cojocaru, Ruxandra and Debbah, Merouane and Goffinet, Etienne and Heslow, Daniel and Launay, Julien and Malartic, Quentin and Noune, Badreddine and Pannier, Baptiste and Penedo, Guilherme.
+1. **[FastSpeech2Conformer](model_doc/fastspeech2_conformer)** (ESPnet and Microsoft Research 에서 제공)은 Pengcheng Guo, Florian Boyer, Xuankai Chang, Tomoki Hayashi, Yosuke Higuchi, Hirofumi Inaguma, Naoyuki Kamo, Chenda Li, Daniel Garcia-Romero, Jiatong Shi, Jing Shi, Shinji Watanabe, Kun Wei, Wangyou Zhang, and Yuekai Zhang.의 [Fastspeech 2: Fast And High-quality End-to-End Text To Speech](https://arxiv.org/pdf/2006.04558.pdf)논문과 함께 발표했습니다.
 1. **[FLAN-T5](https://huggingface.co/docs/transformers/model_doc/flan-t5)** (from Google AI) released in the repository [google-research/t5x](https://github.com/google-research/t5x/blob/main/docs/models.md#flan-t5-checkpoints) by Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Eric Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Albert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdhery, Sharan Narang, Gaurav Mishra, Adams Yu, Vincent Zhao, Yanping Huang, Andrew Dai, Hongkun Yu, Slav Petrov, Ed H. Chi, Jeff Dean, Jacob Devlin, Adam Roberts, Denny Zhou, Quoc V. Le, and Jason Wei
 1. **[FLAN-UL2](https://huggingface.co/docs/transformers/model_doc/flan-ul2)** (from Google AI) released in the repository [google-research/t5x](https://github.com/google-research/t5x/blob/main/docs/models.md#flan-ul2-checkpoints) by Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Eric Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Albert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdhery, Sharan Narang, Gaurav Mishra, Adams Yu, Vincent Zhao, Yanping Huang, Andrew Dai, Hongkun Yu, Slav Petrov, Ed H. Chi, Jeff Dean, Jacob Devlin, Adam Roberts, Denny Zhou, Quoc V. Le, and Jason Wei
 1. **[FlauBERT](https://huggingface.co/docs/transformers/model_doc/flaubert)** (from CNRS) released with the paper [FlauBERT: Unsupervised Language Model Pre-training for French](https://arxiv.org/abs/1912.05372) by Hang Le, Loïc Vial, Jibril Frej, Vincent Segonne, Maximin Coavoux, Benjamin Lecouteux, Alexandre Allauzen, Benoît Crabbé, Laurent Besacier, Didier Schwab.
@@ -306,7 +307,7 @@ Flax, PyTorch, TensorFlow 설치 페이지에서 이들을 conda로 설치하는
 1. **[HerBERT](https://huggingface.co/docs/transformers/model_doc/herbert)** (Allegro.pl, AGH University of Science and Technology 에서 제공)은 Piotr Rybak, Robert Mroczkowski, Janusz Tracz, Ireneusz Gawlik.의 [KLEJ: Comprehensive Benchmark for Polish Language Understanding](https://www.aclweb.org/anthology/2020.acl-main.111.pdf)논문과 함께 발표했습니다.
 1. **[Hubert](https://huggingface.co/docs/transformers/model_doc/hubert)** (Facebook 에서) Wei-Ning Hsu, Benjamin Bolte, Yao-Hung Hubert Tsai, Kushal Lakhotia, Ruslan Salakhutdinov, Abdelrahman Mohamed 의 [HuBERT: Self-Supervised Speech Representation Learning by Masked Prediction of Hidden Units](https://arxiv.org/abs/2106.07447) 논문과 함께 발표했습니다.
 1. **[I-BERT](https://huggingface.co/docs/transformers/model_doc/ibert)** (Berkeley 에서) Sehoon Kim, Amir Gholami, Zhewei Yao, Michael W. Mahoney, Kurt Keutzer 의 [I-BERT: Integer-only BERT Quantization](https://arxiv.org/abs/2101.01321) 논문과 함께 발표했습니다.
-1. **[IDEFICS](https://huggingface.co/docs/transformers/model_doc/idefics)** (from HuggingFace) released with the paper [OBELICS: An Open Web-Scale Filtered Dataset of Interleaved Image-Text Documents](https://huggingface.co/papers/2306.16527) by Hugo Laurençon, Lucile Saulnier, Léo Tronchon, Stas Bekman, Amanpreet Singh, Anton Lozhkov, Thomas Wang, Siddharth Karamcheti, Alexander M. Rush, Douwe Kiela, Matthieu Cord, Victor Sanh. 
+1. **[IDEFICS](https://huggingface.co/docs/transformers/model_doc/idefics)** (from HuggingFace) released with the paper [OBELICS: An Open Web-Scale Filtered Dataset of Interleaved Image-Text Documents](https://huggingface.co/papers/2306.16527) by Hugo Laurençon, Lucile Saulnier, Léo Tronchon, Stas Bekman, Amanpreet Singh, Anton Lozhkov, Thomas Wang, Siddharth Karamcheti, Alexander M. Rush, Douwe Kiela, Matthieu Cord, Victor Sanh.
 1. **[ImageGPT](https://huggingface.co/docs/transformers/model_doc/imagegpt)** (OpenAI 에서) Mark Chen, Alec Radford, Rewon Child, Jeffrey Wu, Heewoo Jun, David Luan, Ilya Sutskever 의 [Generative Pretraining from Pixels](https://openai.com/blog/image-gpt/) 논문과 함께 발표했습니다.
 1. **[Informer](https://huggingface.co/docs/transformers/model_doc/informer)** (from Beihang University, UC Berkeley, Rutgers University, SEDD Company) released with the paper [Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecasting](https://arxiv.org/abs/2012.07436) by Haoyi Zhou, Shanghang Zhang, Jieqi Peng, Shuai Zhang, Jianxin Li, Hui Xiong, and Wancai Zhang.
 1. **[InstructBLIP](https://huggingface.co/docs/transformers/model_doc/instructblip)** (Salesforce 에서 제공)은 Wenliang Dai, Junnan Li, Dongxu Li, Anthony Meng Huat Tiong, Junqi Zhao, Weisheng Wang, Boyang Li, Pascale Fung, Steven Hoi.의 [InstructBLIP: Towards General-purpose Vision-Language Models with Instruction Tuning](https://arxiv.org/abs/2305.06500)논문과 함께 발표했습니다.
@@ -340,8 +341,8 @@ Flax, PyTorch, TensorFlow 설치 페이지에서 이들을 conda로 설치하는
 1. **[Megatron-BERT](https://huggingface.co/docs/transformers/model_doc/megatron-bert)** (NVIDIA 에서) Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro 의 [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) 논문과 함께 발표했습니다.
 1. **[Megatron-GPT2](https://huggingface.co/docs/transformers/model_doc/megatron_gpt2)** (NVIDIA 에서) Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro 의 [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) 논문과 함께 발표했습니다.
 1. **[MGP-STR](https://huggingface.co/docs/transformers/model_doc/mgp-str)** (Alibaba Research 에서 제공)은 Peng Wang, Cheng Da, and Cong Yao.의 [Multi-Granularity Prediction for Scene Text Recognition](https://arxiv.org/abs/2209.03592)논문과 함께 발표했습니다.
-1. **[Mistral](https://huggingface.co/docs/transformers/model_doc/mistral)** (from Mistral AI) by The Mistral AI team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed.. 
-1. **[Mixtral](https://huggingface.co/docs/transformers/model_doc/mixtral)** (from Mistral AI) by The [Mistral AI](https://mistral.ai) team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed. 
+1. **[Mistral](https://huggingface.co/docs/transformers/model_doc/mistral)** (from Mistral AI) by The Mistral AI team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed..
+1. **[Mixtral](https://huggingface.co/docs/transformers/model_doc/mixtral)** (from Mistral AI) by The [Mistral AI](https://mistral.ai) team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed.
 1. **[mLUKE](https://huggingface.co/docs/transformers/model_doc/mluke)** (Studio Ousia 에서) Ryokan Ri, Ikuya Yamada, and Yoshimasa Tsuruoka 의 [mLUKE: The Power of Entity Representations in Multilingual Pretrained Language Models](https://arxiv.org/abs/2110.08151) 논문과 함께 발표했습니다.
 1. **[MMS](https://huggingface.co/docs/transformers/model_doc/mms)** (Facebook 에서 제공)은 Vineel Pratap, Andros Tjandra, Bowen Shi, Paden Tomasello, Arun Babu, Sayani Kundu, Ali Elkahky, Zhaoheng Ni, Apoorv Vyas, Maryam Fazel-Zarandi, Alexei Baevski, Yossi Adi, Xiaohui Zhang, Wei-Ning Hsu, Alexis Conneau, Michael Auli.의 [Scaling Speech Technology to 1,000+ Languages](https://arxiv.org/abs/2305.13516)논문과 함께 발표했습니다.
 1. **[MobileBERT](https://huggingface.co/docs/transformers/model_doc/mobilebert)** (CMU/Google Brain 에서) Zhiqing Sun, Hongkun Yu, Xiaodan Song, Renjie Liu, Yiming Yang, and Denny Zhou 의 [MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices](https://arxiv.org/abs/2004.02984) 논문과 함께 발표했습니다.
@@ -377,10 +378,11 @@ Flax, PyTorch, TensorFlow 설치 페이지에서 이들을 conda로 설치하는
 1. **[Pix2Struct](https://huggingface.co/docs/transformers/model_doc/pix2struct)** (Google 에서 제공)은 Kenton Lee, Mandar Joshi, Iulia Turc, Hexiang Hu, Fangyu Liu, Julian Eisenschlos, Urvashi Khandelwal, Peter Shaw, Ming-Wei Chang, Kristina Toutanova.의 [Pix2Struct: Screenshot Parsing as Pretraining for Visual Language Understanding](https://arxiv.org/abs/2210.03347)논문과 함께 발표했습니다.
 1. **[PLBart](https://huggingface.co/docs/transformers/model_doc/plbart)** (UCLA NLP 에서) Wasi Uddin Ahmad, Saikat Chakraborty, Baishakhi Ray, Kai-Wei Chang 의 [Unified Pre-training for Program Understanding and Generation](https://arxiv.org/abs/2103.06333) 논문과 함께 발표했습니다.
 1. **[PoolFormer](https://huggingface.co/docs/transformers/model_doc/poolformer)** (Sea AI Labs 에서) Yu, Weihao and Luo, Mi and Zhou, Pan and Si, Chenyang and Zhou, Yichen and Wang, Xinchao and Feng, Jiashi and Yan, Shuicheng 의 [MetaFormer is Actually What You Need for Vision](https://arxiv.org/abs/2111.11418) 논문과 함께 발표했습니다.
-1. **[Pop2Piano](https://huggingface.co/docs/transformers/model_doc/pop2piano)** released with the paper [Pop2Piano : Pop Audio-based Piano Cover Generation](https://arxiv.org/abs/2211.00895) by Jongho Choi, Kyogu Lee. 
+1. **[Pop2Piano](https://huggingface.co/docs/transformers/model_doc/pop2piano)** released with the paper [Pop2Piano : Pop Audio-based Piano Cover Generation](https://arxiv.org/abs/2211.00895) by Jongho Choi, Kyogu Lee.
 1. **[ProphetNet](https://huggingface.co/docs/transformers/model_doc/prophetnet)** (Microsoft Research 에서) Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou 의 [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) 논문과 함께 발표했습니다.
 1. **[PVT](https://huggingface.co/docs/transformers/model_doc/pvt)** (Nanjing University, The University of Hong Kong etc. 에서 제공)은 Wenhai Wang, Enze Xie, Xiang Li, Deng-Ping Fan, Kaitao Song, Ding Liang, Tong Lu, Ping Luo, Ling Shao.의 [Pyramid Vision Transformer: A Versatile Backbone for Dense Prediction without Convolutions](https://arxiv.org/pdf/2102.12122.pdf)논문과 함께 발표했습니다.
 1. **[QDQBert](https://huggingface.co/docs/transformers/model_doc/qdqbert)** (NVIDIA 에서) Hao Wu, Patrick Judd, Xiaojie Zhang, Mikhail Isaev and Paulius Micikevicius 의 [Integer Quantization for Deep Learning Inference: Principles and Empirical Evaluation](https://arxiv.org/abs/2004.09602) 논문과 함께 발표했습니다.
+1. **[Qwen2](https://huggingface.co/docs/transformers/model_doc/qwen2)** (the Qwen team, Alibaba Group 에서 제공)은 Jinze Bai, Shuai Bai, Yunfei Chu, Zeyu Cui, Kai Dang, Xiaodong Deng, Yang Fan, Wenbin Ge, Yu Han, Fei Huang, Binyuan Hui, Luo Ji, Mei Li, Junyang Lin, Runji Lin, Dayiheng Liu, Gao Liu, Chengqiang Lu, Keming Lu, Jianxin Ma, Rui Men, Xingzhang Ren, Xuancheng Ren, Chuanqi Tan, Sinan Tan, Jianhong Tu, Peng Wang, Shijie Wang, Wei Wang, Shengguang Wu, Benfeng Xu, Jin Xu, An Yang, Hao Yang, Jian Yang, Shusheng Yang, Yang Yao, Bowen Yu, Hongyi Yuan, Zheng Yuan, Jianwei Zhang, Xingxuan Zhang, Yichang Zhang, Zhenru Zhang, Chang Zhou, Jingren Zhou, Xiaohuan Zhou and Tianhang Zhu.의 [Qwen Technical Report](https://arxiv.org/abs/2309.16609)논문과 함께 발표했습니다.
 1. **[RAG](https://huggingface.co/docs/transformers/model_doc/rag)** (Facebook 에서) Patrick Lewis, Ethan Perez, Aleksandara Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, Sebastian Riedel, Douwe Kiela 의 [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://arxiv.org/abs/2005.11401) 논문과 함께 발표했습니다.
 1. **[REALM](https://huggingface.co/docs/transformers/model_doc/realm.html)** (Google Research 에서) Kelvin Guu, Kenton Lee, Zora Tung, Panupong Pasupat and Ming-Wei Chang 의 [REALM: Retrieval-Augmented Language Model Pre-Training](https://arxiv.org/abs/2002.08909) 논문과 함께 발표했습니다.
 1. **[Reformer](https://huggingface.co/docs/transformers/model_doc/reformer)** (Google Research 에서) Nikita Kitaev, Łukasz Kaiser, Anselm Levskaya 의 [Reformer: The Efficient Transformer](https://arxiv.org/abs/2001.04451) 논문과 함께 발표했습니다.
@@ -398,6 +400,7 @@ Flax, PyTorch, TensorFlow 설치 페이지에서 이들을 conda로 설치하는
 1. **[Segment Anything](https://huggingface.co/docs/transformers/model_doc/sam)** (Meta AI 에서 제공)은 Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alex Berg, Wan-Yen Lo, Piotr Dollar, Ross Girshick.의 [Segment Anything](https://arxiv.org/pdf/2304.02643v1.pdf)논문과 함께 발표했습니다.
 1. **[SEW](https://huggingface.co/docs/transformers/model_doc/sew)** (ASAPP 에서) Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi 의 [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) 논문과 함께 발표했습니다.
 1. **[SEW-D](https://huggingface.co/docs/transformers/model_doc/sew_d)** (ASAPP 에서) Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi 의 [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) 논문과 함께 발표했습니다.
+1. **[SigLIP](https://huggingface.co/docs/transformers/model_doc/siglip)** (Google AI 에서 제공)은 Xiaohua Zhai, Basil Mustafa, Alexander Kolesnikov, Lucas Beyer.의 [Sigmoid Loss for Language Image Pre-Training](https://arxiv.org/abs/2303.15343)논문과 함께 발표했습니다.
 1. **[SpeechT5](https://huggingface.co/docs/transformers/model_doc/speecht5)** (Microsoft Research 에서 제공)은 Junyi Ao, Rui Wang, Long Zhou, Chengyi Wang, Shuo Ren, Yu Wu, Shujie Liu, Tom Ko, Qing Li, Yu Zhang, Zhihua Wei, Yao Qian, Jinyu Li, Furu Wei.의 [SpeechT5: Unified-Modal Encoder-Decoder Pre-Training for Spoken Language Processing](https://arxiv.org/abs/2110.07205)논문과 함께 발표했습니다.
 1. **[SpeechToTextTransformer](https://huggingface.co/docs/transformers/model_doc/speech_to_text)** (Facebook 에서) Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Dmytro Okhonko, Juan Pino 의 [fairseq S2T: Fast Speech-to-Text Modeling with fairseq](https://arxiv.org/abs/2010.05171) 논문과 함께 발표했습니다.
 1. **[SpeechToTextTransformer2](https://huggingface.co/docs/transformers/model_doc/speech_to_text_2)** (Facebook 에서) Changhan Wang, Anne Wu, Juan Pino, Alexei Baevski, Michael Auli, Alexis Conneau 의 [Large-Scale Self- and Semi-Supervised Learning for Speech Translation](https://arxiv.org/abs/2104.06678) 논문과 함께 발표했습니다.
@@ -424,7 +427,7 @@ Flax, PyTorch, TensorFlow 설치 페이지에서 이들을 conda로 설치하는
 1. **[UMT5](https://huggingface.co/docs/transformers/model_doc/umt5)** (Google Research 에서 제공)은 Hyung Won Chung, Xavier Garcia, Adam Roberts, Yi Tay, Orhan Firat, Sharan Narang, Noah Constant.의 [UniMax: Fairer and More Effective Language Sampling for Large-Scale Multilingual Pretraining](https://openreview.net/forum?id=kXwdL1cWOAi)논문과 함께 발표했습니다.
 1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (Microsoft Research 에서) Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang 의 [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597) 논문과 함께 발표했습니다.
 1. **[UniSpeechSat](https://huggingface.co/docs/transformers/model_doc/unispeech-sat)** (Microsoft Research 에서) Sanyuan Chen, Yu Wu, Chengyi Wang, Zhengyang Chen, Zhuo Chen, Shujie Liu, Jian Wu, Yao Qian, Furu Wei, Jinyu Li, Xiangzhan Yu 의 [UNISPEECH-SAT: UNIVERSAL SPEECH REPRESENTATION LEARNING WITH SPEAKER AWARE PRE-TRAINING](https://arxiv.org/abs/2110.05752) 논문과 함께 발표했습니다.
-1. **[UnivNet](https://huggingface.co/docs/transformers/model_doc/univnet)** (from Kakao Corporation) released with the paper [UnivNet: A Neural Vocoder with Multi-Resolution Spectrogram Discriminators for High-Fidelity Waveform Generation](https://arxiv.org/abs/2106.07889) by Won Jang, Dan Lim, Jaesam Yoon, Bongwan Kim, and Juntae Kim. 
+1. **[UnivNet](https://huggingface.co/docs/transformers/model_doc/univnet)** (from Kakao Corporation) released with the paper [UnivNet: A Neural Vocoder with Multi-Resolution Spectrogram Discriminators for High-Fidelity Waveform Generation](https://arxiv.org/abs/2106.07889) by Won Jang, Dan Lim, Jaesam Yoon, Bongwan Kim, and Juntae Kim.
 1. **[UPerNet](https://huggingface.co/docs/transformers/model_doc/upernet)** (Peking University 에서 제공)은 Tete Xiao, Yingcheng Liu, Bolei Zhou, Yuning Jiang, Jian Sun.의 [Unified Perceptual Parsing for Scene Understanding](https://arxiv.org/abs/1807.10221)논문과 함께 발표했습니다.
 1. **[VAN](https://huggingface.co/docs/transformers/model_doc/van)** (Tsinghua University and Nankai University 에서) Meng-Hao Guo, Cheng-Ze Lu, Zheng-Ning Liu, Ming-Ming Cheng, Shi-Min Hu 의 [Visual Attention Network](https://arxiv.org/pdf/2202.09741.pdf) 논문과 함께 발표했습니다.
 1. **[VideoMAE](https://huggingface.co/docs/transformers/model_doc/videomae)** (Multimedia Computing Group, Nanjing University 에서) Zhan Tong, Yibing Song, Jue Wang, Limin Wang 의 [VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training](https://arxiv.org/abs/2203.12602) 논문과 함께 발표했습니다.
@@ -440,6 +443,7 @@ Flax, PyTorch, TensorFlow 설치 페이지에서 이들을 conda로 설치하는
 1. **[VITS](https://huggingface.co/docs/transformers/model_doc/vits)** (Kakao Enterprise 에서 제공)은 Jaehyeon Kim, Jungil Kong, Juhee Son.의 [Conditional Variational Autoencoder with Adversarial Learning for End-to-End Text-to-Speech](https://arxiv.org/abs/2106.06103)논문과 함께 발표했습니다.
 1. **[ViViT](https://huggingface.co/docs/transformers/model_doc/vivit)** (from Google Research) released with the paper [ViViT: A Video Vision Transformer](https://arxiv.org/abs/2103.15691) by Anurag Arnab, Mostafa Dehghani, Georg Heigold, Chen Sun, Mario Lučić, Cordelia Schmid.
 1. **[Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/wav2vec2)** (Facebook AI 에서) Alexei Baevski, Henry Zhou, Abdelrahman Mohamed, Michael Auli 의 [wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations](https://arxiv.org/abs/2006.11477) 논문과 함께 발표했습니다.
+1. **[Wav2Vec2-BERT](https://huggingface.co/docs/transformers/model_doc/wav2vec2-bert)** (from Meta AI) released with the paper [Seamless: Multilingual Expressive and Streaming Speech Translation](https://ai.meta.com/research/publications/seamless-multilingual-expressive-and-streaming-speech-translation/) by the Seamless Communication team.
 1. **[Wav2Vec2-Conformer](https://huggingface.co/docs/transformers/model_doc/wav2vec2-conformer)** (Facebook AI 에서) Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Sravya Popuri, Dmytro Okhonko, Juan Pino 의 [FAIRSEQ S2T: Fast Speech-to-Text Modeling with FAIRSEQ](https://arxiv.org/abs/2010.05171) 논문과 함께 발표했습니다.
 1. **[Wav2Vec2Phoneme](https://huggingface.co/docs/transformers/model_doc/wav2vec2_phoneme)** (Facebook AI 에서) Qiantong Xu, Alexei Baevski, Michael Auli 의 [Simple and Effective Zero-shot Cross-lingual Phoneme Recognition](https://arxiv.org/abs/2109.11680) 논문과 함께 발표했습니다.
 1. **[WavLM](https://huggingface.co/docs/transformers/model_doc/wavlm)** (Microsoft Research 에서) Sanyuan Chen, Chengyi Wang, Zhengyang Chen, Yu Wu, Shujie Liu, Zhuo Chen, Jinyu Li, Naoyuki Kanda, Takuya Yoshioka, Xiong Xiao, Jian Wu, Long Zhou, Shuo Ren, Yanmin Qian, Yao Qian, Jian Wu, Michael Zeng, Furu Wei 의 [WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing](https://arxiv.org/abs/2110.13900) 논문과 함께 발표했습니다.

README_pt-br.md

@@ -258,7 +258,7 @@ O modelo em si é um [Pytorch `nn.Module`](https://pytorch.org/docs/stable/nn.ht
 
 ### Com pip
 
-Este repositório é testado no Python 3.8+, Flax 0.4.1+, PyTorch 1.10+ e TensorFlow 2.6+.
+Este repositório é testado no Python 3.8+, Flax 0.4.1+, PyTorch 1.11+ e TensorFlow 2.6+.
 
 Você deve instalar o 🤗 Transformers em um [ambiente virtual](https://docs.python.org/3/library/venv.html). Se você não está familiarizado com ambientes virtuais em Python, confira o [guia do usuário](https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/).
 
@@ -276,15 +276,15 @@ Se você deseja experimentar com os exemplos ou precisa da versão mais recente
 
 ### Com conda
 
-Desde a versão v4.0.0 do Transformers, agora temos um canal conda: `huggingface`.
-
 O 🤗 Transformers pode ser instalado com conda da seguinte forma:
 
 ```bash
-conda install -c huggingface transformers
+conda install conda-forge::transformers
 ```
 
-Siga as páginas de instalação do Flax, PyTorch ou TensorFlow para ver como instalá-los com conda. 
+> **_NOTA:_**  Instalar `transformers` pelo canal `huggingface` está obsoleto.
+
+Siga as páginas de instalação do Flax, PyTorch ou TensorFlow para ver como instalá-los com conda.
 
 Siga as páginas de instalação do Flax, PyTorch ou TensorFlow para ver como instalá-los com o conda.
 
@@ -527,7 +527,7 @@ Número atual de pontos de verificação: ![](https://img.shields.io/endpoint?ur
 1. **[XLS-R](https://huggingface.co/docs/transformers/model_doc/xls_r)** (from Facebook AI) released with the paper [XLS-R: Self-supervised Cross-lingual Speech Representation Learning at Scale](https://arxiv.org/abs/2111.09296) by Arun Babu, Changhan Wang, Andros Tjandra, Kushal Lakhotia, Qiantong Xu, Naman Goyal, Kritika Singh, Patrick von Platen, Yatharth Saraf, Juan Pino, Alexei Baevski, Alexis Conneau, Michael Auli.
 1. **[XLSR-Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/xlsr_wav2vec2)** (from Facebook AI) released with the paper [Unsupervised Cross-Lingual Representation Learning For Speech Recognition](https://arxiv.org/abs/2006.13979) by Alexis Conneau, Alexei Baevski, Ronan Collobert, Abdelrahman Mohamed, Michael Auli.
 1. **[YOLOS](https://huggingface.co/docs/transformers/model_doc/yolos)** (from Huazhong University of Science & Technology) released with the paper [You Only Look at One Sequence: Rethinking Transformer in Vision through Object Detection](https://arxiv.org/abs/2106.00666) by Yuxin Fang, Bencheng Liao, Xinggang Wang, Jiemin Fang, Jiyang Qi, Rui Wu, Jianwei Niu, Wenyu Liu.
-1. **[YOSO](https://huggingface.co/docs/transformers/model_doc/yoso)** (from the University of Wisconsin - Madison) released with the paper [You Only Sample (Almost) Once: Linear Cost Self-Attention Via Bernoulli Sampling](https://arxiv.org/abs/2111.09714) by Zhanpeng Zeng, 
+1. **[YOSO](https://huggingface.co/docs/transformers/model_doc/yoso)** (from the University of Wisconsin - Madison) released with the paper [You Only Sample (Almost) Once: Linear Cost Self-Attention Via Bernoulli Sampling](https://arxiv.org/abs/2111.09714) by Zhanpeng Zeng,
 Yunyang Xiong, Sathya N. Ravi, Shailesh Acharya, Glenn Fung, Vikas Singh.
 
 1. Quer contribuir com um novo modelo? Adicionamos um **guia detalhado e modelos de exemplo** para orientar você no processo de adição de um novo modelo. Você pode encontrá-los na pasta [`templates`](./templates) do repositório. Certifique-se de verificar as [diretrizes de contribuição](./CONTRIBUTING.md) e entrar em contato com os mantenedores ou abrir uma issue para coletar feedback antes de iniciar sua PR.

README_ru.md

@@ -248,7 +248,7 @@ Hugging Face Hub. Мы хотим, чтобы Transformers позволил ра
 
 ### С помощью pip
 
-Данный репозиторий протестирован на Python 3.8+, Flax 0.4.1+, PyTorch 1.10+ и TensorFlow 2.6+.
+Данный репозиторий протестирован на Python 3.8+, Flax 0.4.1+, PyTorch 1.11+ и TensorFlow 2.6+.
 
 Устанавливать 🤗 Transformers следует в [виртуальной среде](https://docs.python.org/3/library/venv.html). Если вы не знакомы с виртуальными средами Python, ознакомьтесь с [руководством пользователя](https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/).
 
@@ -267,14 +267,14 @@ pip install transformers
 
 ### С помощью conda
 
-Начиная с версии Transformers v4.0.0, у нас появилсась поддержка conda: `huggingface`.
-
 Установить Transformers с помощью conda можно следующим образом:
 
 ```bash
-conda install -c huggingface transformers
+conda install conda-forge::transformers
 ```
 
+> **_ЗАМЕТКА:_** Установка `transformers` через канал `huggingface` устарела.
+
 О том, как установить Flax, PyTorch или TensorFlow с помощью conda, читайте на страницах, посвященных их установке.
 
 > **_ЗАМЕТКА:_** В операционной системе Windows вам может быть предложено активировать режим разработчика, чтобы воспользоваться преимуществами кэширования. Если для вас это невозможно, сообщите нам об этом [здесь](https://github.com/huggingface/huggingface_hub/issues/1062).

README_te.md

@@ -225,12 +225,12 @@ limitations under the License.
     - విద్యావేత్తలు మరియు అభ్యాసకుల ప్రవేశానికి తక్కువ అవరోధం.
     - తెలుసుకోవడానికి కేవలం మూడు తరగతులతో కొన్ని వినియోగదారు-ముఖ సంగ్రహణలు.
     - మా అన్ని ప్రీట్రైన్డ్ మోడల్‌లను ఉపయోగించడం కోసం ఏకీకృత API.
-  
+
 2. తక్కువ గణన ఖర్చులు, చిన్న కార్బన్ పాదముద్ర:
     - పరిశోధకులు ఎల్లప్పుడూ మళ్లీ శిక్షణ పొందే బదులు శిక్షణ పొందిన నమూనాలను పంచుకోవచ్చు.
     - అభ్యాసకులు గణన సమయాన్ని మరియు ఉత్పత్తి ఖర్చులను తగ్గించగలరు.
     - అన్ని పద్ధతుల్లో 60,000 కంటే ఎక్కువ ప్రీట్రైన్డ్ మోడల్‌లతో డజన్ల కొద్దీ ఆర్కిటెక్చర్‌లు.
-  
+
 3. మోడల్ జీవితకాలంలో ప్రతి భాగానికి సరైన ఫ్రేమ్‌వర్క్‌ను ఎంచుకోండి:
     - 3 లైన్ల కోడ్‌లో స్టేట్ ఆఫ్ ది ఆర్ట్ మోడల్‌లకు శిక్షణ ఇవ్వండి.
     - TF2.0/PyTorch/JAX ఫ్రేమ్‌వర్క్‌ల మధ్య ఒకే మోడల్‌ను ఇష్టానుసారంగా తరలించండి.
@@ -251,7 +251,7 @@ limitations under the License.
 
 ### పిప్ తో
 
-ఈ రిపోజిటరీ పైథాన్ 3.8+, ఫ్లాక్స్ 0.4.1+, PyTorch 1.10+ మరియు TensorFlow 2.6+లో పరీక్షించబడింది.
+ఈ రిపోజిటరీ పైథాన్ 3.8+, ఫ్లాక్స్ 0.4.1+, PyTorch 1.11+ మరియు TensorFlow 2.6+లో పరీక్షించబడింది.
 
 మీరు [వర్చువల్ వాతావరణం](https://docs.python.org/3/library/venv.html)లో 🤗 ట్రాన్స్‌ఫార్మర్‌లను ఇన్‌స్టాల్ చేయాలి. మీకు పైథాన్ వర్చువల్ పరిసరాల గురించి తెలియకుంటే, [యూజర్ గైడ్](https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/) చూడండి.
 
@@ -270,14 +270,14 @@ pip install transformers
 
 ### కొండా తో
 
-ట్రాన్స్‌ఫార్మర్స్ వెర్షన్ v4.0.0 నుండి, మేము ఇప్పుడు కొండా ఛానెల్‌ని కలిగి ఉన్నాము: `huggingface`.
-
 🤗 కింది విధంగా కొండా ఉపయోగించి ట్రాన్స్‌ఫార్మర్‌లను ఇన్‌స్టాల్ చేయవచ్చు:
 
 ```shell script
-conda install -c huggingface transformers
+conda install conda-forge::transformers
 ```
 
+> **_గమనిక:_** `huggingface` ఛానెల్ నుండి `transformers` ఇన్‌స్టాల్ చేయడం పురాతనంగా ఉంది.
+
 Flax, PyTorch లేదా TensorFlow యొక్క ఇన్‌స్టాలేషన్ పేజీలను కొండాతో ఎలా ఇన్‌స్టాల్ చేయాలో చూడటానికి వాటిని అనుసరించండి.
 
 > **_గమనిక:_** Windowsలో, కాషింగ్ నుండి ప్రయోజనం పొందేందుకు మీరు డెవలపర్ మోడ్‌ని సక్రియం చేయమని ప్రాంప్ట్ చేయబడవచ్చు. ఇది మీకు ఎంపిక కాకపోతే, దయచేసి [ఈ సంచిక](https://github.com/huggingface/huggingface_hub/issues/1062)లో మాకు తెలియజేయండి.

README_zh-hans.md

@@ -201,7 +201,7 @@ checkpoint: 检查点
 
 ### 使用 pip
 
-这个仓库已在 Python 3.8+、Flax 0.4.1+、PyTorch 1.10+ 和 TensorFlow 2.6+ 下经过测试。
+这个仓库已在 Python 3.8+、Flax 0.4.1+、PyTorch 1.11+ 和 TensorFlow 2.6+ 下经过测试。
 
 你可以在[虚拟环境](https://docs.python.org/3/library/venv.html)中安装 🤗 Transformers。如果你还不熟悉 Python 的虚拟环境，请阅此[用户说明](https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/)。
 
@@ -219,14 +219,14 @@ pip install transformers
 
 ### 使用 conda
 
-自 Transformers 4.0.0 版始，我们有了一个 conda 频道： `huggingface`。
-
 🤗 Transformers 可以通过 conda 依此安装：
 
 ```shell script
-conda install -c huggingface transformers
+conda install conda-forge::transformers
 ```
 
+> **_笔记:_** 从 `huggingface` 渠道安装 `transformers` 已被废弃。
+
 要通过 conda 安装 Flax、PyTorch 或 TensorFlow 其中之一，请参阅它们各自安装页的说明。
 
 ## 模型架构
@@ -269,7 +269,7 @@ conda install -c huggingface transformers
 1. **[CLAP](https://huggingface.co/docs/transformers/model_doc/clap)** (来自 LAION-AI) 伴随论文 [Large-scale Contrastive Language-Audio Pretraining with Feature Fusion and Keyword-to-Caption Augmentation](https://arxiv.org/abs/2211.06687) 由 Yusong Wu, Ke Chen, Tianyu Zhang, Yuchen Hui, Taylor Berg-Kirkpatrick, Shlomo Dubnov 发布。
 1. **[CLIP](https://huggingface.co/docs/transformers/model_doc/clip)** (来自 OpenAI) 伴随论文 [Learning Transferable Visual Models From Natural Language Supervision](https://arxiv.org/abs/2103.00020) 由 Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, Ilya Sutskever 发布。
 1. **[CLIPSeg](https://huggingface.co/docs/transformers/model_doc/clipseg)** (来自 University of Göttingen) 伴随论文 [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003) 由 Timo Lüddecke and Alexander Ecker 发布。
-1. **[CLVP](https://huggingface.co/docs/transformers/model_doc/clvp)** released with the paper [Better speech synthesis through scaling](https://arxiv.org/abs/2305.07243) by James Betker. 
+1. **[CLVP](https://huggingface.co/docs/transformers/model_doc/clvp)** released with the paper [Better speech synthesis through scaling](https://arxiv.org/abs/2305.07243) by James Betker.
 1. **[CodeGen](https://huggingface.co/docs/transformers/model_doc/codegen)** (来自 Salesforce) 伴随论文 [A Conversational Paradigm for Program Synthesis](https://arxiv.org/abs/2203.13474) 由 Erik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu, Huan Wang, Yingbo Zhou, Silvio Savarese, Caiming Xiong 发布。
 1. **[CodeLlama](https://huggingface.co/docs/transformers/model_doc/llama_code)** (来自 MetaAI) 伴随论文 [Code Llama: Open Foundation Models for Code](https://ai.meta.com/research/publications/code-llama-open-foundation-models-for-code/) 由 Baptiste Rozière, Jonas Gehring, Fabian Gloeckle, Sten Sootla, Itai Gat, Xiaoqing Ellen Tan, Yossi Adi, Jingyu Liu, Tal Remez, Jérémy Rapin, Artyom Kozhevnikov, Ivan Evtimov, Joanna Bitton, Manish Bhatt, Cristian Canton Ferrer, Aaron Grattafiori, Wenhan Xiong, Alexandre Défossez, Jade Copet, Faisal Azhar, Hugo Touvron, Louis Martin, Nicolas Usunier, Thomas Scialom, Gabriel Synnaeve 发布。
 1. **[Conditional DETR](https://huggingface.co/docs/transformers/model_doc/conditional_detr)** (来自 Microsoft Research Asia) 伴随论文 [Conditional DETR for Fast Training Convergence](https://arxiv.org/abs/2108.06152) 由 Depu Meng, Xiaokang Chen, Zejia Fan, Gang Zeng, Houqiang Li, Yuhui Yuan, Lei Sun, Jingdong Wang 发布。
@@ -306,6 +306,7 @@ conda install -c huggingface transformers
 1. **[ErnieM](https://huggingface.co/docs/transformers/model_doc/ernie_m)** (来自 Baidu) 伴随论文 [ERNIE-M: Enhanced Multilingual Representation by Aligning Cross-lingual Semantics with Monolingual Corpora](https://arxiv.org/abs/2012.15674) 由 Xuan Ouyang, Shuohuan Wang, Chao Pang, Yu Sun, Hao Tian, Hua Wu, Haifeng Wang 发布。
 1. **[ESM](https://huggingface.co/docs/transformers/model_doc/esm)** (from Meta AI) are transformer protein language models.  **ESM-1b** was released with the paper [Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences](https://www.pnas.org/content/118/15/e2016239118) by Alexander Rives, Joshua Meier, Tom Sercu, Siddharth Goyal, Zeming Lin, Jason Liu, Demi Guo, Myle Ott, C. Lawrence Zitnick, Jerry Ma, and Rob Fergus. **ESM-1v** was released with the paper [Language models enable zero-shot prediction of the effects of mutations on protein function](https://doi.org/10.1101/2021.07.09.450648) by Joshua Meier, Roshan Rao, Robert Verkuil, Jason Liu, Tom Sercu and Alexander Rives. **ESM-2** was released with the paper [Language models of protein sequences at the scale of evolution enable accurate structure prediction](https://doi.org/10.1101/2022.07.20.500902) by Zeming Lin, Halil Akin, Roshan Rao, Brian Hie, Zhongkai Zhu, Wenting Lu, Allan dos Santos Costa, Maryam Fazel-Zarandi, Tom Sercu, Sal Candido, Alexander Rives.
 1. **[Falcon](https://huggingface.co/docs/transformers/model_doc/falcon)** (from Technology Innovation Institute) by Almazrouei, Ebtesam and Alobeidli, Hamza and Alshamsi, Abdulaziz and Cappelli, Alessandro and Cojocaru, Ruxandra and Debbah, Merouane and Goffinet, Etienne and Heslow, Daniel and Launay, Julien and Malartic, Quentin and Noune, Badreddine and Pannier, Baptiste and Penedo, Guilherme.
+1. **[FastSpeech2Conformer](model_doc/fastspeech2_conformer)** (来自 ESPnet and Microsoft Research) 伴随论文 [Fastspeech 2: Fast And High-quality End-to-End Text To Speech](https://arxiv.org/pdf/2006.04558.pdf) 由 Pengcheng Guo, Florian Boyer, Xuankai Chang, Tomoki Hayashi, Yosuke Higuchi, Hirofumi Inaguma, Naoyuki Kamo, Chenda Li, Daniel Garcia-Romero, Jiatong Shi, Jing Shi, Shinji Watanabe, Kun Wei, Wangyou Zhang, and Yuekai Zhang 发布。
 1. **[FLAN-T5](https://huggingface.co/docs/transformers/model_doc/flan-t5)** (from Google AI) released in the repository [google-research/t5x](https://github.com/google-research/t5x/blob/main/docs/models.md#flan-t5-checkpoints) by Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Eric Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Albert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdhery, Sharan Narang, Gaurav Mishra, Adams Yu, Vincent Zhao, Yanping Huang, Andrew Dai, Hongkun Yu, Slav Petrov, Ed H. Chi, Jeff Dean, Jacob Devlin, Adam Roberts, Denny Zhou, Quoc V. Le, and Jason Wei
 1. **[FLAN-UL2](https://huggingface.co/docs/transformers/model_doc/flan-ul2)** (from Google AI) released in the repository [google-research/t5x](https://github.com/google-research/t5x/blob/main/docs/models.md#flan-ul2-checkpoints) by Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Eric Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Albert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdhery, Sharan Narang, Gaurav Mishra, Adams Yu, Vincent Zhao, Yanping Huang, Andrew Dai, Hongkun Yu, Slav Petrov, Ed H. Chi, Jeff Dean, Jacob Devlin, Adam Roberts, Denny Zhou, Quoc V. Le, and Jason Wei
 1. **[FlauBERT](https://huggingface.co/docs/transformers/model_doc/flaubert)** (来自 CNRS) 伴随论文 [FlauBERT: Unsupervised Language Model Pre-training for French](https://arxiv.org/abs/1912.05372) 由 Hang Le, Loïc Vial, Jibril Frej, Vincent Segonne, Maximin Coavoux, Benjamin Lecouteux, Alexandre Allauzen, Benoît Crabbé, Laurent Besacier, Didier Schwab 发布。
@@ -330,7 +331,7 @@ conda install -c huggingface transformers
 1. **[HerBERT](https://huggingface.co/docs/transformers/model_doc/herbert)** (来自 Allegro.pl, AGH University of Science and Technology) 伴随论文 [KLEJ: Comprehensive Benchmark for Polish Language Understanding](https://www.aclweb.org/anthology/2020.acl-main.111.pdf) 由 Piotr Rybak, Robert Mroczkowski, Janusz Tracz, Ireneusz Gawlik 发布。
 1. **[Hubert](https://huggingface.co/docs/transformers/model_doc/hubert)** (来自 Facebook) 伴随论文 [HuBERT: Self-Supervised Speech Representation Learning by Masked Prediction of Hidden Units](https://arxiv.org/abs/2106.07447) 由 Wei-Ning Hsu, Benjamin Bolte, Yao-Hung Hubert Tsai, Kushal Lakhotia, Ruslan Salakhutdinov, Abdelrahman Mohamed 发布。
 1. **[I-BERT](https://huggingface.co/docs/transformers/model_doc/ibert)** (来自 Berkeley) 伴随论文 [I-BERT: Integer-only BERT Quantization](https://arxiv.org/abs/2101.01321) 由 Sehoon Kim, Amir Gholami, Zhewei Yao, Michael W. Mahoney, Kurt Keutzer 发布。
-1. **[IDEFICS](https://huggingface.co/docs/transformers/model_doc/idefics)** (from HuggingFace) released with the paper [OBELICS: An Open Web-Scale Filtered Dataset of Interleaved Image-Text Documents](https://huggingface.co/papers/2306.16527) by Hugo Laurençon, Lucile Saulnier, Léo Tronchon, Stas Bekman, Amanpreet Singh, Anton Lozhkov, Thomas Wang, Siddharth Karamcheti, Alexander M. Rush, Douwe Kiela, Matthieu Cord, Victor Sanh. 
+1. **[IDEFICS](https://huggingface.co/docs/transformers/model_doc/idefics)** (from HuggingFace) released with the paper [OBELICS: An Open Web-Scale Filtered Dataset of Interleaved Image-Text Documents](https://huggingface.co/papers/2306.16527) by Hugo Laurençon, Lucile Saulnier, Léo Tronchon, Stas Bekman, Amanpreet Singh, Anton Lozhkov, Thomas Wang, Siddharth Karamcheti, Alexander M. Rush, Douwe Kiela, Matthieu Cord, Victor Sanh.
 1. **[ImageGPT](https://huggingface.co/docs/transformers/model_doc/imagegpt)** (来自 OpenAI) 伴随论文 [Generative Pretraining from Pixels](https://openai.com/blog/image-gpt/) 由 Mark Chen, Alec Radford, Rewon Child, Jeffrey Wu, Heewoo Jun, David Luan, Ilya Sutskever 发布。
 1. **[Informer](https://huggingface.co/docs/transformers/model_doc/informer)** (from Beihang University, UC Berkeley, Rutgers University, SEDD Company) released with the paper [Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecasting](https://arxiv.org/abs/2012.07436) by Haoyi Zhou, Shanghang Zhang, Jieqi Peng, Shuai Zhang, Jianxin Li, Hui Xiong, and Wancai Zhang.
 1. **[InstructBLIP](https://huggingface.co/docs/transformers/model_doc/instructblip)** (来自 Salesforce) 伴随论文 [InstructBLIP: Towards General-purpose Vision-Language Models with Instruction Tuning](https://arxiv.org/abs/2305.06500) 由 Wenliang Dai, Junnan Li, Dongxu Li, Anthony Meng Huat Tiong, Junqi Zhao, Weisheng Wang, Boyang Li, Pascale Fung, Steven Hoi 发布。
@@ -364,8 +365,8 @@ conda install -c huggingface transformers
 1. **[Megatron-BERT](https://huggingface.co/docs/transformers/model_doc/megatron-bert)** (来自 NVIDIA) 伴随论文 [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) 由 Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro 发布。
 1. **[Megatron-GPT2](https://huggingface.co/docs/transformers/model_doc/megatron_gpt2)** (来自 NVIDIA) 伴随论文 [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) 由 Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro 发布。
 1. **[MGP-STR](https://huggingface.co/docs/transformers/model_doc/mgp-str)** (来自 Alibaba Research) 伴随论文 [Multi-Granularity Prediction for Scene Text Recognition](https://arxiv.org/abs/2209.03592) 由 Peng Wang, Cheng Da, and Cong Yao 发布。
-1. **[Mistral](https://huggingface.co/docs/transformers/model_doc/mistral)** (from Mistral AI) by The Mistral AI team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed.. 
-1. **[Mixtral](https://huggingface.co/docs/transformers/model_doc/mixtral)** (from Mistral AI) by The [Mistral AI](https://mistral.ai) team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed. 
+1. **[Mistral](https://huggingface.co/docs/transformers/model_doc/mistral)** (from Mistral AI) by The Mistral AI team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed..
+1. **[Mixtral](https://huggingface.co/docs/transformers/model_doc/mixtral)** (from Mistral AI) by The [Mistral AI](https://mistral.ai) team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed.
 1. **[mLUKE](https://huggingface.co/docs/transformers/model_doc/mluke)** (来自 Studio Ousia) 伴随论文 [mLUKE: The Power of Entity Representations in Multilingual Pretrained Language Models](https://arxiv.org/abs/2110.08151) 由 Ryokan Ri, Ikuya Yamada, and Yoshimasa Tsuruoka 发布。
 1. **[MMS](https://huggingface.co/docs/transformers/model_doc/mms)** (来自 Facebook) 伴随论文 [Scaling Speech Technology to 1,000+ Languages](https://arxiv.org/abs/2305.13516) 由 Vineel Pratap, Andros Tjandra, Bowen Shi, Paden Tomasello, Arun Babu, Sayani Kundu, Ali Elkahky, Zhaoheng Ni, Apoorv Vyas, Maryam Fazel-Zarandi, Alexei Baevski, Yossi Adi, Xiaohui Zhang, Wei-Ning Hsu, Alexis Conneau, Michael Auli 发布。
 1. **[MobileBERT](https://huggingface.co/docs/transformers/model_doc/mobilebert)** (来自 CMU/Google Brain) 伴随论文 [MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices](https://arxiv.org/abs/2004.02984) 由 Zhiqing Sun, Hongkun Yu, Xiaodan Song, Renjie Liu, Yiming Yang, and Denny Zhou 发布。
@@ -401,10 +402,11 @@ conda install -c huggingface transformers
 1. **[Pix2Struct](https://huggingface.co/docs/transformers/model_doc/pix2struct)** (来自 Google) 伴随论文 [Pix2Struct: Screenshot Parsing as Pretraining for Visual Language Understanding](https://arxiv.org/abs/2210.03347) 由 Kenton Lee, Mandar Joshi, Iulia Turc, Hexiang Hu, Fangyu Liu, Julian Eisenschlos, Urvashi Khandelwal, Peter Shaw, Ming-Wei Chang, Kristina Toutanova 发布。
 1. **[PLBart](https://huggingface.co/docs/transformers/model_doc/plbart)** (来自 UCLA NLP) 伴随论文 [Unified Pre-training for Program Understanding and Generation](https://arxiv.org/abs/2103.06333) 由 Wasi Uddin Ahmad, Saikat Chakraborty, Baishakhi Ray, Kai-Wei Chang 发布。
 1. **[PoolFormer](https://huggingface.co/docs/transformers/model_doc/poolformer)** (来自 Sea AI Labs) 伴随论文 [MetaFormer is Actually What You Need for Vision](https://arxiv.org/abs/2111.11418) 由 Yu, Weihao and Luo, Mi and Zhou, Pan and Si, Chenyang and Zhou, Yichen and Wang, Xinchao and Feng, Jiashi and Yan, Shuicheng 发布。
-1. **[Pop2Piano](https://huggingface.co/docs/transformers/model_doc/pop2piano)** released with the paper [Pop2Piano : Pop Audio-based Piano Cover Generation](https://arxiv.org/abs/2211.00895) by Jongho Choi, Kyogu Lee. 
+1. **[Pop2Piano](https://huggingface.co/docs/transformers/model_doc/pop2piano)** released with the paper [Pop2Piano : Pop Audio-based Piano Cover Generation](https://arxiv.org/abs/2211.00895) by Jongho Choi, Kyogu Lee.
 1. **[ProphetNet](https://huggingface.co/docs/transformers/model_doc/prophetnet)** (来自 Microsoft Research) 伴随论文 [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) 由 Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou 发布。
 1. **[PVT](https://huggingface.co/docs/transformers/model_doc/pvt)** (来自 Nanjing University, The University of Hong Kong etc.) 伴随论文 [Pyramid Vision Transformer: A Versatile Backbone for Dense Prediction without Convolutions](https://arxiv.org/pdf/2102.12122.pdf) 由 Wenhai Wang, Enze Xie, Xiang Li, Deng-Ping Fan, Kaitao Song, Ding Liang, Tong Lu, Ping Luo, Ling Shao 发布。
 1. **[QDQBert](https://huggingface.co/docs/transformers/model_doc/qdqbert)** (来自 NVIDIA) 伴随论文 [Integer Quantization for Deep Learning Inference: Principles and Empirical Evaluation](https://arxiv.org/abs/2004.09602) 由 Hao Wu, Patrick Judd, Xiaojie Zhang, Mikhail Isaev and Paulius Micikevicius 发布。
+1. **[Qwen2](https://huggingface.co/docs/transformers/model_doc/qwen2)** (来自 the Qwen team, Alibaba Group) 伴随论文 [Qwen Technical Report](https://arxiv.org/abs/2309.16609) 由 Jinze Bai, Shuai Bai, Yunfei Chu, Zeyu Cui, Kai Dang, Xiaodong Deng, Yang Fan, Wenbin Ge, Yu Han, Fei Huang, Binyuan Hui, Luo Ji, Mei Li, Junyang Lin, Runji Lin, Dayiheng Liu, Gao Liu, Chengqiang Lu, Keming Lu, Jianxin Ma, Rui Men, Xingzhang Ren, Xuancheng Ren, Chuanqi Tan, Sinan Tan, Jianhong Tu, Peng Wang, Shijie Wang, Wei Wang, Shengguang Wu, Benfeng Xu, Jin Xu, An Yang, Hao Yang, Jian Yang, Shusheng Yang, Yang Yao, Bowen Yu, Hongyi Yuan, Zheng Yuan, Jianwei Zhang, Xingxuan Zhang, Yichang Zhang, Zhenru Zhang, Chang Zhou, Jingren Zhou, Xiaohuan Zhou and Tianhang Zhu 发布。
 1. **[RAG](https://huggingface.co/docs/transformers/model_doc/rag)** (来自 Facebook) 伴随论文 [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://arxiv.org/abs/2005.11401) 由 Patrick Lewis, Ethan Perez, Aleksandara Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, Sebastian Riedel, Douwe Kiela 发布。
 1. **[REALM](https://huggingface.co/docs/transformers/model_doc/realm.html)** (来自 Google Research) 伴随论文 [REALM: Retrieval-Augmented Language Model Pre-Training](https://arxiv.org/abs/2002.08909) 由 Kelvin Guu, Kenton Lee, Zora Tung, Panupong Pasupat and Ming-Wei Chang 发布。
 1. **[Reformer](https://huggingface.co/docs/transformers/model_doc/reformer)** (来自 Google Research) 伴随论文 [Reformer: The Efficient Transformer](https://arxiv.org/abs/2001.04451) 由 Nikita Kitaev, Łukasz Kaiser, Anselm Levskaya 发布。
@@ -422,6 +424,7 @@ conda install -c huggingface transformers
 1. **[Segment Anything](https://huggingface.co/docs/transformers/model_doc/sam)** (来自 Meta AI) 伴随论文 [Segment Anything](https://arxiv.org/pdf/2304.02643v1.pdf) 由 Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alex Berg, Wan-Yen Lo, Piotr Dollar, Ross Girshick 发布。
 1. **[SEW](https://huggingface.co/docs/transformers/model_doc/sew)** (来自 ASAPP) 伴随论文 [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) 由 Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi 发布。
 1. **[SEW-D](https://huggingface.co/docs/transformers/model_doc/sew_d)** (来自 ASAPP) 伴随论文 [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) 由 Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi 发布。
+1. **[SigLIP](https://huggingface.co/docs/transformers/model_doc/siglip)** (来自 Google AI) 伴随论文 [Sigmoid Loss for Language Image Pre-Training](https://arxiv.org/abs/2303.15343) 由 Xiaohua Zhai, Basil Mustafa, Alexander Kolesnikov, Lucas Beyer 发布。
 1. **[SpeechT5](https://huggingface.co/docs/transformers/model_doc/speecht5)** (来自 Microsoft Research) 伴随论文 [SpeechT5: Unified-Modal Encoder-Decoder Pre-Training for Spoken Language Processing](https://arxiv.org/abs/2110.07205) 由 Junyi Ao, Rui Wang, Long Zhou, Chengyi Wang, Shuo Ren, Yu Wu, Shujie Liu, Tom Ko, Qing Li, Yu Zhang, Zhihua Wei, Yao Qian, Jinyu Li, Furu Wei 发布。
 1. **[SpeechToTextTransformer](https://huggingface.co/docs/transformers/model_doc/speech_to_text)** (来自 Facebook), 伴随论文 [fairseq S2T: Fast Speech-to-Text Modeling with fairseq](https://arxiv.org/abs/2010.05171) 由 Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Dmytro Okhonko, Juan Pino 发布。
 1. **[SpeechToTextTransformer2](https://huggingface.co/docs/transformers/model_doc/speech_to_text_2)** (来自 Facebook) 伴随论文 [Large-Scale Self- and Semi-Supervised Learning for Speech Translation](https://arxiv.org/abs/2104.06678) 由 Changhan Wang, Anne Wu, Juan Pino, Alexei Baevski, Michael Auli, Alexis Conneau 发布。
@@ -448,7 +451,7 @@ conda install -c huggingface transformers
 1. **[UMT5](https://huggingface.co/docs/transformers/model_doc/umt5)** (来自 Google Research) 伴随论文 [UniMax: Fairer and More Effective Language Sampling for Large-Scale Multilingual Pretraining](https://openreview.net/forum?id=kXwdL1cWOAi) 由 Hyung Won Chung, Xavier Garcia, Adam Roberts, Yi Tay, Orhan Firat, Sharan Narang, Noah Constant 发布。
 1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (来自 Microsoft Research) 伴随论文 [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597) 由 Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang 发布。
 1. **[UniSpeechSat](https://huggingface.co/docs/transformers/model_doc/unispeech-sat)** (来自 Microsoft Research) 伴随论文 [UNISPEECH-SAT: UNIVERSAL SPEECH REPRESENTATION LEARNING WITH SPEAKER AWARE PRE-TRAINING](https://arxiv.org/abs/2110.05752) 由 Sanyuan Chen, Yu Wu, Chengyi Wang, Zhengyang Chen, Zhuo Chen, Shujie Liu, Jian Wu, Yao Qian, Furu Wei, Jinyu Li, Xiangzhan Yu 发布。
-1. **[UnivNet](https://huggingface.co/docs/transformers/model_doc/univnet)** (from Kakao Corporation) released with the paper [UnivNet: A Neural Vocoder with Multi-Resolution Spectrogram Discriminators for High-Fidelity Waveform Generation](https://arxiv.org/abs/2106.07889) by Won Jang, Dan Lim, Jaesam Yoon, Bongwan Kim, and Juntae Kim. 
+1. **[UnivNet](https://huggingface.co/docs/transformers/model_doc/univnet)** (from Kakao Corporation) released with the paper [UnivNet: A Neural Vocoder with Multi-Resolution Spectrogram Discriminators for High-Fidelity Waveform Generation](https://arxiv.org/abs/2106.07889) by Won Jang, Dan Lim, Jaesam Yoon, Bongwan Kim, and Juntae Kim.
 1. **[UPerNet](https://huggingface.co/docs/transformers/model_doc/upernet)** (来自 Peking University) 伴随论文 [Unified Perceptual Parsing for Scene Understanding](https://arxiv.org/abs/1807.10221) 由 Tete Xiao, Yingcheng Liu, Bolei Zhou, Yuning Jiang, Jian Sun 发布。
 1. **[VAN](https://huggingface.co/docs/transformers/model_doc/van)** (来自 Tsinghua University and Nankai University) 伴随论文 [Visual Attention Network](https://arxiv.org/pdf/2202.09741.pdf) 由 Meng-Hao Guo, Cheng-Ze Lu, Zheng-Ning Liu, Ming-Ming Cheng, Shi-Min Hu 发布。
 1. **[VideoMAE](https://huggingface.co/docs/transformers/model_doc/videomae)** (来自 Multimedia Computing Group, Nanjing University) 伴随论文 [VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training](https://arxiv.org/abs/2203.12602) 由 Zhan Tong, Yibing Song, Jue Wang, Limin Wang 发布。
@@ -464,6 +467,7 @@ conda install -c huggingface transformers
 1. **[VITS](https://huggingface.co/docs/transformers/model_doc/vits)** (来自 Kakao Enterprise) 伴随论文 [Conditional Variational Autoencoder with Adversarial Learning for End-to-End Text-to-Speech](https://arxiv.org/abs/2106.06103) 由 Jaehyeon Kim, Jungil Kong, Juhee Son 发布。
 1. **[ViViT](https://huggingface.co/docs/transformers/model_doc/vivit)** (来自 Google Research) released with the paper [ViViT: A Video Vision Transformer](https://arxiv.org/abs/2103.15691) 由 Anurag Arnab, Mostafa Dehghani, Georg Heigold, Chen Sun, Mario Lučić, Cordelia Schmid.
 1. **[Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/wav2vec2)** (来自 Facebook AI) 伴随论文 [wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations](https://arxiv.org/abs/2006.11477) 由 Alexei Baevski, Henry Zhou, Abdelrahman Mohamed, Michael Auli 发布。
+1. **[Wav2Vec2-BERT](https://huggingface.co/docs/transformers/model_doc/wav2vec2-bert)** (from Meta AI) released with the paper [Seamless: Multilingual Expressive and Streaming Speech Translation](https://ai.meta.com/research/publications/seamless-multilingual-expressive-and-streaming-speech-translation/) by the Seamless Communication team.
 1. **[Wav2Vec2-Conformer](https://huggingface.co/docs/transformers/model_doc/wav2vec2-conformer)** (来自 Facebook AI) 伴随论文 [FAIRSEQ S2T: Fast Speech-to-Text Modeling with FAIRSEQ](https://arxiv.org/abs/2010.05171) 由 Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Sravya Popuri, Dmytro Okhonko, Juan Pino 发布。
 1. **[Wav2Vec2Phoneme](https://huggingface.co/docs/transformers/model_doc/wav2vec2_phoneme)** (来自 Facebook AI) 伴随论文 [Simple and Effective Zero-shot Cross-lingual Phoneme Recognition](https://arxiv.org/abs/2109.11680) 由 Qiantong Xu, Alexei Baevski, Michael Auli 发布。
 1. **[WavLM](https://huggingface.co/docs/transformers/model_doc/wavlm)** (from Microsoft Research) released with the paper [WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing](https://arxiv.org/abs/2110.13900) by Sanyuan Chen, Chengyi Wang, Zhengyang Chen, Yu Wu, Shujie Liu, Zhuo Chen, Jinyu Li, Naoyuki Kanda, Takuya Yoshioka, Xiong Xiao, Jian Wu, Long Zhou, Shuo Ren, Yanmin Qian, Yao Qian, Jian Wu, Michael Zeng, Furu Wei.

README_zh-hant.md

@@ -213,7 +213,7 @@ Tokenizer 為所有的預訓練模型提供了預處理，並可以直接轉換
 
 ### 使用 pip
 
-這個 Repository 已在 Python 3.8+、Flax 0.4.1+、PyTorch 1.10+ 和 TensorFlow 2.6+ 下經過測試。
+這個 Repository 已在 Python 3.8+、Flax 0.4.1+、PyTorch 1.11+ 和 TensorFlow 2.6+ 下經過測試。
 
 你可以在[虛擬環境](https://docs.python.org/3/library/venv.html)中安裝 🤗 Transformers。如果你還不熟悉 Python 的虛擬環境，請閱此[使用者指引](https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/)。
 
@@ -231,14 +231,14 @@ pip install transformers
 
 ### 使用 conda
 
-自 Transformers 4.0.0 版始，我們有了一個 conda channel： `huggingface`。
-
 🤗 Transformers 可以藉由 conda 依此安裝：
 
 ```shell script
-conda install -c huggingface transformers
+conda install conda-forge::transformers
 ```
 
+> **_筆記:_** 從 `huggingface` 頻道安裝 `transformers` 已被淘汰。
+
 要藉由 conda 安裝 Flax、PyTorch 或 TensorFlow 其中之一，請參閱它們各自安裝頁面的說明。
 
 ## 模型架構
@@ -281,7 +281,7 @@ conda install -c huggingface transformers
 1. **[CLAP](https://huggingface.co/docs/transformers/model_doc/clap)** (from LAION-AI) released with the paper [Large-scale Contrastive Language-Audio Pretraining with Feature Fusion and Keyword-to-Caption Augmentation](https://arxiv.org/abs/2211.06687) by Yusong Wu, Ke Chen, Tianyu Zhang, Yuchen Hui, Taylor Berg-Kirkpatrick, Shlomo Dubnov.
 1. **[CLIP](https://huggingface.co/docs/transformers/model_doc/clip)** (from OpenAI) released with the paper [Learning Transferable Visual Models From Natural Language Supervision](https://arxiv.org/abs/2103.00020) by Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, Ilya Sutskever.
 1. **[CLIPSeg](https://huggingface.co/docs/transformers/model_doc/clipseg)** (from University of Göttingen) released with the paper [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003) by Timo Lüddecke and Alexander Ecker.
-1. **[CLVP](https://huggingface.co/docs/transformers/model_doc/clvp)** released with the paper [Better speech synthesis through scaling](https://arxiv.org/abs/2305.07243) by James Betker. 
+1. **[CLVP](https://huggingface.co/docs/transformers/model_doc/clvp)** released with the paper [Better speech synthesis through scaling](https://arxiv.org/abs/2305.07243) by James Betker.
 1. **[CodeGen](https://huggingface.co/docs/transformers/model_doc/codegen)** (from Salesforce) released with the paper [A Conversational Paradigm for Program Synthesis](https://arxiv.org/abs/2203.13474) by Erik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu, Huan Wang, Yingbo Zhou, Silvio Savarese, Caiming Xiong.
 1. **[CodeLlama](https://huggingface.co/docs/transformers/model_doc/llama_code)** (from MetaAI) released with the paper [Code Llama: Open Foundation Models for Code](https://ai.meta.com/research/publications/code-llama-open-foundation-models-for-code/) by Baptiste Rozière, Jonas Gehring, Fabian Gloeckle, Sten Sootla, Itai Gat, Xiaoqing Ellen Tan, Yossi Adi, Jingyu Liu, Tal Remez, Jérémy Rapin, Artyom Kozhevnikov, Ivan Evtimov, Joanna Bitton, Manish Bhatt, Cristian Canton Ferrer, Aaron Grattafiori, Wenhan Xiong, Alexandre Défossez, Jade Copet, Faisal Azhar, Hugo Touvron, Louis Martin, Nicolas Usunier, Thomas Scialom, Gabriel Synnaeve.
 1. **[Conditional DETR](https://huggingface.co/docs/transformers/model_doc/conditional_detr)** (from Microsoft Research Asia) released with the paper [Conditional DETR for Fast Training Convergence](https://arxiv.org/abs/2108.06152) by Depu Meng, Xiaokang Chen, Zejia Fan, Gang Zeng, Houqiang Li, Yuhui Yuan, Lei Sun, Jingdong Wang.
@@ -318,6 +318,7 @@ conda install -c huggingface transformers
 1. **[ErnieM](https://huggingface.co/docs/transformers/model_doc/ernie_m)** (from Baidu) released with the paper [ERNIE-M: Enhanced Multilingual Representation by Aligning Cross-lingual Semantics with Monolingual Corpora](https://arxiv.org/abs/2012.15674) by Xuan Ouyang, Shuohuan Wang, Chao Pang, Yu Sun, Hao Tian, Hua Wu, Haifeng Wang.
 1. **[ESM](https://huggingface.co/docs/transformers/model_doc/esm)** (from Meta AI) are transformer protein language models.  **ESM-1b** was released with the paper [Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences](https://www.pnas.org/content/118/15/e2016239118) by Alexander Rives, Joshua Meier, Tom Sercu, Siddharth Goyal, Zeming Lin, Jason Liu, Demi Guo, Myle Ott, C. Lawrence Zitnick, Jerry Ma, and Rob Fergus. **ESM-1v** was released with the paper [Language models enable zero-shot prediction of the effects of mutations on protein function](https://doi.org/10.1101/2021.07.09.450648) by Joshua Meier, Roshan Rao, Robert Verkuil, Jason Liu, Tom Sercu and Alexander Rives. **ESM-2** was released with the paper [Language models of protein sequences at the scale of evolution enable accurate structure prediction](https://doi.org/10.1101/2022.07.20.500902) by Zeming Lin, Halil Akin, Roshan Rao, Brian Hie, Zhongkai Zhu, Wenting Lu, Allan dos Santos Costa, Maryam Fazel-Zarandi, Tom Sercu, Sal Candido, Alexander Rives.
 1. **[Falcon](https://huggingface.co/docs/transformers/model_doc/falcon)** (from Technology Innovation Institute) by Almazrouei, Ebtesam and Alobeidli, Hamza and Alshamsi, Abdulaziz and Cappelli, Alessandro and Cojocaru, Ruxandra and Debbah, Merouane and Goffinet, Etienne and Heslow, Daniel and Launay, Julien and Malartic, Quentin and Noune, Badreddine and Pannier, Baptiste and Penedo, Guilherme.
+1. **[FastSpeech2Conformer](model_doc/fastspeech2_conformer)** (from ESPnet and Microsoft Research) released with the paper [Fastspeech 2: Fast And High-quality End-to-End Text To Speech](https://arxiv.org/pdf/2006.04558.pdf) by Pengcheng Guo, Florian Boyer, Xuankai Chang, Tomoki Hayashi, Yosuke Higuchi, Hirofumi Inaguma, Naoyuki Kamo, Chenda Li, Daniel Garcia-Romero, Jiatong Shi, Jing Shi, Shinji Watanabe, Kun Wei, Wangyou Zhang, and Yuekai Zhang.
 1. **[FLAN-T5](https://huggingface.co/docs/transformers/model_doc/flan-t5)** (from Google AI) released in the repository [google-research/t5x](https://github.com/google-research/t5x/blob/main/docs/models.md#flan-t5-checkpoints) by Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Eric Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Albert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdhery, Sharan Narang, Gaurav Mishra, Adams Yu, Vincent Zhao, Yanping Huang, Andrew Dai, Hongkun Yu, Slav Petrov, Ed H. Chi, Jeff Dean, Jacob Devlin, Adam Roberts, Denny Zhou, Quoc V. Le, and Jason Wei
 1. **[FLAN-UL2](https://huggingface.co/docs/transformers/model_doc/flan-ul2)** (from Google AI) released in the repository [google-research/t5x](https://github.com/google-research/t5x/blob/main/docs/models.md#flan-ul2-checkpoints) by Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Eric Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Albert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdhery, Sharan Narang, Gaurav Mishra, Adams Yu, Vincent Zhao, Yanping Huang, Andrew Dai, Hongkun Yu, Slav Petrov, Ed H. Chi, Jeff Dean, Jacob Devlin, Adam Roberts, Denny Zhou, Quoc V. Le, and Jason Wei
 1. **[FlauBERT](https://huggingface.co/docs/transformers/model_doc/flaubert)** (from CNRS) released with the paper [FlauBERT: Unsupervised Language Model Pre-training for French](https://arxiv.org/abs/1912.05372) by Hang Le, Loïc Vial, Jibril Frej, Vincent Segonne, Maximin Coavoux, Benjamin Lecouteux, Alexandre Allauzen, Benoît Crabbé, Laurent Besacier, Didier Schwab.
@@ -342,7 +343,7 @@ conda install -c huggingface transformers
 1. **[HerBERT](https://huggingface.co/docs/transformers/model_doc/herbert)** (from Allegro.pl, AGH University of Science and Technology) released with the paper [KLEJ: Comprehensive Benchmark for Polish Language Understanding](https://www.aclweb.org/anthology/2020.acl-main.111.pdf) by Piotr Rybak, Robert Mroczkowski, Janusz Tracz, Ireneusz Gawlik.
 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.
-1. **[IDEFICS](https://huggingface.co/docs/transformers/model_doc/idefics)** (from HuggingFace) released with the paper [OBELICS: An Open Web-Scale Filtered Dataset of Interleaved Image-Text Documents](https://huggingface.co/papers/2306.16527) by Hugo Laurençon, Lucile Saulnier, Léo Tronchon, Stas Bekman, Amanpreet Singh, Anton Lozhkov, Thomas Wang, Siddharth Karamcheti, Alexander M. Rush, Douwe Kiela, Matthieu Cord, Victor Sanh. 
+1. **[IDEFICS](https://huggingface.co/docs/transformers/model_doc/idefics)** (from HuggingFace) released with the paper [OBELICS: An Open Web-Scale Filtered Dataset of Interleaved Image-Text Documents](https://huggingface.co/papers/2306.16527) by Hugo Laurençon, Lucile Saulnier, Léo Tronchon, Stas Bekman, Amanpreet Singh, Anton Lozhkov, Thomas Wang, Siddharth Karamcheti, Alexander M. Rush, Douwe Kiela, Matthieu Cord, Victor Sanh.
 1. **[ImageGPT](https://huggingface.co/docs/transformers/model_doc/imagegpt)** (from OpenAI) released with the paper [Generative Pretraining from Pixels](https://openai.com/blog/image-gpt/) by Mark Chen, Alec Radford, Rewon Child, Jeffrey Wu, Heewoo Jun, David Luan, Ilya Sutskever.
 1. **[Informer](https://huggingface.co/docs/transformers/model_doc/informer)** (from Beihang University, UC Berkeley, Rutgers University, SEDD Company) released with the paper [Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecasting](https://arxiv.org/abs/2012.07436) by Haoyi Zhou, Shanghang Zhang, Jieqi Peng, Shuai Zhang, Jianxin Li, Hui Xiong, and Wancai Zhang.
 1. **[InstructBLIP](https://huggingface.co/docs/transformers/model_doc/instructblip)** (from Salesforce) released with the paper [InstructBLIP: Towards General-purpose Vision-Language Models with Instruction Tuning](https://arxiv.org/abs/2305.06500) by Wenliang Dai, Junnan Li, Dongxu Li, Anthony Meng Huat Tiong, Junqi Zhao, Weisheng Wang, Boyang Li, Pascale Fung, Steven Hoi.
@@ -376,8 +377,8 @@ conda install -c huggingface transformers
 1. **[Megatron-BERT](https://huggingface.co/docs/transformers/model_doc/megatron-bert)** (from NVIDIA) released with the paper [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) by Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro.
 1. **[Megatron-GPT2](https://huggingface.co/docs/transformers/model_doc/megatron_gpt2)** (from NVIDIA) released with the paper [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) by Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro.
 1. **[MGP-STR](https://huggingface.co/docs/transformers/model_doc/mgp-str)** (from Alibaba Research) released with the paper [Multi-Granularity Prediction for Scene Text Recognition](https://arxiv.org/abs/2209.03592) by Peng Wang, Cheng Da, and Cong Yao.
-1. **[Mistral](https://huggingface.co/docs/transformers/model_doc/mistral)** (from Mistral AI) by The Mistral AI team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed.. 
-1. **[Mixtral](https://huggingface.co/docs/transformers/model_doc/mixtral)** (from Mistral AI) by The [Mistral AI](https://mistral.ai) team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed. 
+1. **[Mistral](https://huggingface.co/docs/transformers/model_doc/mistral)** (from Mistral AI) by The Mistral AI team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed..
+1. **[Mixtral](https://huggingface.co/docs/transformers/model_doc/mixtral)** (from Mistral AI) by The [Mistral AI](https://mistral.ai) team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed.
 1. **[mLUKE](https://huggingface.co/docs/transformers/model_doc/mluke)** (from Studio Ousia) released with the paper [mLUKE: The Power of Entity Representations in Multilingual Pretrained Language Models](https://arxiv.org/abs/2110.08151) by Ryokan Ri, Ikuya Yamada, and Yoshimasa Tsuruoka.
 1. **[MMS](https://huggingface.co/docs/transformers/model_doc/mms)** (from Facebook) released with the paper [Scaling Speech Technology to 1,000+ Languages](https://arxiv.org/abs/2305.13516) by Vineel Pratap, Andros Tjandra, Bowen Shi, Paden Tomasello, Arun Babu, Sayani Kundu, Ali Elkahky, Zhaoheng Ni, Apoorv Vyas, Maryam Fazel-Zarandi, Alexei Baevski, Yossi Adi, Xiaohui Zhang, Wei-Ning Hsu, Alexis Conneau, Michael Auli.
 1. **[MobileBERT](https://huggingface.co/docs/transformers/model_doc/mobilebert)** (from CMU/Google Brain) released with the paper [MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices](https://arxiv.org/abs/2004.02984) by Zhiqing Sun, Hongkun Yu, Xiaodan Song, Renjie Liu, Yiming Yang, and Denny Zhou.
@@ -413,10 +414,11 @@ conda install -c huggingface transformers
 1. **[Pix2Struct](https://huggingface.co/docs/transformers/model_doc/pix2struct)** (from Google) released with the paper [Pix2Struct: Screenshot Parsing as Pretraining for Visual Language Understanding](https://arxiv.org/abs/2210.03347) by Kenton Lee, Mandar Joshi, Iulia Turc, Hexiang Hu, Fangyu Liu, Julian Eisenschlos, Urvashi Khandelwal, Peter Shaw, Ming-Wei Chang, Kristina Toutanova.
 1. **[PLBart](https://huggingface.co/docs/transformers/model_doc/plbart)** (from UCLA NLP) released with the paper [Unified Pre-training for Program Understanding and Generation](https://arxiv.org/abs/2103.06333) by Wasi Uddin Ahmad, Saikat Chakraborty, Baishakhi Ray, Kai-Wei Chang.
 1. **[PoolFormer](https://huggingface.co/docs/transformers/model_doc/poolformer)** (from Sea AI Labs) released with the paper [MetaFormer is Actually What You Need for Vision](https://arxiv.org/abs/2111.11418) by Yu, Weihao and Luo, Mi and Zhou, Pan and Si, Chenyang and Zhou, Yichen and Wang, Xinchao and Feng, Jiashi and Yan, Shuicheng.
-1. **[Pop2Piano](https://huggingface.co/docs/transformers/model_doc/pop2piano)** released with the paper [Pop2Piano : Pop Audio-based Piano Cover Generation](https://arxiv.org/abs/2211.00895) by Jongho Choi, Kyogu Lee. 
+1. **[Pop2Piano](https://huggingface.co/docs/transformers/model_doc/pop2piano)** released with the paper [Pop2Piano : Pop Audio-based Piano Cover Generation](https://arxiv.org/abs/2211.00895) by Jongho Choi, Kyogu Lee.
 1. **[ProphetNet](https://huggingface.co/docs/transformers/model_doc/prophetnet)** (from Microsoft Research) released with the paper [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) by Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou.
 1. **[PVT](https://huggingface.co/docs/transformers/model_doc/pvt)** (from Nanjing University, The University of Hong Kong etc.) released with the paper [Pyramid Vision Transformer: A Versatile Backbone for Dense Prediction without Convolutions](https://arxiv.org/pdf/2102.12122.pdf) by Wenhai Wang, Enze Xie, Xiang Li, Deng-Ping Fan, Kaitao Song, Ding Liang, Tong Lu, Ping Luo, Ling Shao.
 1. **[QDQBert](https://huggingface.co/docs/transformers/model_doc/qdqbert)** (from NVIDIA) released with the paper [Integer Quantization for Deep Learning Inference: Principles and Empirical Evaluation](https://arxiv.org/abs/2004.09602) by Hao Wu, Patrick Judd, Xiaojie Zhang, Mikhail Isaev and Paulius Micikevicius.
+1. **[Qwen2](https://huggingface.co/docs/transformers/model_doc/qwen2)** (from the Qwen team, Alibaba Group) released with the paper [Qwen Technical Report](https://arxiv.org/abs/2309.16609) by Jinze Bai, Shuai Bai, Yunfei Chu, Zeyu Cui, Kai Dang, Xiaodong Deng, Yang Fan, Wenbin Ge, Yu Han, Fei Huang, Binyuan Hui, Luo Ji, Mei Li, Junyang Lin, Runji Lin, Dayiheng Liu, Gao Liu, Chengqiang Lu, Keming Lu, Jianxin Ma, Rui Men, Xingzhang Ren, Xuancheng Ren, Chuanqi Tan, Sinan Tan, Jianhong Tu, Peng Wang, Shijie Wang, Wei Wang, Shengguang Wu, Benfeng Xu, Jin Xu, An Yang, Hao Yang, Jian Yang, Shusheng Yang, Yang Yao, Bowen Yu, Hongyi Yuan, Zheng Yuan, Jianwei Zhang, Xingxuan Zhang, Yichang Zhang, Zhenru Zhang, Chang Zhou, Jingren Zhou, Xiaohuan Zhou and Tianhang Zhu.
 1. **[RAG](https://huggingface.co/docs/transformers/model_doc/rag)** (from Facebook) released with the paper [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://arxiv.org/abs/2005.11401) by Patrick Lewis, Ethan Perez, Aleksandara Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, Sebastian Riedel, Douwe Kiela.
 1. **[REALM](https://huggingface.co/docs/transformers/model_doc/realm.html)** (from Google Research) released with the paper [REALM: Retrieval-Augmented Language Model Pre-Training](https://arxiv.org/abs/2002.08909) by Kelvin Guu, Kenton Lee, Zora Tung, Panupong Pasupat and Ming-Wei Chang.
 1. **[Reformer](https://huggingface.co/docs/transformers/model_doc/reformer)** (from Google Research) released with the paper [Reformer: The Efficient Transformer](https://arxiv.org/abs/2001.04451) by Nikita Kitaev, Łukasz Kaiser, Anselm Levskaya.
@@ -434,6 +436,7 @@ conda install -c huggingface transformers
 1. **[Segment Anything](https://huggingface.co/docs/transformers/model_doc/sam)** (from Meta AI) released with the paper [Segment Anything](https://arxiv.org/pdf/2304.02643v1.pdf) by Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alex Berg, Wan-Yen Lo, Piotr Dollar, Ross Girshick.
 1. **[SEW](https://huggingface.co/docs/transformers/model_doc/sew)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
 1. **[SEW-D](https://huggingface.co/docs/transformers/model_doc/sew_d)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
+1. **[SigLIP](https://huggingface.co/docs/transformers/model_doc/siglip)** (from Google AI) released with the paper [Sigmoid Loss for Language Image Pre-Training](https://arxiv.org/abs/2303.15343) by Xiaohua Zhai, Basil Mustafa, Alexander Kolesnikov, Lucas Beyer.
 1. **[SpeechT5](https://huggingface.co/docs/transformers/model_doc/speecht5)** (from Microsoft Research) released with the paper [SpeechT5: Unified-Modal Encoder-Decoder Pre-Training for Spoken Language Processing](https://arxiv.org/abs/2110.07205) by Junyi Ao, Rui Wang, Long Zhou, Chengyi Wang, Shuo Ren, Yu Wu, Shujie Liu, Tom Ko, Qing Li, Yu Zhang, Zhihua Wei, Yao Qian, Jinyu Li, Furu Wei.
 1. **[SpeechToTextTransformer](https://huggingface.co/docs/transformers/model_doc/speech_to_text)** (from Facebook), released together with the paper [fairseq S2T: Fast Speech-to-Text Modeling with fairseq](https://arxiv.org/abs/2010.05171) by Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Dmytro Okhonko, Juan Pino.
 1. **[SpeechToTextTransformer2](https://huggingface.co/docs/transformers/model_doc/speech_to_text_2)** (from Facebook) released with the paper [Large-Scale Self- and Semi-Supervised Learning for Speech Translation](https://arxiv.org/abs/2104.06678) by Changhan Wang, Anne Wu, Juan Pino, Alexei Baevski, Michael Auli, Alexis Conneau.
@@ -460,7 +463,7 @@ conda install -c huggingface transformers
 1. **[UMT5](https://huggingface.co/docs/transformers/model_doc/umt5)** (from Google Research) released with the paper [UniMax: Fairer and More Effective Language Sampling for Large-Scale Multilingual Pretraining](https://openreview.net/forum?id=kXwdL1cWOAi) by Hyung Won Chung, Xavier Garcia, Adam Roberts, Yi Tay, Orhan Firat, Sharan Narang, Noah Constant.
 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. **[UnivNet](https://huggingface.co/docs/transformers/model_doc/univnet)** (from Kakao Corporation) released with the paper [UnivNet: A Neural Vocoder with Multi-Resolution Spectrogram Discriminators for High-Fidelity Waveform Generation](https://arxiv.org/abs/2106.07889) by Won Jang, Dan Lim, Jaesam Yoon, Bongwan Kim, and Juntae Kim. 
+1. **[UnivNet](https://huggingface.co/docs/transformers/model_doc/univnet)** (from Kakao Corporation) released with the paper [UnivNet: A Neural Vocoder with Multi-Resolution Spectrogram Discriminators for High-Fidelity Waveform Generation](https://arxiv.org/abs/2106.07889) by Won Jang, Dan Lim, Jaesam Yoon, Bongwan Kim, and Juntae Kim.
 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/pdf/2202.09741.pdf) 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.
@@ -476,6 +479,7 @@ conda install -c huggingface transformers
 1. **[VITS](https://huggingface.co/docs/transformers/model_doc/vits)** (from Kakao Enterprise) released with the paper [Conditional Variational Autoencoder with Adversarial Learning for End-to-End Text-to-Speech](https://arxiv.org/abs/2106.06103) by Jaehyeon Kim, Jungil Kong, Juhee Son.
 1. **[ViViT](https://huggingface.co/docs/transformers/model_doc/vivit)** (from Google Research) released with the paper [ViViT: A Video Vision Transformer](https://arxiv.org/abs/2103.15691) by Anurag Arnab, Mostafa Dehghani, Georg Heigold, Chen Sun, Mario Lučić, Cordelia Schmid.
 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.
+1. **[Wav2Vec2-BERT](https://huggingface.co/docs/transformers/model_doc/wav2vec2-bert)** (from Meta AI) released with the paper [Seamless: Multilingual Expressive and Streaming Speech Translation](https://ai.meta.com/research/publications/seamless-multilingual-expressive-and-streaming-speech-translation/) by the Seamless Communication team.
 1. **[Wav2Vec2-Conformer](https://huggingface.co/docs/transformers/model_doc/wav2vec2-conformer)** (from Facebook AI) released with the paper [FAIRSEQ S2T: Fast Speech-to-Text Modeling with FAIRSEQ](https://arxiv.org/abs/2010.05171) by Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Sravya Popuri, Dmytro Okhonko, Juan Pino.
 1. **[Wav2Vec2Phoneme](https://huggingface.co/docs/transformers/model_doc/wav2vec2_phoneme)** (from Facebook AI) released with the paper [Simple and Effective Zero-shot Cross-lingual Phoneme Recognition](https://arxiv.org/abs/2109.11680) by Qiantong Xu, Alexei Baevski, Michael Auli.
 1. **[WavLM](https://huggingface.co/docs/transformers/model_doc/wavlm)** (from Microsoft Research) released with the paper [WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing](https://arxiv.org/abs/2110.13900) by Sanyuan Chen, Chengyi Wang, Zhengyang Chen, Yu Wu, Shujie Liu, Zhuo Chen, Jinyu Li, Naoyuki Kanda, Takuya Yoshioka, Xiong Xiao, Jian Wu, Long Zhou, Shuo Ren, Yanmin Qian, Yao Qian, Jian Wu, Michael Zeng, Furu Wei.

conftest.py

@@ -26,7 +26,7 @@ from transformers.testing_utils import HfDoctestModule, HfDocTestParser
 
 
 # allow having multiple repository checkouts and not needing to remember to rerun
-# 'pip install -e .[dev]' when switching between checkouts and running tests.
+# `pip install -e '.[dev]'` when switching between checkouts and running tests.
 git_repo_path = abspath(join(dirname(__file__), "src"))
 sys.path.insert(1, git_repo_path)
 

docker/transformers-all-latest-gpu/Dockerfile

@@ -9,9 +9,9 @@ SHELL ["sh", "-lc"]
 # The following `ARG` are mainly used to specify the versions explicitly & directly in this docker file, and not meant
 # to be used as arguments for docker build (so far).
 
-ARG PYTORCH='2.1.0'
+ARG PYTORCH='2.1.1'
 # (not always a valid torch version)
-ARG INTEL_TORCH_EXT='2.1.0'
+ARG INTEL_TORCH_EXT='2.1.100'
 # Example: `cu102`, `cu113`, etc.
 ARG CUDA='cu118'
 
@@ -56,16 +56,16 @@ RUN python3 -m pip install --no-cache-dir auto-gptq --extra-index-url https://hu
 RUN python3 -m pip install --no-cache-dir einops
 
 # Add autoawq for quantization testing
-RUN python3 -m pip install --no-cache-dir https://github.com/casper-hansen/AutoAWQ/releases/download/v0.1.7/autoawq-0.1.7+cu118-cp38-cp38-linux_x86_64.whl
+RUN python3 -m pip install --no-cache-dir https://github.com/casper-hansen/AutoAWQ/releases/download/v0.1.8/autoawq-0.1.8+cu118-cp38-cp38-linux_x86_64.whl
 
-# For bettertransformer + gptq 
+# For bettertransformer + gptq
 RUN python3 -m pip install --no-cache-dir git+https://github.com/huggingface/optimum@main#egg=optimum
 
 # For video model testing
 RUN python3 -m pip install --no-cache-dir decord av==9.2.0
 
 # For `dinat` model
-RUN python3 -m pip install --no-cache-dir natten -f https://shi-labs.com/natten/wheels/$CUDA/
+RUN python3 -m pip install --no-cache-dir 'natten<0.15.0' -f https://shi-labs.com/natten/wheels/$CUDA/
 
 # For `nougat` tokenizer
 RUN python3 -m pip install --no-cache-dir python-Levenshtein

docker/transformers-cpu/Dockerfile

@@ -1,26 +0,0 @@
-FROM ubuntu:18.04
-LABEL maintainer="Hugging Face"
-LABEL repository="transformers"
-
-RUN apt update && \
-    apt install -y bash \
-                   build-essential \
-                   git \
-                   curl \
-                   ca-certificates \
-                   python3 \
-                   python3-pip && \
-    rm -rf /var/lib/apt/lists
-
-RUN python3 -m pip install --no-cache-dir --upgrade pip && \
-    python3 -m pip install --no-cache-dir \
-    jupyter \
-    tensorflow-cpu \
-    torch
-
-WORKDIR /workspace
-COPY . transformers/
-RUN cd transformers/ && \
-    python3 -m pip install --no-cache-dir .
-
-CMD ["/bin/bash"]

docker/transformers-doc-builder/Dockerfile

@@ -1,4 +1,4 @@
-FROM python:3.8
+FROM python:3.10
 LABEL maintainer="Hugging Face"
 
 RUN apt update

docker/transformers-pytorch-cpu/Dockerfile

@@ -1,25 +0,0 @@
-FROM ubuntu:18.04
-LABEL maintainer="Hugging Face"
-LABEL repository="transformers"
-
-RUN apt update && \
-    apt install -y bash \
-                   build-essential \
-                   git \
-                   curl \
-                   ca-certificates \
-                   python3 \
-                   python3-pip && \
-    rm -rf /var/lib/apt/lists
-
-RUN python3 -m pip install --no-cache-dir --upgrade pip && \
-    python3 -m pip install --no-cache-dir \
-    jupyter \
-    torch
-
-WORKDIR /workspace
-COPY . transformers/
-RUN cd transformers/ && \
-    python3 -m pip install --no-cache-dir .
-
-CMD ["/bin/bash"]

docker/transformers-pytorch-gpu/Dockerfile

@@ -11,7 +11,7 @@ ARG REF=main
 RUN git clone https://github.com/huggingface/transformers && cd transformers && git checkout $REF
 
 # If set to nothing, will install the latest version
-ARG PYTORCH='2.1.0'
+ARG PYTORCH='2.1.1'
 ARG TORCH_VISION=''
 ARG TORCH_AUDIO=''
 # Example: `cu102`, `cu113`, etc.

docker/transformers-tensorflow-cpu/Dockerfile

@@ -1,25 +0,0 @@
-FROM ubuntu:18.04
-LABEL maintainer="Hugging Face"
-LABEL repository="transformers"
-
-RUN apt update && \
-    apt install -y bash \
-                   build-essential \
-                   git \
-                   curl \
-                   ca-certificates \
-                   python3 \
-                   python3-pip && \
-    rm -rf /var/lib/apt/lists
-
-RUN python3 -m pip install --no-cache-dir --upgrade pip && \
-    python3 -m pip install --no-cache-dir \
-    mkl \
-    tensorflow-cpu
-
-WORKDIR /workspace
-COPY . transformers/
-RUN cd transformers/ && \
-    python3 -m pip install --no-cache-dir .
-
-CMD ["/bin/bash"]

docs/README.md

@@ -250,7 +250,7 @@ then its documentation should look like this:
 
 Note that we always omit the "defaults to \`None\`" when None is the default for any argument. Also note that even
 if the first line describing your argument type and its default gets long, you can't break it on several lines. You can
-however write as many lines as you want in the indented description (see the example above with `input_ids`).
+however, write as many lines as you want in the indented description (see the example above with `input_ids`).
 
 #### Writing a multi-line code block
 

docs/source/de/add_new_pipeline.md

@@ -15,7 +15,7 @@ rendered properly in your Markdown viewer.
 
 # Wie erstellt man eine benutzerdefinierte Pipeline?
 
-In dieser Anleitung sehen wir uns an, wie Sie eine benutzerdefinierte Pipeline erstellen und sie auf dem [Hub](hf.co/models) freigeben oder sie der
+In dieser Anleitung sehen wir uns an, wie Sie eine benutzerdefinierte Pipeline erstellen und sie auf dem [Hub](https://hf.co/models) freigeben oder sie der
 🤗 Transformers-Bibliothek hinzufügen.
 
 Zuallererst müssen Sie entscheiden, welche Roheingaben die Pipeline verarbeiten kann. Es kann sich um Strings, rohe Bytes,

docs/source/de/installation.md

@@ -139,10 +139,10 @@ Ihre Python-Umgebung wird beim nächsten Ausführen die `main`-Version von 🤗
 
 ## Installation mit conda
 
-Installation von dem conda Kanal `huggingface`:
+Installation von dem conda Kanal `conda-forge`:
 
 ```bash
-conda install -c huggingface transformers
+conda install conda-forge::transformers
 ```
 
 ## Cache Einrichtung
@@ -157,7 +157,7 @@ Vorgefertigte Modelle werden heruntergeladen und lokal zwischengespeichert unter
 <Tip>
 
 Transformers verwendet die Shell-Umgebungsvariablen `PYTORCH_TRANSFORMERS_CACHE` oder `PYTORCH_PRETRAINED_BERT_CACHE`, wenn Sie von einer früheren Iteration dieser Bibliothek kommen und diese Umgebungsvariablen gesetzt haben, sofern Sie nicht die Shell-Umgebungsvariable `TRANSFORMERS_CACHE` angeben.
-  
+
 </Tip>
 
 ## Offline Modus
@@ -246,5 +246,5 @@ Sobald Ihre Datei heruntergeladen und lokal zwischengespeichert ist, geben Sie d
 <Tip>
 
 Weitere Informationen zum Herunterladen von Dateien, die auf dem Hub gespeichert sind, finden Sie im Abschnitt [Wie man Dateien vom Hub herunterlädt] (https://huggingface.co/docs/hub/how-to-downstream).
-  
+
 </Tip>

docs/source/en/_toctree.yml

@@ -111,6 +111,8 @@
     title: Share a custom model
   - local: chat_templating
     title: Templates for chat models
+  - local: trainer
+    title: Trainer
   - local: sagemaker
     title: Run training on Amazon SageMaker
   - local: serialization
@@ -140,16 +142,16 @@
       title: Methods and tools for efficient training on a single GPU
     - local: perf_train_gpu_many
       title: Multiple GPUs and parallelism
+    - local: fsdp
+      title: Fully Sharded Data Parallel
     - local: perf_train_cpu
       title: Efficient training on CPU
     - local: perf_train_cpu_many
       title: Distributed CPU training
-    - local: perf_train_tpu
-      title: Training on TPUs
     - local: perf_train_tpu_tf
       title: Training on TPU with TensorFlow
     - local: perf_train_special
-      title: Training on Specialized Hardware
+      title: PyTorch training on Apple silicon
     - local: perf_hardware
       title: Custom hardware for training
     - local: hpo_train
@@ -164,7 +166,7 @@
   - local: big_models
     title: Instantiating a big model
   - local: debugging
-    title: Troubleshooting
+    title: Debugging
   - local: tf_xla
     title: XLA Integration for TensorFlow Models
   - local: perf_torch_compile
@@ -330,6 +332,8 @@
         title: ESM
       - local: model_doc/falcon
         title: Falcon
+      - local: model_doc/fastspeech2_conformer
+        title: FastSpeech2Conformer
       - local: model_doc/flan-t5
         title: FLAN-T5
       - local: model_doc/flan-ul2
@@ -442,6 +446,8 @@
         title: ProphetNet
       - local: model_doc/qdqbert
         title: QDQBert
+      - local: model_doc/qwen2
+        title: Qwen2
       - local: model_doc/rag
         title: RAG
       - local: model_doc/realm
@@ -644,6 +650,8 @@
         title: VITS
       - local: model_doc/wav2vec2
         title: Wav2Vec2
+      - local: model_doc/wav2vec2-bert
+        title: Wav2Vec2-BERT
       - local: model_doc/wav2vec2-conformer
         title: Wav2Vec2-Conformer
       - local: model_doc/wav2vec2_phoneme
@@ -729,6 +737,8 @@
         title: Pix2Struct
       - local: model_doc/sam
         title: Segment Anything
+      - local: model_doc/siglip
+        title: SigLIP
       - local: model_doc/speech-encoder-decoder
         title: Speech Encoder Decoder Models
       - local: model_doc/tapas

docs/source/en/add_new_pipeline.md

@@ -15,7 +15,7 @@ rendered properly in your Markdown viewer.
 
 # How to create a custom pipeline?
 
-In this guide, we will see how to create a custom pipeline and share it on the [Hub](hf.co/models) or add it to the
+In this guide, we will see how to create a custom pipeline and share it on the [Hub](https://hf.co/models) or add it to the
 🤗 Transformers library.
 
 First and foremost, you need to decide the raw entries the pipeline will be able to take. It can be strings, raw bytes,

docs/source/en/debugging.md

@@ -16,6 +16,74 @@ rendered properly in your Markdown viewer.
 
 # Debugging
 
+Training on multiple GPUs can be a tricky endeavor whether you're running into installation issues or communication problems between your GPUs. This debugging guide covers some issues you may run into and how to resolve them.
+
+## DeepSpeed CUDA installation
+
+If you're using DeepSpeed, you've probably already installed it with the following command.
+
+```bash
+pip install deepspeed
+```
+
+DeepSpeed compiles CUDA C++ code and it can be a potential source of errors when building PyTorch extensions that require CUDA. These errors depend on how CUDA is installed on your system, and this section focuses on PyTorch built with *CUDA 10.2*.
+
+<Tip>
+
+For any other installation issues, please [open an issue](https://github.com/microsoft/DeepSpeed/issues) with the DeepSpeed team.
+
+</Tip>
+
+### Non-identical CUDA toolkits
+
+PyTorch comes with its own CUDA toolkit, but to use DeepSpeed with PyTorch, you need to have an identical version of CUDA installed system-wide. For example, if you installed PyTorch with `cudatoolkit==10.2` in your Python environment, then you'll also need to have CUDA 10.2 installed system-wide. If you don't have CUDA installed system-wide, you should install it first.
+
+The exact location may vary from system to system, but `usr/local/cuda-10.2` is the most common location on many Unix systems. When CUDA is correctly setup and added to your `PATH` environment variable, you can find the installation location with the following command:
+
+```bash
+which nvcc
+```
+
+### Multiple CUDA toolkits
+
+You may also have more than one CUDA toolkit installed system-wide.
+
+```bash
+/usr/local/cuda-10.2
+/usr/local/cuda-11.0
+```
+
+Typically, package installers set the paths to whatever the last version was installed. If the package build fails because it can't find the right CUDA version (despite it being installed system-wide already), then you need to configure the `PATH` and `LD_LIBRARY_PATH` environment variables to point to the correct path.
+
+Take a look at the contents of these environment variables first:
+
+```bash
+echo $PATH
+echo $LD_LIBRARY_PATH
+```
+
+`PATH` lists the locations of the executables and `LD_LIBRARY_PATH` lists where to look for shared libraries. Earlier entries are prioritized over later ones, and `:` is used to separate multiple entries. To tell the build program where to find the specific CUDA toolkit you want, insert the correct path to list first. This command prepends rather than overwrites the existing values.
+
+```bash
+# adjust the version and full path if needed
+export PATH=/usr/local/cuda-10.2/bin:$PATH
+export LD_LIBRARY_PATH=/usr/local/cuda-10.2/lib64:$LD_LIBRARY_PATH
+```
+
+In addition, you should also check the directories you assign actually exist. The `lib64` sub-directory contains various CUDA `.so` objects (like `libcudart.so`) and while it is unlikely your system names them differently, you should check the actual names and change them accordingly.
+
+### Older CUDA versions
+
+Sometimes, older CUDA versions may refuse to build with newer compilers. For example, if you have `gcc-9` but CUDA wants `gcc-7`. Usually, installing the latest CUDA toolkit enables support for the newer compiler.
+
+You could also install an older version of the compiler in addition to the one you're currently using (or it may already be installed but it's not used by default and the build system can't see it). To resolve this, you can create a symlink to give the build system visibility to the older compiler.
+
+```bash
+# adapt the path to your system
+sudo ln -s /usr/bin/gcc-7  /usr/local/cuda-10.2/bin/gcc
+sudo ln -s /usr/bin/g++-7  /usr/local/cuda-10.2/bin/g++
+```
+
 ## Multi-GPU Network Issues Debug
 
 When training or inferencing with `DistributedDataParallel` and multiple GPU, if you run into issue of inter-communication between processes and/or nodes, you can use the following script to diagnose network issues.

docs/source/en/fsdp.md

@@ -0,0 +1,138 @@
+<!--Copyright 2023 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+
+⚠️ Note that this file is in Markdown but contain specific syntax for our doc-builder (similar to MDX) that may not be
+rendered properly in your Markdown viewer.
+
+-->
+
+# Fully Sharded Data Parallel
+
+[Fully Sharded Data Parallel (FSDP)](https://pytorch.org/blog/introducing-pytorch-fully-sharded-data-parallel-api/) is a data parallel method that shards a model's parameters, gradients and optimizer states across the number of available GPUs (also called workers or *rank*). Unlike [DistributedDataParallel (DDP)](https://pytorch.org/docs/stable/generated/torch.nn.parallel.DistributedDataParallel.html), FSDP reduces memory-usage because a model is replicated on each GPU. This improves GPU memory-efficiency and allows you to train much larger models on fewer GPUs. FSDP is integrated with the Accelerate, a library for easily managing training in distributed environments, which means it is available for use from the [`Trainer`] class.
+
+Before you start, make sure Accelerate is installed and at least PyTorch 2.1.0 or newer.
+
+```bash
+pip install accelerate
+```
+
+## FSDP configuration
+
+To start, run the [`accelerate config`](https://huggingface.co/docs/accelerate/package_reference/cli#accelerate-config) command to create a configuration file for your training environment. Accelerate uses this configuration file to automatically setup the correct training environment based on your selected training options in `accelerate config`.
+
+```bash
+accelerate config
+```
+
+When you run `accelerate config`, you'll be prompted with a series of options to configure your training environment. This section covers some of the most important FSDP options. To learn more about the other available FSDP options, take a look at the [fsdp_config](https://huggingface.co/docs/transformers/main_classes/trainer#transformers.TrainingArguments.fsdp_config) parameters.
+
+### Sharding strategy
+
+FSDP offers a number of sharding strategies to select from:
+
+* `FULL_SHARD` - shards model parameters, gradients and optimizer states across workers; select `1` for this option
+* `SHARD_GRAD_OP`- shard gradients and optimizer states across workers; select `2` for this option
+* `NO_SHARD` - don't shard anything (this is equivalent to DDP); select `3` for this option
+* `HYBRID_SHARD` - shard model parameters, gradients and optimizer states within each worker where each worker also has a full copy; select `4` for this option
+* `HYBRID_SHARD_ZERO2` - shard gradients and optimizer states within each worker where each worker also has a full copy; select `5` for this option
+
+This is enabled by the `fsdp_sharding_strategy` flag.
+
+### CPU offload
+
+You could also offload parameters and gradients when they are not in use to the CPU to save even more GPU memory and help you fit large models where even FSDP may not be sufficient. This is enabled by setting `fsdp_offload_params: true` when running `accelerate config`.
+
+### Wrapping policy
+
+FSDP is applied by wrapping each layer in the network. The wrapping is usually applied in a nested way where the full weights are discarded after each forward pass to save memory for use in the next layer. The *auto wrapping* policy is the simplest way to implement this and you don't need to change any code. You should select `fsdp_auto_wrap_policy: TRANSFORMER_BASED_WRAP` to wrap a Transformer layer and `fsdp_transformer_layer_cls_to_wrap` to specify which layer to wrap (for example `BertLayer`).
+
+Otherwise, you can choose a size-based wrapping policy where FSDP is applied to a layer if it exceeds a certain number of parameters. This is enabled by setting `fsdp_wrap_policy: SIZE_BASED_WRAP` and `min_num_param` to the desired size threshold.
+
+### Checkpointing
+
+Intermediate checkpoints should be saved with `fsdp_state_dict_type: SHARDED_STATE_DICT` because saving the full state dict with CPU offloading on rank 0 takes a lot of time and often results in `NCCL Timeout` errors due to indefinite hanging during broadcasting. You can resume training with the sharded state dicts with the [`~accelerate.Accelerator.load_state`]` method.
+
+```py
+# directory containing checkpoints
+accelerator.load_state("ckpt")
+```
+
+However, when training ends, you want to save the full state dict because sharded state dict is only compatible with FSDP.
+
+```py
+if trainer.is_fsdp_enabled:
+    trainer.accelerator.state.fsdp_plugin.set_state_dict_type("FULL_STATE_DICT")
+
+trainer.save_model(script_args.output_dir)
+```
+
+### TPU
+
+[PyTorch XLA](https://pytorch.org/xla/release/2.1/index.html) supports FSDP training for TPUs and it can be enabled by modifying the FSDP configuration file generated by `accelerate config`. In addition to the sharding strategies and wrapping options specified above, you can add the parameters shown below to the file.
+
+```yaml
+xla: True # must be set to True to enable PyTorch/XLA
+xla_fsdp_settings: # XLA-specific FSDP parameters
+xla_fsdp_grad_ckpt: True # use gradient checkpointing
+```
+
+The [`xla_fsdp_settings`](https://github.com/pytorch/xla/blob/2e6e183e0724818f137c8135b34ef273dea33318/torch_xla/distributed/fsdp/xla_fully_sharded_data_parallel.py#L128) allow you to configure additional XLA-specific parameters for FSDP.
+
+## Launch training
+
+An example FSDP configuration file may look like:
+
+```yaml
+compute_environment: LOCAL_MACHINE
+debug: false
+distributed_type: FSDP
+downcast_bf16: 'no'
+fsdp_config:
+  fsdp_auto_wrap_policy: TRANSFORMER_BASED_WRAP
+  fsdp_backward_prefetch_policy: BACKWARD_PRE
+  fsdp_cpu_ram_efficient_loading: true
+  fsdp_forward_prefetch: false
+  fsdp_offload_params: true
+  fsdp_sharding_strategy: 1
+  fsdp_state_dict_type: SHARDED_STATE_DICT
+  fsdp_sync_module_states: true
+  fsdp_transformer_layer_cls_to_wrap: BertLayer
+  fsdp_use_orig_params: true
+machine_rank: 0
+main_training_function: main
+mixed_precision: bf16
+num_machines: 1
+num_processes: 2
+rdzv_backend: static
+same_network: true
+tpu_env: []
+tpu_use_cluster: false
+tpu_use_sudo: false
+use_cpu: false
+```
+
+To launch training, run the [`accelerate launch`](https://huggingface.co/docs/accelerate/package_reference/cli#accelerate-launch) command and it'll automatically use the configuration file you previously created with `accelerate config`.
+
+```bash
+accelerate launch my-trainer-script.py
+```
+
+```bash
+accelerate launch --fsdp="full shard" --fsdp_config="path/to/fsdp_config/ my-trainer-script.py
+```
+
+## Next steps
+
+FSDP can be a powerful tool for training really large models and you have access to more than one GPU or TPU. By sharding the model parameters, optimizer and gradient states, and even offloading them to the CPU when they're inactive, FSDP can reduce the high cost of large-scale training. If you're interested in learning more, the following may be helpful:
+
+* Follow along with the more in-depth Accelerate guide for [FSDP](https://huggingface.co/docs/accelerate/usage_guides/fsdp).
+* Read the [Introducing PyTorch Fully Sharded Data Parallel (FSDP) API](https://pytorch.org/blog/introducing-pytorch-fully-sharded-data-parallel-api/) blog post.
+* Read the [Scaling PyTorch models on Cloud TPUs with FSDP](https://pytorch.org/blog/scaling-pytorch-models-on-cloud-tpus-with-fsdp/) blog post.

docs/source/en/generation_strategies.md

@@ -82,7 +82,7 @@ Even if the default decoding strategy mostly works for your task, you can still
 commonly adjusted parameters include:
 
 - `max_new_tokens`: the maximum number of tokens to generate. In other words, the size of the output sequence, not
-including the tokens in the prompt. As an alternative to using the output's length as a stopping criteria, you can choose 
+including the tokens in the prompt. As an alternative to using the output's length as a stopping criteria, you can choose
 to stop generation whenever the full generation exceeds some amount of time. To learn more, check [`StoppingCriteria`].
 - `num_beams`: by specifying a number of beams higher than 1, you are effectively switching from greedy search to
 beam search. This strategy evaluates several hypotheses at each time step and eventually chooses the hypothesis that
@@ -339,13 +339,16 @@ This guide illustrates the main parameters that enable various decoding strategi
 [`generate`] method, which gives you even further control over the [`generate`] method's behavior.
 For the complete list of the available parameters, refer to the [API documentation](./main_classes/text_generation.md).
 
-### Assisted Decoding
+### Speculative Decoding
 
-Assisted decoding is a modification of the decoding strategies above that uses an assistant model with the same
-tokenizer (ideally a much smaller model) to greedily generate a few candidate tokens. The main model then validates
-the candidate tokens in a single forward pass, which speeds up the decoding process. Currently, only greedy search
-and sampling are supported with assisted decoding, and doesn't support batched inputs. To learn more about assisted
-decoding, check [this blog post](https://huggingface.co/blog/assisted-generation).
+Speculative decoding (also known as assisted decoding) is a modification of the decoding strategies above, that uses an
+assistant model (ideally a much smaller one) with the same tokenizer, to generate a few candidate tokens. The main
+model then validates the candidate tokens in a single forward pass, which speeds up the decoding process. If
+`do_sample=True`, then the token validation with resampling introduced in the
+[speculative decoding paper](https://arxiv.org/pdf/2211.17192.pdf) is used.
+
+Currently, only greedy search and sampling are supported with assisted decoding, and assisted decoding doesn't support batched inputs.
+To learn more about assisted decoding, check [this blog post](https://huggingface.co/blog/assisted-generation).
 
 To enable assisted decoding, set the `assistant_model` argument with a model.
 
@@ -366,8 +369,8 @@ To enable assisted decoding, set the `assistant_model` argument with a model.
 ['Alice and Bob are sitting in a bar. Alice is drinking a beer and Bob is drinking a']
 ```
 
-When using assisted decoding with sampling methods, you can use the `temperature` argument to control the randomness
-just like in multinomial sampling. However, in assisted decoding, reducing the temperature will help improving latency.
+When using assisted decoding with sampling methods, you can use the `temperature` argument to control the randomness,
+just like in multinomial sampling. However, in assisted decoding, reducing the temperature may help improve the latency.
 
 ```python
 >>> from transformers import AutoModelForCausalLM, AutoTokenizer, set_seed

docs/source/en/index.md

@@ -132,6 +132,7 @@ Flax), PyTorch, and/or TensorFlow.
 |                           [ESM](model_doc/esm)                           |       ✅        |         ✅         |      ❌      |
 |              [FairSeq Machine-Translation](model_doc/fsmt)               |       ✅        |         ❌         |      ❌      |
 |                        [Falcon](model_doc/falcon)                        |       ✅        |         ❌         |      ❌      |
+|         [FastSpeech2Conformer](model_doc/fastspeech2_conformer)          |       ✅        |         ❌         |      ❌      |
 |                       [FLAN-T5](model_doc/flan-t5)                       |       ✅        |         ✅         |      ✅      |
 |                      [FLAN-UL2](model_doc/flan-ul2)                      |       ✅        |         ✅         |      ✅      |
 |                      [FlauBERT](model_doc/flaubert)                      |       ✅        |         ✅         |      ❌      |
@@ -231,6 +232,7 @@ Flax), PyTorch, and/or TensorFlow.
 |                    [ProphetNet](model_doc/prophetnet)                    |       ✅        |         ❌         |      ❌      |
 |                           [PVT](model_doc/pvt)                           |       ✅        |         ❌         |      ❌      |
 |                       [QDQBert](model_doc/qdqbert)                       |       ✅        |         ❌         |      ❌      |
+|                         [Qwen2](model_doc/qwen2)                         |       ✅        |         ❌         |      ❌      |
 |                           [RAG](model_doc/rag)                           |       ✅        |         ✅         |      ❌      |
 |                         [REALM](model_doc/realm)                         |       ✅        |         ❌         |      ❌      |
 |                      [Reformer](model_doc/reformer)                      |       ✅        |         ❌         |      ❌      |
@@ -249,6 +251,7 @@ Flax), PyTorch, and/or TensorFlow.
 |                     [SegFormer](model_doc/segformer)                     |       ✅        |         ✅         |      ❌      |
 |                           [SEW](model_doc/sew)                           |       ✅        |         ❌         |      ❌      |
 |                         [SEW-D](model_doc/sew-d)                         |       ✅        |         ❌         |      ❌      |
+|                        [SigLIP](model_doc/siglip)                        |       ✅        |         ❌         |      ❌      |
 |        [Speech Encoder decoder](model_doc/speech-encoder-decoder)        |       ✅        |         ❌         |      ✅      |
 |                 [Speech2Text](model_doc/speech_to_text)                  |       ✅        |         ✅         |      ❌      |
 |                      [SpeechT5](model_doc/speecht5)                      |       ✅        |         ❌         |      ❌      |
@@ -293,6 +296,7 @@ Flax), PyTorch, and/or TensorFlow.
 |                          [VITS](model_doc/vits)                          |       ✅        |         ❌         |      ❌      |
 |                         [ViViT](model_doc/vivit)                         |       ✅        |         ❌         |      ❌      |
 |                      [Wav2Vec2](model_doc/wav2vec2)                      |       ✅        |         ✅         |      ✅      |
+|                 [Wav2Vec2-BERT](model_doc/wav2vec2-bert)                 |       ✅        |         ❌         |      ❌      |
 |            [Wav2Vec2-Conformer](model_doc/wav2vec2-conformer)            |       ✅        |         ❌         |      ❌      |
 |              [Wav2Vec2Phoneme](model_doc/wav2vec2_phoneme)               |       ✅        |         ✅         |      ✅      |
 |                         [WavLM](model_doc/wavlm)                         |       ✅        |         ❌         |      ❌      |

docs/source/en/installation.md

@@ -70,7 +70,7 @@ pip install 'transformers[tf-cpu]'
 <Tip warning={true}>
 
 M1 / ARM Users
-    
+
 You will need to install the following before installing TensorFLow 2.0
 ```
 brew install cmake
@@ -147,10 +147,10 @@ Your Python environment will find the `main` version of 🤗 Transformers on the
 
 ## Install with conda
 
-Install from the conda channel `huggingface`:
+Install from the conda channel `conda-forge`:
 
 ```bash
-conda install -c huggingface transformers
+conda install conda-forge::transformers
 ```
 
 ## Cache setup

docs/source/en/internal/generation_utils.md

@@ -45,7 +45,7 @@ inputs = tokenizer("Hello, my dog is cute and ", return_tensors="pt")
 generation_output = model.generate(**inputs, return_dict_in_generate=True, output_scores=True)
 ```
 
-The `generation_output` object is a [`~generation.GreedySearchDecoderOnlyOutput`], as we can
+The `generation_output` object is a [`~generation.GenerateDecoderOnlyOutput`], as we can
 see in the documentation of that class below, it means it has the following attributes:
 
 - `sequences`: the generated sequences of tokens
@@ -77,25 +77,13 @@ We document here all output types.
 
 ### PyTorch
 
-[[autodoc]] generation.GreedySearchEncoderDecoderOutput
+[[autodoc]] generation.GenerateDecoderOnlyOutput
 
-[[autodoc]] generation.GreedySearchDecoderOnlyOutput
+[[autodoc]] generation.GenerateEncoderDecoderOutput
 
-[[autodoc]] generation.SampleEncoderDecoderOutput
+[[autodoc]] generation.GenerateBeamDecoderOnlyOutput
 
-[[autodoc]] generation.SampleDecoderOnlyOutput
-
-[[autodoc]] generation.BeamSearchEncoderDecoderOutput
-
-[[autodoc]] generation.BeamSearchDecoderOnlyOutput
-
-[[autodoc]] generation.BeamSampleEncoderDecoderOutput
-
-[[autodoc]] generation.BeamSampleDecoderOnlyOutput
-
-[[autodoc]] generation.ContrastiveSearchEncoderDecoderOutput
-
-[[autodoc]] generation.ContrastiveSearchDecoderOnlyOutput
+[[autodoc]] generation.GenerateBeamEncoderDecoderOutput
 
 ### TensorFlow
 
@@ -317,7 +305,7 @@ generation.
 
 ## StoppingCriteria
 
-A [`StoppingCriteria`] can be used to change when to stop generation (other than EOS token). Please note that this is exclusivelly available to our PyTorch implementations.
+A [`StoppingCriteria`] can be used to change when to stop generation (other than EOS token). Please note that this is exclusively available to our PyTorch implementations.
 
 [[autodoc]] StoppingCriteria
     - __call__
@@ -333,7 +321,7 @@ A [`StoppingCriteria`] can be used to change when to stop generation (other than
 
 ## Constraints
 
-A [`Constraint`] can be used to force the generation to include specific tokens or sequences in the output. Please note that this is exclusivelly available to our PyTorch implementations.
+A [`Constraint`] can be used to force the generation to include specific tokens or sequences in the output. Please note that this is exclusively available to our PyTorch implementations.
 
 [[autodoc]] Constraint
 

docs/source/en/main_classes/deepspeed.md

@@ -1221,12 +1221,7 @@ Therefore you have two ways to take advantage of this very beneficial feature:
 ### Optimizer and Scheduler
 
 As long as you don't enable `offload_optimizer` you can mix and match DeepSpeed and HuggingFace schedulers and
-optimizers, with the exception of using the combination of HuggingFace scheduler and DeepSpeed optimizer:
-
-| Combos       | HF Scheduler | DS Scheduler |
-|:-------------|:-------------|:-------------|
-| HF Optimizer | Yes          | Yes          |
-| DS Optimizer | No           | Yes          |
+optimizers.
 
 It is possible to use a non-DeepSpeed optimizer when `offload_optimizer` is enabled, as long as it has both CPU and
 GPU implementation (except LAMB).
@@ -2054,7 +2049,6 @@ In this case you usually need to raise the value of `initial_scale_power`. Setti
 
 ### Notes
 
-- DeepSpeed works with the PyTorch [`Trainer`] but not TF [`TFTrainer`].
 - While DeepSpeed has a pip installable PyPI package, it is highly recommended that it gets installed from [source](https://github.com/microsoft/deepspeed#installation) to best match your hardware and also if you need to enable
   certain features, like 1-bit Adam, which aren't available in the pypi distribution.
 - You don't have to use the [`Trainer`] to use DeepSpeed with 🤗 Transformers - you can use any model

docs/source/en/main_classes/pipelines.md

@@ -400,12 +400,6 @@ Pipelines available for natural language processing tasks include the following.
     - __call__
     - all
 
-### NerPipeline
-
-[[autodoc]] NerPipeline
-
-See [`TokenClassificationPipeline`] for all details.
-
 ### QuestionAnsweringPipeline
 
 [[autodoc]] QuestionAnsweringPipeline

docs/source/en/main_classes/trainer.md

@@ -16,70 +16,23 @@ rendered properly in your Markdown viewer.
 
 # Trainer
 
-The [`Trainer`] class provides an API for feature-complete training in PyTorch for most standard use cases. It's used in most of the [example scripts](https://github.com/huggingface/transformers/tree/main/examples).
+The [`Trainer`] class provides an API for feature-complete training in PyTorch, and it supports distributed training on multiple GPUs/TPUs, mixed precision for [NVIDIA GPUs](https://nvidia.github.io/apex/), [AMD GPUs](https://rocm.docs.amd.com/en/latest/rocm.html), and [`torch.amp`](https://pytorch.org/docs/stable/amp.html) for PyTorch. [`Trainer`] goes hand-in-hand with the [`TrainingArguments`] class, which offers a wide range of options to customize how a model is trained. Together, these two classes provide a complete training API.
 
-<Tip>
-
-If you're looking to fine-tune a language model like Llama-2 or Mistral on a text dataset using autoregressive techniques, consider using [`trl`](https://github.com/huggingface/trl)'s [`~trl.SFTTrainer`]. The [`~trl.SFTTrainer`] wraps the [`Trainer`] and is specially optimized for this particular task and supports sequence packing, LoRA, quantization, and DeepSpeed for efficient scaling to any model size. On the other hand, the [`Trainer`] is a more versatile option, suitable for a broader spectrum of tasks.
-
-</Tip>
-
-Before instantiating your [`Trainer`], create a [`TrainingArguments`] to access all the points of customization during training.
-
-The API supports distributed training on multiple GPUs/TPUs, mixed precision through [NVIDIA Apex] for NVIDIA GPUs, [ROCm APEX](https://github.com/ROCmSoftwarePlatform/apex) for AMD GPUs, and Native AMP for PyTorch.
-
-The [`Trainer`] contains the basic training loop which supports the above features. To inject custom behavior you can subclass them and override the following methods:
-
-- **get_train_dataloader** -- Creates the training DataLoader.
-- **get_eval_dataloader** -- Creates the evaluation DataLoader.
-- **get_test_dataloader** -- Creates the test DataLoader.
-- **log** -- Logs information on the various objects watching training.
-- **create_optimizer_and_scheduler** -- Sets up the optimizer and learning rate scheduler if they were not passed at
-  init. Note, that you can also subclass or override the `create_optimizer` and `create_scheduler` methods
-  separately.
-- **create_optimizer** -- Sets up the optimizer if it wasn't passed at init.
-- **create_scheduler** -- Sets up the learning rate scheduler if it wasn't passed at init.
-- **compute_loss** - Computes the loss on a batch of training inputs.
-- **training_step** -- Performs a training step.
-- **prediction_step** -- Performs an evaluation/test step.
-- **evaluate** -- Runs an evaluation loop and returns metrics.
-- **predict** -- Returns predictions (with metrics if labels are available) on a test set.
+[`Seq2SeqTrainer`] and [`Seq2SeqTrainingArguments`] inherit from the [`Trainer`] and [`TrainingArgument`] classes and they're adapted for training models for sequence-to-sequence tasks such as summarization or translation.
 
 <Tip warning={true}>
 
 The [`Trainer`] class is optimized for 🤗 Transformers models and can have surprising behaviors
-when you use it on other models. When using it on your own model, make sure:
+when used with other models. When using it with your own model, make sure:
 
-- your model always return tuples or subclasses of [`~utils.ModelOutput`].
+- your model always return tuples or subclasses of [`~utils.ModelOutput`]
 - your model can compute the loss if a `labels` argument is provided and that loss is returned as the first
   element of the tuple (if your model returns tuples)
-- your model can accept multiple label arguments (use the `label_names` in your [`TrainingArguments`] to indicate their name to the [`Trainer`]) but none of them should be named `"label"`.
+- your model can accept multiple label arguments (use `label_names` in [`TrainingArguments`] to indicate their name to the [`Trainer`]) but none of them should be named `"label"`
 
 </Tip>
 
-Here is an example of how to customize [`Trainer`] to use a weighted loss (useful when you have an unbalanced training set):
-
-```python
-from torch import nn
-from transformers import Trainer
-
-
-class CustomTrainer(Trainer):
-    def compute_loss(self, model, inputs, return_outputs=False):
-        labels = inputs.pop("labels")
-        # forward pass
-        outputs = model(**inputs)
-        logits = outputs.get("logits")
-        # compute custom loss (suppose one has 3 labels with different weights)
-        loss_fct = nn.CrossEntropyLoss(weight=torch.tensor([1.0, 2.0, 3.0], device=model.device))
-        loss = loss_fct(logits.view(-1, self.model.config.num_labels), labels.view(-1))
-        return (loss, outputs) if return_outputs else loss
-```
-
-Another way to customize the training loop behavior for the PyTorch [`Trainer`] is to use [callbacks](callback) that can inspect the training loop state (for progress reporting, logging on TensorBoard or other ML platforms...) and take decisions (like early stopping).
-
-
-## Trainer
+## Trainer[[api-reference]]
 
 [[autodoc]] Trainer
     - all
@@ -99,683 +52,3 @@ Another way to customize the training loop behavior for the PyTorch [`Trainer`]
 
 [[autodoc]] Seq2SeqTrainingArguments
     - all
-
-## Checkpoints
-
-By default, [`Trainer`] will save all checkpoints in the `output_dir` you set in the
-[`TrainingArguments`] you are using. Those will go in subfolder named `checkpoint-xxx` with xxx
-being the step at which the training was at.
-
-Resuming training from a checkpoint can be done when calling [`Trainer.train`] with either:
-
-- `resume_from_checkpoint=True` which will resume training from the latest checkpoint
-- `resume_from_checkpoint=checkpoint_dir` which will resume training from the specific checkpoint in the directory
-  passed.
-
-In addition, you can easily save your checkpoints on the Model Hub when using `push_to_hub=True`. By default, all
-the models saved in intermediate checkpoints are saved in different commits, but not the optimizer state. You can adapt
-the `hub-strategy` value of your [`TrainingArguments`] to either:
-
-- `"checkpoint"`: the latest checkpoint is also pushed in a subfolder named last-checkpoint, allowing you to
-  resume training easily with `trainer.train(resume_from_checkpoint="output_dir/last-checkpoint")`.
-- `"all_checkpoints"`: all checkpoints are pushed like they appear in the output folder (so you will get one
-  checkpoint folder per folder in your final repository)
-
-
-## Logging
-
-By default [`Trainer`] will use `logging.INFO` for the main process and `logging.WARNING` for the replicas if any.
-
-These defaults can be overridden to use any of the 5 `logging` levels with [`TrainingArguments`]'s
-arguments:
-
-- `log_level` - for the main process
-- `log_level_replica` - for the replicas
-
-Further, if [`TrainingArguments`]'s `log_on_each_node` is set to `False` only the main node will
-use the log level settings for its main process, all other nodes will use the log level settings for replicas.
-
-Note that [`Trainer`] is going to set `transformers`'s log level separately for each node in its
-[`Trainer.__init__`]. So you may want to set this sooner (see the next example) if you tap into other
-`transformers` functionality before creating the [`Trainer`] object.
-
-Here is an example of how this can be used in an application:
-
-```python
-[...]
-logger = logging.getLogger(__name__)
-
-# Setup logging
-logging.basicConfig(
-    format="%(asctime)s - %(levelname)s - %(name)s - %(message)s",
-    datefmt="%m/%d/%Y %H:%M:%S",
-    handlers=[logging.StreamHandler(sys.stdout)],
-)
-
-# set the main code and the modules it uses to the same log-level according to the node
-log_level = training_args.get_process_log_level()
-logger.setLevel(log_level)
-datasets.utils.logging.set_verbosity(log_level)
-transformers.utils.logging.set_verbosity(log_level)
-
-trainer = Trainer(...)
-```
-
-And then if you only want to see warnings on the main node and all other nodes to not print any most likely duplicated
-warnings you could run it as:
-
-```bash
-my_app.py ... --log_level warning --log_level_replica error
-```
-
-In the multi-node environment if you also don't want the logs to repeat for each node's main process, you will want to
-change the above to:
-
-```bash
-my_app.py ... --log_level warning --log_level_replica error --log_on_each_node 0
-```
-
-and then only the main process of the first node will log at the "warning" level, and all other processes on the main
-node and all processes on other nodes will log at the "error" level.
-
-If you need your application to be as quiet as possible you could do:
-
-```bash
-my_app.py ... --log_level error --log_level_replica error --log_on_each_node 0
-```
-
-(add `--log_on_each_node 0` if on multi-node environment)
-
-
-## Randomness
-
-When resuming from a checkpoint generated by [`Trainer`] all efforts are made to restore the
-_python_, _numpy_ and _pytorch_ RNG states to the same states as they were at the moment of saving that checkpoint,
-which should make the "stop and resume" style of training as close as possible to non-stop training.
-
-However, due to various default non-deterministic pytorch settings this might not fully work. If you want full
-determinism please refer to [Controlling sources of randomness](https://pytorch.org/docs/stable/notes/randomness). As explained in the document, that some of those settings
-that make things deterministic (.e.g., `torch.backends.cudnn.deterministic`) may slow things down, therefore this
-can't be done by default, but you can enable those yourself if needed.
-
-
-## Specific GPUs Selection
-
-Let's discuss how you can tell your program which GPUs are to be used and in what order.
-
-When using [`DistributedDataParallel`](https://pytorch.org/docs/stable/generated/torch.nn.parallel.DistributedDataParallel.html) to use only a subset of your GPUs, you simply specify the number of GPUs to use. For example, if you have 4 GPUs, but you wish to use the first 2 you can do:
-
-```bash
-torchrun --nproc_per_node=2  trainer-program.py ...
-```
-
-if you have either [`accelerate`](https://github.com/huggingface/accelerate) or [`deepspeed`](https://github.com/microsoft/DeepSpeed) installed you can also accomplish the same by using one of:
-
-```bash
-accelerate launch --num_processes 2 trainer-program.py ...
-```
-
-```bash
-deepspeed --num_gpus 2 trainer-program.py ...
-```
-
-You don't need to use the Accelerate or [the Deepspeed integration](deepspeed) features to use these launchers.
-
-
-Until now you were able to tell the program how many GPUs to use. Now let's discuss how to select specific GPUs and control their order.
-
-The following environment variables help you control which GPUs to use and their order.
-
-**`CUDA_VISIBLE_DEVICES`**
-
-If you have multiple GPUs and you'd like to use only 1 or a few of those GPUs, set the environment variable `CUDA_VISIBLE_DEVICES` to a list of the GPUs to be used.
-
-For example, let's say you have 4 GPUs: 0, 1, 2 and 3. To run only on the physical GPUs 0 and 2, you can do:
-
-```bash
-CUDA_VISIBLE_DEVICES=0,2 torchrun trainer-program.py ...
-```
-
-So now pytorch will see only 2 GPUs, where your physical GPUs 0 and 2 are mapped to `cuda:0` and `cuda:1` correspondingly.
-
-You can even change their order:
-
-```bash
-CUDA_VISIBLE_DEVICES=2,0 torchrun trainer-program.py ...
-```
-
-Here your physical GPUs 0 and 2 are mapped to `cuda:1` and `cuda:0` correspondingly.
-
-The above examples were all for `DistributedDataParallel` use pattern, but the same method works for [`DataParallel`](https://pytorch.org/docs/stable/generated/torch.nn.DataParallel.html) as well:
-
-```bash
-CUDA_VISIBLE_DEVICES=2,0 python trainer-program.py ...
-```
-
-To emulate an environment without GPUs simply set this environment variable to an empty value like so:
-
-```bash
-CUDA_VISIBLE_DEVICES= python trainer-program.py ...
-```
-
-As with any environment variable you can, of course, export those instead of adding these to the command line, as in:
-
-
-```bash
-export CUDA_VISIBLE_DEVICES=0,2
-torchrun trainer-program.py ...
-```
-
-but this approach can be confusing since you may forget you set up the environment variable earlier and not understand why the wrong GPUs are used. Therefore, it's a common practice to set the environment variable just for a specific run on the same command line as it's shown in most examples of this section.
-
-**`CUDA_DEVICE_ORDER`**
-
-There is an additional environment variable `CUDA_DEVICE_ORDER` that controls how the physical devices are ordered. The two choices are:
-
-1. ordered by PCIe bus IDs (matches `nvidia-smi` and `rocm-smi`'s order) - this is the default.
-
-```bash
-export CUDA_DEVICE_ORDER=PCI_BUS_ID
-```
-
-2. ordered by GPU compute capabilities
-
-```bash
-export CUDA_DEVICE_ORDER=FASTEST_FIRST
-```
-
-Most of the time you don't need to care about this environment variable, but it's very helpful if you have a lopsided setup where you have an old and a new GPUs physically inserted in such a way so that the slow older card appears to be first. One way to fix that is to swap the cards. But if you can't swap the cards (e.g., if the cooling of the devices gets impacted) then setting `CUDA_DEVICE_ORDER=FASTEST_FIRST` will always put the newer faster card first. It'll be somewhat confusing though since `nvidia-smi` (or `rocm-smi`) will still report them in the PCIe order.
-
-The other solution to swapping the order is to use:
-
-```bash
-export CUDA_VISIBLE_DEVICES=1,0
-```
-In this example we are working with just 2 GPUs, but of course the same would apply to as many GPUs as your computer has.
-
-Also if you do set this environment variable it's the best to set it in your `~/.bashrc` file or some other startup config file and forget about it.
-
-
-
-
-## Trainer Integrations
-
-The [`Trainer`] has been extended to support libraries that may dramatically improve your training
-time and fit much bigger models.
-
-Currently it supports third party solutions, [DeepSpeed](https://github.com/microsoft/DeepSpeed) and [PyTorch FSDP](https://pytorch.org/docs/stable/fsdp.html), which implement parts of the paper [ZeRO: Memory Optimizations
-Toward Training Trillion Parameter Models, by Samyam Rajbhandari, Jeff Rasley, Olatunji Ruwase, Yuxiong He](https://arxiv.org/abs/1910.02054).
-
-This provided support is new and experimental as of this writing. While the support for DeepSpeed and PyTorch FSDP is active and we welcome issues around it, we don't support the FairScale integration anymore since it has been integrated in PyTorch main (see the [PyTorch FSDP integration](#pytorch-fully-sharded-data-parallel))
-
-<a id='zero-install-notes'></a>
-
-### CUDA Extension Installation Notes
-
-As of this writing, Deepspeed require compilation of CUDA C++ code, before it can be used.
-
-While all installation issues should be dealt with through the corresponding GitHub Issues of [Deepspeed](https://github.com/microsoft/DeepSpeed/issues), there are a few common issues that one may encounter while building
-any PyTorch extension that needs to build CUDA extensions.
-
-Therefore, if you encounter a CUDA-related build issue while doing the following:
-
-```bash
-pip install deepspeed
-```
-
-please, read the following notes first.
-
-In these notes we give examples for what to do when `pytorch` has been built with CUDA `10.2`. If your situation is
-different remember to adjust the version number to the one you are after.
-
-#### Possible problem #1
-
-While, Pytorch comes with its own CUDA toolkit, to build these two projects you must have an identical version of CUDA
-installed system-wide.
-
-For example, if you installed `pytorch` with `cudatoolkit==10.2` in the Python environment, you also need to have
-CUDA `10.2` installed system-wide.
-
-The exact location may vary from system to system, but `/usr/local/cuda-10.2` is the most common location on many
-Unix systems. When CUDA is correctly set up and added to the `PATH` environment variable, one can find the
-installation location by doing:
-
-```bash
-which nvcc
-```
-
-If you don't have CUDA installed system-wide, install it first. You will find the instructions by using your favorite
-search engine. For example, if you're on Ubuntu you may want to search for: [ubuntu cuda 10.2 install](https://www.google.com/search?q=ubuntu+cuda+10.2+install).
-
-#### Possible problem #2
-
-Another possible common problem is that you may have more than one CUDA toolkit installed system-wide. For example you
-may have:
-
-```bash
-/usr/local/cuda-10.2
-/usr/local/cuda-11.0
-```
-
-Now, in this situation you need to make sure that your `PATH` and `LD_LIBRARY_PATH` environment variables contain
-the correct paths to the desired CUDA version. Typically, package installers will set these to contain whatever the
-last version was installed. If you encounter the problem, where the package build fails because it can't find the right
-CUDA version despite you having it installed system-wide, it means that you need to adjust the 2 aforementioned
-environment variables.
-
-First, you may look at their contents:
-
-```bash
-echo $PATH
-echo $LD_LIBRARY_PATH
-```
-
-so you get an idea of what is inside.
-
-It's possible that `LD_LIBRARY_PATH` is empty.
-
-`PATH` lists the locations of where executables can be found and `LD_LIBRARY_PATH` is for where shared libraries
-are to looked for. In both cases, earlier entries have priority over the later ones. `:` is used to separate multiple
-entries.
-
-Now, to tell the build program where to find the specific CUDA toolkit, insert the desired paths to be listed first by
-doing:
-
-```bash
-export PATH=/usr/local/cuda-10.2/bin:$PATH
-export LD_LIBRARY_PATH=/usr/local/cuda-10.2/lib64:$LD_LIBRARY_PATH
-```
-
-Note that we aren't overwriting the existing values, but prepending instead.
-
-Of course, adjust the version number, the full path if need be. Check that the directories you assign actually do
-exist. `lib64` sub-directory is where the various CUDA `.so` objects, like `libcudart.so` reside, it's unlikely
-that your system will have it named differently, but if it is adjust it to reflect your reality.
-
-
-#### Possible problem #3
-
-Some older CUDA versions may refuse to build with newer compilers. For example, you my have `gcc-9` but it wants
-`gcc-7`.
-
-There are various ways to go about it.
-
-If you can install the latest CUDA toolkit it typically should support the newer compiler.
-
-Alternatively, you could install the lower version of the compiler in addition to the one you already have, or you may
-already have it but it's not the default one, so the build system can't see it. If you have `gcc-7` installed but the
-build system complains it can't find it, the following might do the trick:
-
-```bash
-sudo ln -s /usr/bin/gcc-7  /usr/local/cuda-10.2/bin/gcc
-sudo ln -s /usr/bin/g++-7  /usr/local/cuda-10.2/bin/g++
-```
-
-Here, we are making a symlink to `gcc-7` from `/usr/local/cuda-10.2/bin/gcc` and since
-`/usr/local/cuda-10.2/bin/` should be in the `PATH` environment variable (see the previous problem's solution), it
-should find `gcc-7` (and `g++7`) and then the build will succeed.
-
-As always make sure to edit the paths in the example to match your situation.
-
-
-### PyTorch Fully Sharded Data parallel
-
-To accelerate training huge models on larger batch sizes, we can use a fully sharded data parallel model.
-This type of data parallel paradigm enables fitting more data and larger models by sharding the optimizer states, gradients and parameters.
-To read more about it and the benefits, check out the [Fully Sharded Data Parallel blog](https://pytorch.org/blog/introducing-pytorch-fully-sharded-data-parallel-api/).
-We have integrated the latest PyTorch's Fully Sharded Data Parallel (FSDP) training feature.
-All you need to do is enable it through the config.
-
-**Required PyTorch version for FSDP support**: PyTorch >=2.1.0
-
-**Usage**:
-
-- Make sure you have added the distributed launcher
-`-m torch.distributed.launch --nproc_per_node=NUMBER_OF_GPUS_YOU_HAVE` if you haven't been using it already.
-
-- **Sharding Strategy**: 
-  - FULL_SHARD : Shards optimizer states + gradients + model parameters across data parallel workers/GPUs.
-    For this, add `--fsdp full_shard` to the command line arguments. 
-  - SHARD_GRAD_OP : Shards optimizer states + gradients across data parallel workers/GPUs.
-    For this, add `--fsdp shard_grad_op` to the command line arguments.
-  - NO_SHARD : No sharding. For this, add `--fsdp no_shard` to the command line arguments.
-  - HYBRID_SHARD : No sharding. For this, add `--fsdp hybrid_shard` to the command line arguments.
-  - HYBRID_SHARD_ZERO2 : No sharding. For this, add `--fsdp hybrid_shard_zero2` to the command line arguments.
-- To offload the parameters and gradients to the CPU, 
-  add `--fsdp "full_shard offload"` or `--fsdp "shard_grad_op offload"` to the command line arguments.
-- To automatically recursively wrap layers with FSDP using `default_auto_wrap_policy`, 
-  add `--fsdp "full_shard auto_wrap"` or `--fsdp "shard_grad_op auto_wrap"` to the command line arguments.
-- To enable both CPU offloading and auto wrapping, 
-  add `--fsdp "full_shard offload auto_wrap"` or `--fsdp "shard_grad_op offload auto_wrap"` to the command line arguments.
-- Remaining FSDP config is passed via `--fsdp_config <path_to_fsdp_config.json>`. It is either a location of
-  FSDP json config file (e.g., `fsdp_config.json`) or an already loaded json file as `dict`. 
-  - If auto wrapping is enabled, you can either use transformer based auto wrap policy or size based auto wrap policy.
-    - For transformer based auto wrap policy, it is recommended to specify `transformer_layer_cls_to_wrap` in the config file. If not specified, the default value is `model._no_split_modules` when available.
-      This specifies the list of transformer layer class name (case-sensitive) to wrap ,e.g, [`BertLayer`], [`GPTJBlock`], [`T5Block`] ....
-      This is important because submodules that share weights (e.g., embedding layer) should not end up in different FSDP wrapped units.
-      Using this policy, wrapping happens for each block containing Multi-Head Attention followed by couple of MLP layers. 
-      Remaining layers including the shared embeddings are conveniently wrapped in same outermost FSDP unit.
-      Therefore, use this for transformer based models.
-    - For size based auto wrap policy, please add `min_num_params` in the config file. 
-      It specifies FSDP's minimum number of parameters for auto wrapping.
-  - `backward_prefetch` can be specified in the config file. It controls when to prefetch next set of parameters. 
-    `backward_pre` and `backward_pos` are available options. 
-    For more information refer `torch.distributed.fsdp.fully_sharded_data_parallel.BackwardPrefetch`
-  - `forward_prefetch` can be specified in the config file. It controls when to prefetch next set of parameters. 
-    If `"True"`, FSDP explicitly prefetches the next upcoming all-gather while executing in the forward pass. 
-  - `limit_all_gathers` can be specified in the config file. 
-    If `"True"`, FSDP explicitly synchronizes the CPU thread to prevent too many in-flight all-gathers.
-  - `activation_checkpointing` can be specified in the config file.
-    If `"True"`, FSDP activation checkpointing is a technique to reduce memory usage by clearing activations of
-    certain layers and recomputing them during a backward pass. Effectively, this trades extra computation time
-    for reduced memory usage.
-  - `use_orig_params` can be specified in the config file. 
-    If True, allows non-uniform `requires_grad` during init, which means support for interspersed frozen and trainable paramteres. Useful in cases such as parameter-efficient fine-tuning. This also enables to have different optimizer param groups. This should be `True` when creating optimizer object before preparing/wrapping the model with FSDP.
-    Please refer this [blog](https://dev-discuss.pytorch.org/t/rethinking-pytorch-fully-sharded-data-parallel-fsdp-from-first-principles/1019). 
-
-**Saving and loading**
-Saving entire intermediate checkpoints using `FULL_STATE_DICT` state_dict_type with CPU offloading on rank 0 takes a lot of time and often results in NCCL Timeout errors due to indefinite hanging during broadcasting. However, at the end of training, we want the whole model state dict instead of the sharded state dict which is only compatible with FSDP. Use `SHARDED_STATE_DICT` (default) state_dict_type to save the intermediate checkpoints and optimizer states in this format recommended by the PyTorch team. 
-
-Saving the final checkpoint in transformers format using default `safetensors` format requires below changes.
-```python
-if trainer.is_fsdp_enabled:
-    trainer.accelerator.state.fsdp_plugin.set_state_dict_type("FULL_STATE_DICT")
-
-trainer.save_model(script_args.output_dir)
-```
-
-**Few caveats to be aware of**
-- it is incompatible with `generate`, thus is incompatible with `--predict_with_generate` 
-  in all seq2seq/clm scripts (translation/summarization/clm etc.).  
-  Please refer issue [#21667](https://github.com/huggingface/transformers/issues/21667)
-
-### PyTorch/XLA Fully Sharded Data parallel
-
-For all the TPU users, great news! PyTorch/XLA now supports FSDP.
-All the latest Fully Sharded Data Parallel (FSDP) training are supported.
-For more information refer to the [Scaling PyTorch models on Cloud TPUs with FSDP](https://pytorch.org/blog/scaling-pytorch-models-on-cloud-tpus-with-fsdp/) and [PyTorch/XLA implementation of FSDP](https://github.com/pytorch/xla/tree/master/torch_xla/distributed/fsdp)
-All you need to do is enable it through the config.
-
-**Required PyTorch/XLA version for FSDP support**: >=2.0
-
-**Usage**:
-
-Pass `--fsdp "full shard"` along with following changes to be made in `--fsdp_config <path_to_fsdp_config.json>`:
-- `xla` should be set to `True` to enable PyTorch/XLA FSDP.
-- `xla_fsdp_settings` The value is a dictionary which stores the XLA FSDP wrapping parameters.
-  For a complete list of options, please see [here](
-  https://github.com/pytorch/xla/blob/master/torch_xla/distributed/fsdp/xla_fully_sharded_data_parallel.py).
-- `xla_fsdp_grad_ckpt`. When `True`, uses gradient checkpointing over each nested XLA FSDP wrapped layer. 
-  This setting can only be used when the xla flag is set to true, and an auto wrapping policy is specified through
-  `min_num_params` or `transformer_layer_cls_to_wrap`. 
-- You can either use transformer based auto wrap policy or size based auto wrap policy.
-  - For transformer based auto wrap policy, it is recommended to specify `transformer_layer_cls_to_wrap` in the config file. If not specified, the default value is `model._no_split_modules` when available.
-    This specifies the list of transformer layer class name (case-sensitive) to wrap ,e.g, [`BertLayer`], [`GPTJBlock`], [`T5Block`] ....
-    This is important because submodules that share weights (e.g., embedding layer) should not end up in different FSDP wrapped units.
-    Using this policy, wrapping happens for each block containing Multi-Head Attention followed by couple of MLP layers. 
-    Remaining layers including the shared embeddings are conveniently wrapped in same outermost FSDP unit.
-    Therefore, use this for transformer based models.
-  - For size based auto wrap policy, please add `min_num_params` in the config file. 
-    It specifies FSDP's minimum number of parameters for auto wrapping.
-
-
-### Using Trainer for accelerated PyTorch Training on Mac 
-
-With PyTorch v1.12 release, developers and researchers can take advantage of Apple silicon GPUs for significantly faster model training. 
-This unlocks the ability to perform machine learning workflows like prototyping and fine-tuning locally, right on Mac.
-Apple's Metal Performance Shaders (MPS) as a backend for PyTorch enables this and can be used via the new `"mps"` device. 
-This will map computational graphs and primitives on the MPS Graph framework and tuned kernels provided by MPS.
-For more information please refer official documents [Introducing Accelerated PyTorch Training on Mac](https://pytorch.org/blog/introducing-accelerated-pytorch-training-on-mac/)
-and [MPS BACKEND](https://pytorch.org/docs/stable/notes/mps.html). 
-
-<Tip warning={false}>
-
-We strongly recommend to install PyTorch >= 1.13 (nightly version at the time of writing) on your MacOS machine. 
-It has major fixes related to model correctness and performance improvements for transformer based models.
-Please refer to https://github.com/pytorch/pytorch/issues/82707 for more details.
-
-</Tip>
-
-**Benefits of Training and Inference using Apple Silicon Chips**
-
-1. Enables users to train larger networks or batch sizes locally
-2. Reduces data retrieval latency and provides the GPU with direct access to the full memory store due to unified memory architecture. 
-Therefore, improving end-to-end performance.
-3. Reduces costs associated with cloud-based development or the need for additional local GPUs.
-
-**Pre-requisites**: To install torch with mps support, 
-please follow this nice medium article [GPU-Acceleration Comes to PyTorch on M1 Macs](https://medium.com/towards-data-science/gpu-acceleration-comes-to-pytorch-on-m1-macs-195c399efcc1).
-
-**Usage**:
-`mps` device will be used by default if available similar to the way `cuda` device is used.
-Therefore, no action from user is required. 
-For example, you can run the official Glue text classififcation task (from the root folder) using Apple Silicon GPU with below command:
-
-```bash
-export TASK_NAME=mrpc
-
-python examples/pytorch/text-classification/run_glue.py \
-  --model_name_or_path bert-base-cased \
-  --task_name $TASK_NAME \
-  --do_train \
-  --do_eval \
-  --max_seq_length 128 \
-  --per_device_train_batch_size 32 \
-  --learning_rate 2e-5 \
-  --num_train_epochs 3 \
-  --output_dir /tmp/$TASK_NAME/ \
-  --overwrite_output_dir
-```
-
-**A few caveats to be aware of**
-
-1. Some PyTorch operations have not been implemented in mps and will throw an error. 
-One way to get around that is to set the environment variable `PYTORCH_ENABLE_MPS_FALLBACK=1`, 
-which will fallback to CPU for these operations. It still throws a UserWarning however.
-2. Distributed setups `gloo` and `nccl` are not working with `mps` device. 
-This means that currently only single GPU of `mps` device type can be used.
-
-Finally, please, remember that, 🤗 `Trainer` only integrates MPS backend, therefore if you
-have any problems or questions with regards to MPS backend usage, please, 
-file an issue with [PyTorch GitHub](https://github.com/pytorch/pytorch/issues).
-
-
-## Using Accelerate Launcher with Trainer
-
-Accelerate now powers Trainer. In terms of what users should expect:
-- They can keep using the Trainer ingterations such as FSDP, DeepSpeed vis trainer arguments without any changes on their part.
-- They can now use Accelerate Launcher with Trainer (recommended).
-
-Steps to use Accelerate Launcher with Trainer:
-1. Make sure 🤗 Accelerate is installed, you can't use the `Trainer` without it anyway. If not `pip install accelerate`. You may also need to update your version of Accelerate: `pip install accelerate --upgrade`
-2. Run `accelerate config` and fill the questionnaire. Below are example accelerate configs:
-  a. DDP Multi-node Multi-GPU config:
-    ```yaml
-    compute_environment: LOCAL_MACHINE                                                                                             
-    distributed_type: MULTI_GPU                                                                                                    
-    downcast_bf16: 'no'
-    gpu_ids: all
-    machine_rank: 0 #change rank as per the node
-    main_process_ip: 192.168.20.1
-    main_process_port: 9898
-    main_training_function: main
-    mixed_precision: fp16
-    num_machines: 2
-    num_processes: 8
-    rdzv_backend: static
-    same_network: true
-    tpu_env: []
-    tpu_use_cluster: false
-    tpu_use_sudo: false
-    use_cpu: false
-    ```
-
-  b. FSDP config:
-    ```yaml
-    compute_environment: LOCAL_MACHINE
-    distributed_type: FSDP
-    downcast_bf16: 'no'
-    fsdp_config:
-      fsdp_auto_wrap_policy: TRANSFORMER_BASED_WRAP
-      fsdp_backward_prefetch_policy: BACKWARD_PRE
-      fsdp_forward_prefetch: true
-      fsdp_offload_params: false
-      fsdp_sharding_strategy: 1
-      fsdp_state_dict_type: FULL_STATE_DICT
-      fsdp_sync_module_states: true
-      fsdp_transformer_layer_cls_to_wrap: BertLayer
-      fsdp_use_orig_params: true
-    machine_rank: 0
-    main_training_function: main
-    mixed_precision: bf16
-    num_machines: 1
-    num_processes: 2
-    rdzv_backend: static
-    same_network: true
-    tpu_env: []
-    tpu_use_cluster: false
-    tpu_use_sudo: false
-    use_cpu: false
-    ```
-  c. DeepSpeed config pointing to a file:
-    ```yaml
-    compute_environment: LOCAL_MACHINE
-    deepspeed_config:
-      deepspeed_config_file: /home/user/configs/ds_zero3_config.json
-      zero3_init_flag: true
-    distributed_type: DEEPSPEED
-    downcast_bf16: 'no'
-    machine_rank: 0
-    main_training_function: main
-    num_machines: 1
-    num_processes: 4
-    rdzv_backend: static
-    same_network: true
-    tpu_env: []
-    tpu_use_cluster: false
-    tpu_use_sudo: false
-    use_cpu: false
-    ```
-
-  d. DeepSpeed config using accelerate plugin:
-    ```yaml
-    compute_environment: LOCAL_MACHINE                                                                                             
-    deepspeed_config:                                                                                                              
-      gradient_accumulation_steps: 1
-      gradient_clipping: 0.7
-      offload_optimizer_device: cpu
-      offload_param_device: cpu
-      zero3_init_flag: true
-      zero_stage: 2
-    distributed_type: DEEPSPEED
-    downcast_bf16: 'no'
-    machine_rank: 0
-    main_training_function: main
-    mixed_precision: bf16
-    num_machines: 1
-    num_processes: 4
-    rdzv_backend: static
-    same_network: true
-    tpu_env: []
-    tpu_use_cluster: false
-    tpu_use_sudo: false
-    use_cpu: false
-    ```
-
-3. Run the Trainer script with args other than the ones handled above by accelerate config or launcher args.
-Below is an example to run `run_glue.py` using `accelerate launcher` with FSDP config from above. 
-
-```bash
-cd transformers
-
-accelerate launch \
-./examples/pytorch/text-classification/run_glue.py \
---model_name_or_path bert-base-cased \
---task_name $TASK_NAME \
---do_train \
---do_eval \
---max_seq_length 128 \
---per_device_train_batch_size 16 \
---learning_rate 5e-5 \
---num_train_epochs 3 \
---output_dir /tmp/$TASK_NAME/ \
---overwrite_output_dir
-```
-
-4. You can also directly use the cmd args for `accelerate launch`. Above example would map to:
-
-```bash
-cd transformers
-
-accelerate launch --num_processes=2 \
---use_fsdp \
---mixed_precision=bf16 \
---fsdp_auto_wrap_policy=TRANSFORMER_BASED_WRAP  \
---fsdp_transformer_layer_cls_to_wrap="BertLayer" \
---fsdp_sharding_strategy=1 \
---fsdp_state_dict_type=FULL_STATE_DICT \
-./examples/pytorch/text-classification/run_glue.py
---model_name_or_path bert-base-cased \
---task_name $TASK_NAME \
---do_train \
---do_eval \
---max_seq_length 128 \
---per_device_train_batch_size 16 \
---learning_rate 5e-5 \
---num_train_epochs 3 \
---output_dir /tmp/$TASK_NAME/ \
---overwrite_output_dir
-```
-
-For more information, please refer the 🤗 Accelerate CLI guide: [Launching your 🤗 Accelerate scripts](https://huggingface.co/docs/accelerate/basic_tutorials/launch).
-
-Sections that were moved:
-
-[ <a href="./deepspeed#deepspeed-trainer-integration">DeepSpeed</a><a id="deepspeed"></a>
-| <a href="./deepspeed#deepspeed-installation">Installation</a><a id="installation"></a>
-| <a href="./deepspeed#deepspeed-multi-gpu">Deployment with multiple GPUs</a><a id="deployment-with-multiple-gpus"></a>
-| <a href="./deepspeed#deepspeed-one-gpu">Deployment with one GPU</a><a id="deployment-with-one-gpu"></a>
-| <a href="./deepspeed#deepspeed-notebook">Deployment in Notebooks</a><a id="deployment-in-notebooks"></a>
-| <a href="./deepspeed#deepspeed-config">Configuration</a><a id="configuration"></a>
-| <a href="./deepspeed#deepspeed-config-passing">Passing Configuration</a><a id="passing-configuration"></a>
-| <a href="./deepspeed#deepspeed-config-shared">Shared Configuration</a><a id="shared-configuration"></a>
-| <a href="./deepspeed#deepspeed-zero">ZeRO</a><a id="zero"></a>
-| <a href="./deepspeed#deepspeed-zero2-config">ZeRO-2 Config</a><a id="zero-2-config"></a>
-| <a href="./deepspeed#deepspeed-zero3-config">ZeRO-3 Config</a><a id="zero-3-config"></a>
-| <a href="./deepspeed#deepspeed-nvme">NVMe Support</a><a id="nvme-support"></a>
-| <a href="./deepspeed#deepspeed-zero2-zero3-performance">ZeRO-2 vs ZeRO-3 Performance</a><a id="zero-2-vs-zero-3-performance"></a>
-| <a href="./deepspeed#deepspeed-zero2-example">ZeRO-2 Example</a><a id="zero-2-example"></a>
-| <a href="./deepspeed#deepspeed-zero3-example">ZeRO-3 Example</a><a id="zero-3-example"></a>
-| <a href="./deepspeed#deepspeed-optimizer">Optimizer</a><a id="optimizer"></a>
-| <a href="./deepspeed#deepspeed-scheduler">Scheduler</a><a id="scheduler"></a>
-| <a href="./deepspeed#deepspeed-fp32">fp32 Precision</a><a id="fp32-precision"></a>
-| <a href="./deepspeed#deepspeed-amp">Automatic Mixed Precision</a><a id="automatic-mixed-precision"></a>
-| <a href="./deepspeed#deepspeed-bs">Batch Size</a><a id="batch-size"></a>
-| <a href="./deepspeed#deepspeed-grad-acc">Gradient Accumulation</a><a id="gradient-accumulation"></a>
-| <a href="./deepspeed#deepspeed-grad-clip">Gradient Clipping</a><a id="gradient-clipping"></a>
-| <a href="./deepspeed#deepspeed-weight-extraction">Getting The Model Weights Out</a><a id="getting-the-model-weights-out"></a>
-]
-
-## Boost your fine-tuning performances using NEFTune
-
-
-NEFTune is a technique to boost the performance of chat models and was introduced by the paper “NEFTune: Noisy Embeddings Improve Instruction Finetuning” from Jain et al. it consists of adding noise to the embedding vectors during training. According to the abstract of the paper:
-
-> Standard finetuning of LLaMA-2-7B using Alpaca achieves 29.79% on AlpacaEval, which rises to 64.69% using noisy embeddings. NEFTune also improves over strong baselines on modern instruction datasets. Models trained with Evol-Instruct see a 10% improvement, with ShareGPT an 8% improvement, and with OpenPlatypus an 8% improvement. Even powerful models further refined with RLHF such as LLaMA-2-Chat benefit from additional training with NEFTune.
-
-<div style="text-align: center">
-<img src="https://huggingface.co/datasets/trl-internal-testing/example-images/resolve/main/images/neft-screenshot.png">
-</div>
-
-To use it in `Trainer` simply pass `neftune_noise_alpha` when creating your `TrainingArguments` instance. Note that to avoid any surprising behaviour, NEFTune is disabled after training to retrieve back the original behaviour of the embedding layer.
-
-```python
-from transformers import Trainer, TrainingArguments
-
-args = TrainingArguments(..., neftune_noise_alpha=0.1)
-trainer = Trainer(..., args=args)
-
-...
-
-trainer.train()
-```

docs/source/en/model_doc/fastspeech2_conformer.md

@@ -0,0 +1,134 @@
+<!--Copyright 2023 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+-->
+
+# FastSpeech2Conformer
+
+## Overview
+
+The FastSpeech2Conformer model was proposed with the paper [Recent Developments On Espnet Toolkit Boosted By Conformer](https://arxiv.org/abs/2010.13956) by Pengcheng Guo, Florian Boyer, Xuankai Chang, Tomoki Hayashi, Yosuke Higuchi, Hirofumi Inaguma, Naoyuki Kamo, Chenda Li, Daniel Garcia-Romero, Jiatong Shi, Jing Shi, Shinji Watanabe, Kun Wei, Wangyou Zhang, and Yuekai Zhang.
+
+The abstract from the original FastSpeech2 paper is the following:
+
+*Non-autoregressive text to speech (TTS) models such as FastSpeech (Ren et al., 2019) can synthesize speech significantly faster than previous autoregressive models with comparable quality. The training of FastSpeech model relies on an autoregressive teacher model for duration prediction (to provide more information as input) and knowledge distillation (to simplify the data distribution in output), which can ease the one-to-many mapping problem (i.e., multiple speech variations correspond to the same text) in TTS. However, FastSpeech has several disadvantages: 1) the teacher-student distillation pipeline is complicated and time-consuming, 2) the duration extracted from the teacher model is not accurate enough, and the target mel-spectrograms distilled from teacher model suffer from information loss due to data simplification, both of which limit the voice quality. In this paper, we propose FastSpeech 2, which addresses the issues in FastSpeech and better solves the one-to-many mapping problem in TTS by 1) directly training the model with ground-truth target instead of the simplified output from teacher, and 2) introducing more variation information of speech (e.g., pitch, energy and more accurate duration) as conditional inputs. Specifically, we extract duration, pitch and energy from speech waveform and directly take them as conditional inputs in training and use predicted values in inference. We further design FastSpeech 2s, which is the first attempt to directly generate speech waveform from text in parallel, enjoying the benefit of fully end-to-end inference. Experimental results show that 1) FastSpeech 2 achieves a 3x training speed-up over FastSpeech, and FastSpeech 2s enjoys even faster inference speed; 2) FastSpeech 2 and 2s outperform FastSpeech in voice quality, and FastSpeech 2 can even surpass autoregressive models. Audio samples are available at https://speechresearch.github.io/fastspeech2/.*
+
+This model was contributed by [Connor Henderson](https://huggingface.co/connor-henderson). The original code can be found [here](https://github.com/espnet/espnet/blob/master/espnet2/tts/fastspeech2/fastspeech2.py).
+
+
+## 🤗 Model Architecture
+FastSpeech2's general structure with a Mel-spectrogram decoder was implemented, and the traditional transformer blocks were replaced with with conformer blocks as done in the ESPnet library.
+
+#### FastSpeech2 Model Architecture
+![FastSpeech2 Model Architecture](https://www.microsoft.com/en-us/research/uploads/prod/2021/04/fastspeech2-1.png)
+
+#### Conformer Blocks
+![Conformer Blocks](https://www.researchgate.net/profile/Hirofumi-Inaguma-2/publication/344911155/figure/fig2/AS:951455406108673@1603856054097/An-overview-of-Conformer-block.png)
+
+#### Convolution Module
+![Convolution Module](https://d3i71xaburhd42.cloudfront.net/8809d0732f6147d4ad9218c8f9b20227c837a746/2-Figure1-1.png)
+
+## 🤗 Transformers Usage
+
+You can run FastSpeech2Conformer locally with the 🤗 Transformers library.
+
+1. First install the 🤗 [Transformers library](https://github.com/huggingface/transformers), g2p-en:
+
+```
+pip install --upgrade pip
+pip install --upgrade transformers g2p-en
+```
+
+2. Run inference via the Transformers modelling code with the model and hifigan separately
+
+```python
+
+from transformers import FastSpeech2ConformerTokenizer, FastSpeech2ConformerModel, FastSpeech2ConformerHifiGan
+import soundfile as sf
+
+tokenizer = FastSpeech2ConformerTokenizer.from_pretrained("espnet/fastspeech2_conformer")
+inputs = tokenizer("Hello, my dog is cute.", return_tensors="pt")
+input_ids = inputs["input_ids"]
+
+model = FastSpeech2ConformerModel.from_pretrained("espnet/fastspeech2_conformer")
+output_dict = model(input_ids, return_dict=True)
+spectrogram = output_dict["spectrogram"]
+
+hifigan = FastSpeech2ConformerHifiGan.from_pretrained("espnet/fastspeech2_conformer_hifigan")
+waveform = hifigan(spectrogram)
+
+sf.write("speech.wav", waveform.squeeze().detach().numpy(), samplerate=22050)
+```
+
+3. Run inference via the Transformers modelling code with the model and hifigan combined
+
+```python
+from transformers import FastSpeech2ConformerTokenizer, FastSpeech2ConformerWithHifiGan
+import soundfile as sf
+
+tokenizer = FastSpeech2ConformerTokenizer.from_pretrained("espnet/fastspeech2_conformer")
+inputs = tokenizer("Hello, my dog is cute.", return_tensors="pt")
+input_ids = inputs["input_ids"]
+
+model = FastSpeech2ConformerWithHifiGan.from_pretrained("espnet/fastspeech2_conformer_with_hifigan")
+output_dict = model(input_ids, return_dict=True)
+waveform = output_dict["waveform"]
+
+sf.write("speech.wav", waveform.squeeze().detach().numpy(), samplerate=22050)
+```
+
+4. Run inference with a pipeline and specify which vocoder to use
+```python
+from transformers import pipeline, FastSpeech2ConformerHifiGan
+import soundfile as sf
+
+vocoder = FastSpeech2ConformerHifiGan.from_pretrained("espnet/fastspeech2_conformer_hifigan")
+synthesiser = pipeline(model="espnet/fastspeech2_conformer", vocoder=vocoder)
+
+speech = synthesiser("Hello, my dog is cooler than you!")
+
+sf.write("speech.wav", speech["audio"].squeeze(), samplerate=speech["sampling_rate"])
+```
+
+
+## FastSpeech2ConformerConfig
+
+[[autodoc]] FastSpeech2ConformerConfig
+
+## FastSpeech2ConformerHifiGanConfig
+
+[[autodoc]] FastSpeech2ConformerHifiGanConfig
+
+## FastSpeech2ConformerWithHifiGanConfig
+
+[[autodoc]] FastSpeech2ConformerWithHifiGanConfig
+
+## FastSpeech2ConformerTokenizer
+
+[[autodoc]] FastSpeech2ConformerTokenizer
+    - __call__
+    - save_vocabulary
+    - decode
+    - batch_decode
+
+## FastSpeech2ConformerModel
+
+[[autodoc]] FastSpeech2ConformerModel
+    - forward
+
+## FastSpeech2ConformerHifiGan
+
+[[autodoc]] FastSpeech2ConformerHifiGan
+    - forward
+
+## FastSpeech2ConformerWithHifiGan
+
+[[autodoc]] FastSpeech2ConformerWithHifiGan
+    - forward

docs/source/en/model_doc/gpt-sw3.md

@@ -30,15 +30,15 @@ in collaboration with RISE and the WASP WARA for Media and Language. GPT-Sw3 has
 320B tokens in Swedish, Norwegian, Danish, Icelandic, English, and programming code. The model was pretrained using a
 causal language modeling (CLM) objective utilizing the NeMo Megatron GPT implementation.
 
-This model was contributed by [AI Sweden](https://huggingface.co/AI-Sweden).
+This model was contributed by [AI Sweden Models](https://huggingface.co/AI-Sweden-Models).
 
 ## Usage example
 
 ```python
 >>> from transformers import AutoTokenizer, AutoModelForCausalLM
 
->>> tokenizer = AutoTokenizer.from_pretrained("AI-Sweden/gpt-sw3-356m")
->>> model = AutoModelForCausalLM.from_pretrained("AI-Sweden/gpt-sw3-356m")
+>>> tokenizer = AutoTokenizer.from_pretrained("AI-Sweden-Models/gpt-sw3-356m")
+>>> model = AutoModelForCausalLM.from_pretrained("AI-Sweden-Models/gpt-sw3-356m")
 
 >>> input_ids = tokenizer("Träd är fina för att", return_tensors="pt")["input_ids"]
 

docs/source/en/model_doc/llama2.md

@@ -67,6 +67,8 @@ come in several checkpoints they each contain a part of each weight of the model
 
 - The LLaMA tokenizer is a BPE model based on [sentencepiece](https://github.com/google/sentencepiece). One quirk of sentencepiece is that when decoding a sequence, if the first token is the start of the word (e.g. "Banana"), the tokenizer does not prepend the prefix space to the string.
 
+- When using Flash Attention 2 via `attn_implementation="flash_attention_2"`, don't pass `torch_dtype` to the `from_pretrained` class method and use Automatic Mixed-Precision training. When using `Trainer`, it is simply specifying either `fp16` or `bf16` to `True`. Otherwise, make sure you are using `torch.autocast`. This is required because the Flash Attention only support `fp16` and `bf16` data type.
+
 
 ## Resources
 

docs/source/en/model_doc/mixtral.md

@@ -20,7 +20,7 @@ rendered properly in your Markdown viewer.
 
 Mixtral-8x7B is Mistral AI's second Large Language Model (LLM). 
 
-The Mixtral model was proposed in the by the [Mistral AI](https://mistral.ai/) team.
+The Mixtral model was proposed by the [Mistral AI](https://mistral.ai/) team.
 
 It was introduced in the [Mixtral of Experts blogpost](https://mistral.ai/news/mixtral-of-experts/) with the following introduction:
 
@@ -42,7 +42,7 @@ Mixtral-45B is a decoder-based LM with the following architectural choices:
 
 * Mixtral is a Mixture of Expert (MOE) model with 8 experts per MLP, with a total of 45B paramateres but the compute required is the same as a 14B model. This is because even though each experts have to be loaded in RAM (70B like ram requirement) each token from the hidden states are dipatched twice (top 2 routing) and thus the compute (the operation required at each foward computation) is just 2 X sequence_length. 
 
-The following implementation details are shared with Mistral AI's first model [mistral](~models/doc/mistral):
+The following implementation details are shared with Mistral AI's first model [mistral](mistral):
 * Sliding Window Attention - Trained with 8k context length and fixed cache size, with a theoretical attention span of 128K tokens
 * GQA (Grouped Query Attention) - allowing faster inference and lower cache size.
 * Byte-fallback BPE tokenizer - ensures that characters are never mapped to out of vocabulary tokens.
@@ -66,7 +66,7 @@ These ready-to-use checkpoints can be downloaded and used via the HuggingFace Hu
 >>> device = "cuda" # the device to load the model onto
 
 >>> model = AutoModelForCausalLM.from_pretrained("mistralai/Mixtral-8x7B-v0.1")
->>> tokenizer = AutoTokenizer.from_pretrained("mistralai/Mistral-8x7B")
+>>> tokenizer = AutoTokenizer.from_pretrained("mistralai/Mixtral-8x7B-v0.1")
 
 >>> prompt = "My favourite condiment is"
 

docs/source/en/model_doc/phi.md

@@ -23,6 +23,7 @@ The Phi-1 model was proposed in [Textbooks Are All You Need](https://arxiv.org/a
 The Phi-1.5 model was proposed in [Textbooks Are All You Need II: phi-1.5 technical report](https://arxiv.org/abs/2309.05463) by Yuanzhi Li, Sébastien Bubeck, Ronen Eldan, Allie Del Giorno, Suriya Gunasekar and Yin Tat Lee.
 
 ### Summary
+
 In Phi-1 and Phi-1.5 papers, the authors showed how important the quality of the data is in training relative to the model size.
 They selected high quality "textbook" data alongside with synthetically generated data for training their small sized Transformer
 based model Phi-1 with 1.3B parameters. Despite this small scale, phi-1 attains pass@1 accuracy 50.6% on HumanEval and 55.5% on MBPP.
@@ -31,7 +32,6 @@ to models 5x larger, and surpassing most non-frontier LLMs. Phi-1.5 exhibits man
 to “think step by step” or perform some rudimentary in-context learning.
 With these two experiments the authors successfully showed the huge impact of quality of training data when training machine learning models.
 
-
 The abstract from the Phi-1 paper is the following:
 
 *We introduce phi-1, a new large language model for code, with significantly smaller size than
@@ -60,16 +60,40 @@ including hallucinations and the potential for toxic and biased generations –e
 are seeing improvement on that front thanks to the absence of web data. We open-source phi-1.5 to
 promote further research on these urgent topics.*
 
-
 This model was contributed by [Susnato Dhar](https://huggingface.co/susnato).
-The original code for Phi-1 and Phi-1.5 can be found [here](https://huggingface.co/microsoft/phi-1/blob/main/modeling_mixformer_sequential.py) and [here](https://huggingface.co/microsoft/phi-1_5/blob/main/modeling_mixformer_sequential.py) respectively.
 
+The original code for Phi-1, Phi-1.5 and Phi-2 can be found [here](https://huggingface.co/microsoft/phi-1), [here](https://huggingface.co/microsoft/phi-1_5) and [here](https://huggingface.co/microsoft/phi-2), respectively.
 
 ## Usage tips
 
 - This model is quite similar to `Llama` with the main difference in [`PhiDecoderLayer`], where they used [`PhiAttention`] and [`PhiMLP`] layers in parallel configuration.
 - The tokenizer used for this model is identical to the [`CodeGenTokenizer`].
 
+## How to use Phi-2
+
+<Tip warning={true}>
+
+Phi-2 has been integrated in the development version (4.37.0.dev) of `transformers`. Until the official version is released through `pip`, ensure that you are doing one of the following:
+
+* When loading the model, ensure that `trust_remote_code=True` is passed as an argument of the `from_pretrained()` function.
+
+* Update your local `transformers` to the development version: `pip uninstall -y transformers && pip install git+https://github.com/huggingface/transformers`. The previous command is an alternative to cloning and installing from the source.
+
+</Tip>
+
+```python
+>>> from transformers import AutoModelForCausalLM, AutoTokenizer
+
+>>> model = AutoModelForCausalLM.from_pretrained("phi-2")
+>>> tokenizer = AutoTokenizer.from_pretrained("phi-2")
+
+>>> inputs = tokenizer('Can you help me write a formal email to a potential business partner proposing a joint venture?', return_tensors="pt", return_attention_mask=False)
+
+>>> outputs = model.generate(**inputs, max_length=30)
+>>> text = tokenizer.batch_decode(outputs)[0]
+>>> print(text)
+'Can you help me write a formal email to a potential business partner proposing a joint venture?\nInput: Company A: ABC Inc.\nCompany B: XYZ Ltd.\nJoint Venture: A new online platform for e-commerce'
+```
 
 ### Example :
 
@@ -77,8 +101,8 @@ The original code for Phi-1 and Phi-1.5 can be found [here](https://huggingface.
 >>> from transformers import PhiForCausalLM, AutoTokenizer
 
 >>> # define the model and tokenizer.
->>> model = PhiForCausalLM.from_pretrained("susnato/phi-1_5_dev")
->>> tokenizer = AutoTokenizer.from_pretrained("susnato/phi-1_5_dev")
+>>> model = PhiForCausalLM.from_pretrained("microsoft/phi-1_5")
+>>> tokenizer = AutoTokenizer.from_pretrained("microsoft/phi-1_5")
 
 >>> # feel free to change the prompt to your liking.
 >>> prompt = "If I were an AI that had just achieved"
@@ -93,7 +117,6 @@ The original code for Phi-1 and Phi-1.5 can be found [here](https://huggingface.
 'If I were an AI that had just achieved a breakthrough in machine learning, I would be thrilled'
 ```
 
-
 ## Combining Phi and Flash Attention 2
 
 First, make sure to install the latest version of Flash Attention 2 to include the sliding window attention feature.
@@ -111,8 +134,8 @@ To load and run a model using Flash Attention 2, refer to the snippet below:
 >>> from transformers import PhiForCausalLM, AutoTokenizer
 
 >>> # define the model and tokenizer and push the model and tokens to the GPU.
->>> model = PhiForCausalLM.from_pretrained("susnato/phi-1_5_dev", torch_dtype=torch.float16, attn_implementation="flash_attention_2").to("cuda")
->>> tokenizer = AutoTokenizer.from_pretrained("susnato/phi-1_5_dev")
+>>> model = PhiForCausalLM.from_pretrained("microsoft/phi-1_5", torch_dtype=torch.float16, attn_implementation="flash_attention_2").to("cuda")
+>>> tokenizer = AutoTokenizer.from_pretrained("microsoft/phi-1_5")
 
 >>> # feel free to change the prompt to your liking.
 >>> prompt = "If I were an AI that had just achieved"
@@ -128,12 +151,13 @@ To load and run a model using Flash Attention 2, refer to the snippet below:
 ```
 
 ### Expected speedups
-Below is an expected speedup diagram that compares pure inference time between the native implementation in transformers using `susnato/phi-1_dev` checkpoint and the Flash Attention 2 version of the model using a sequence length of 2048.
+
+Below is an expected speedup diagram that compares pure inference time between the native implementation in transformers using `microsoft/phi-1` checkpoint and the Flash Attention 2 version of the model using a sequence length of 2048.
+
 <div style="text-align: center">
 <img src="https://huggingface.co/datasets/ybelkada/documentation-images/resolve/main/phi_1_speedup_plot.jpg">
 </div>
 
-
 ## PhiConfig
 
 [[autodoc]] PhiConfig

docs/source/en/model_doc/qwen2.md

@@ -0,0 +1,82 @@
+<!--Copyright 2024 The Qwen Team and The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+
+⚠️ Note that this file is in Markdown but contain specific syntax for our doc-builder (similar to MDX) that may not be
+rendered properly in your Markdown viewer.
+
+-->
+
+# Qwen2
+
+## Overview
+
+Qwen2 is the new model series of large language models from the Qwen team. Previously, we released the Qwen series, including Qwen-72B, Qwen-1.8B, Qwen-VL, Qwen-Audio, etc.
+
+### Model Details
+
+Qwen2 is a language model series including decoder language models of different model sizes. For each size, we release the base language model and the aligned chat model. It is based on the Transformer architecture with SwiGLU activation, attention QKV bias, group query attention, mixture of sliding window attention and full attention, etc. Additionally, we have an improved tokenizer adaptive to multiple natural languages and codes.
+
+
+## Usage tips
+
+`Qwen2-7B-beta` and `Qwen2-7B-Chat-beta` can be found on the [Huggingface Hub](https://huggingface.co/Qwen)
+
+In the following, we demonstrate how to use `Qwen2-7B-Chat-beta` for the inference. Note that we have used the ChatML format for dialog, in this demo we show how to leverage `apply_chat_template` for this purpose.
+
+```python
+>>> from transformers import AutoModelForCausalLM, AutoTokenizer
+>>> device = "cuda" # the device to load the model onto
+
+>>> model = AutoModelForCausalLM.from_pretrained("Qwen2/Qwen2-7B-Chat-beta", device_map="auto")
+>>> tokenizer = AutoTokenizer.from_pretrained("Qwen2/Qwen2-7B-Chat-beta")
+
+>>> prompt = "Give me a short introduction to large language model."
+
+>>> messages = [{"role": "user", "content": prompt}]
+
+>>> text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
+
+>>> model_inputs = tokenizer([text], return_tensors="pt").to(device)
+
+>>> generated_ids = model.generate(model_inputs.input_ids, max_new_tokens=512, do_sample=True)
+
+>>> generated_ids = [output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)]
+
+>>> response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
+```
+
+## Qwen2Config
+
+[[autodoc]] Qwen2Config
+
+## Qwen2Tokenizer
+
+[[autodoc]] Qwen2Tokenizer
+    - save_vocabulary
+
+## Qwen2TokenizerFast
+
+[[autodoc]] Qwen2TokenizerFast
+
+## Qwen2Model
+
+[[autodoc]] Qwen2Model
+    - forward
+
+## Qwen2ForCausalLM
+
+[[autodoc]] Qwen2ForCausalLM
+    - forward
+
+## Qwen2ForSequenceClassification
+
+[[autodoc]] Qwen2ForSequenceClassification
+    - forward

docs/source/en/model_doc/sam.md

@@ -66,6 +66,34 @@ masks = processor.image_processor.post_process_masks(
 scores = outputs.iou_scores
 ```
 
+You can also process your own masks alongside the input images in the processor to be passed to the model.
+
+```python
+import torch
+from PIL import Image
+import requests
+from transformers import SamModel, SamProcessor
+
+device = "cuda" if torch.cuda.is_available() else "cpu"
+model = SamModel.from_pretrained("facebook/sam-vit-huge").to(device)
+processor = SamProcessor.from_pretrained("facebook/sam-vit-huge")
+
+img_url = "https://huggingface.co/ybelkada/segment-anything/resolve/main/assets/car.png"
+raw_image = Image.open(requests.get(img_url, stream=True).raw).convert("RGB")
+mask_url = "https://huggingface.co/ybelkada/segment-anything/resolve/main/assets/car.png"
+segmentation_map = Image.open(requests.get(mask_url, stream=True).raw).convert("RGB")
+input_points = [[[450, 600]]]  # 2D location of a window in the image
+
+inputs = processor(raw_image, input_points=input_points, segmentation_maps=mask, return_tensors="pt").to(device)
+with torch.no_grad():
+    outputs = model(**inputs)
+
+masks = processor.image_processor.post_process_masks(
+    outputs.pred_masks.cpu(), inputs["original_sizes"].cpu(), inputs["reshaped_input_sizes"].cpu()
+)
+scores = outputs.iou_scores
+```
+
 Resources:
 
 - [Demo notebook](https://github.com/huggingface/notebooks/blob/main/examples/segment_anything.ipynb) for using the model.

docs/source/en/model_doc/siglip.md

@@ -0,0 +1,142 @@
+<!--Copyright 2023 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+
+⚠️ Note that this file is in Markdown but contain specific syntax for our doc-builder (similar to MDX) that may not be
+rendered properly in your Markdown viewer.
+
+-->
+
+# SigLIP
+
+## Overview
+
+The SigLIP model was proposed in [Sigmoid Loss for Language Image Pre-Training](https://arxiv.org/abs/2303.15343) by Xiaohua Zhai, Basil Mustafa, Alexander Kolesnikov, Lucas Beyer. SigLIP proposes to replace the loss function used in [CLIP](clip) by a simple pairwise sigmoid loss. This results in better performance in terms of zero-shot classification accuracy on ImageNet.
+
+The abstract from the paper is the following:
+
+*We propose a simple pairwise Sigmoid loss for Language-Image Pre-training (SigLIP). Unlike standard contrastive learning with softmax normalization, the sigmoid loss operates solely on image-text pairs and does not require a global view of the pairwise similarities for normalization. The sigmoid loss simultaneously allows further scaling up the batch size, while also performing better at smaller batch sizes. Combined with Locked-image Tuning, with only four TPUv4 chips, we train a SigLiT model that achieves 84.5% ImageNet zero-shot accuracy in two days. The disentanglement of the batch size from the loss further allows us to study the impact of examples vs pairs and negative to positive ratio. Finally, we push the batch size to the extreme, up to one million, and find that the benefits of growing batch size quickly diminish, with a more reasonable batch size of 32k being sufficient.*
+
+## Usage tips
+
+- Usage of SigLIP is similar to [CLIP](clip). The main difference is the training loss, which does not require a global view of all the pairwise similarities of images and texts within a batch. One needs to apply the sigmoid activation function to the logits, rather than the softmax.
+- Training is not yet supported. If you want to fine-tune SigLIP or train from scratch, refer to the loss function from [OpenCLIP](https://github.com/mlfoundations/open_clip/blob/73ad04ae7fb93ede1c02dc9040a828634cb1edf1/src/open_clip/loss.py#L307), which leverages various `torch.distributed` utilities.
+- When using the standalone [`SiglipTokenizer`] or [`SiglipProcessor`], make sure to pass `padding="max_length"` as that's how the model was trained.
+
+<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers/model_doc/siglip_table.jpeg"
+alt="drawing" width="600"/>
+
+<small> SigLIP evaluation results compared to CLIP. Taken from the <a href="https://arxiv.org/abs/2303.15343">original paper</a>.</small>
+
+This model was contributed by [nielsr](https://huggingface.co/nielsr).
+The original code can be found [here](https://github.com/google-research/big_vision/tree/main).
+
+## Usage example
+
+There are 2 main ways to use SigLIP: either using the pipeline API, which abstracts away all the complexity for you, or by using the `SiglipModel` class yourself.
+
+### Pipeline API
+
+The pipeline allows to use the model in a few lines of code:
+
+```python
+>>> from transformers import pipeline
+>>> from PIL import Image
+>>> import requests
+
+>>> # load pipe
+>>> image_classifier = pipeline(task="zero-shot-image-classification", model="google/siglip-base-patch16-224")
+
+>>> # load image
+>>> url = 'http://images.cocodataset.org/val2017/000000039769.jpg'
+>>> image = Image.open(requests.get(url, stream=True).raw)
+
+>>> # inference
+>>> outputs = image_classifier(image, candidate_labels=["2 cats", "a plane", "a remote"])
+>>> outputs = [{"score": round(output["score"], 4), "label": output["label"] } for output in outputs]
+>>> print(outputs)
+[{'score': 0.1979, 'label': '2 cats'}, {'score': 0.0, 'label': 'a remote'}, {'score': 0.0, 'label': 'a plane'}]
+```
+
+### Using the model yourself
+
+If you want to do the pre- and postprocessing yourself, here's how to do that:
+
+```python
+>>> from PIL import Image
+>>> import requests
+>>> from transformers import AutoProcessor, AutoModel
+>>> import torch
+
+>>> model = AutoModel.from_pretrained("google/siglip-base-patch16-224")
+>>> processor = AutoProcessor.from_pretrained("google/siglip-base-patch16-224")
+
+>>> url = "http://images.cocodataset.org/val2017/000000039769.jpg"
+>>> image = Image.open(requests.get(url, stream=True).raw)
+
+>>> texts = ["a photo of 2 cats", "a photo of 2 dogs"]
+>>> # important: we pass `padding=max_length` since the model was trained with this
+>>> inputs = processor(text=texts, images=image, padding="max_length", return_tensors="pt")
+
+>>> with torch.no_grad():
+...     outputs = model(**inputs)
+
+>>> logits_per_image = outputs.logits_per_image
+>>> probs = torch.sigmoid(logits_per_image) # these are the probabilities
+>>> print(f"{probs[0][0]:.1%} that image 0 is '{texts[0]}'")
+31.9% that image 0 is 'a photo of 2 cats'
+```
+
+## SiglipConfig
+
+[[autodoc]] SiglipConfig
+    - from_text_vision_configs
+
+## SiglipTextConfig
+
+[[autodoc]] SiglipTextConfig
+
+## SiglipVisionConfig
+
+[[autodoc]] SiglipVisionConfig
+
+## SiglipTokenizer
+
+[[autodoc]] SiglipTokenizer
+    - build_inputs_with_special_tokens
+    - get_special_tokens_mask
+    - create_token_type_ids_from_sequences
+    - save_vocabulary
+
+## SiglipImageProcessor
+
+[[autodoc]] SiglipImageProcessor
+    - preprocess
+
+## SiglipProcessor
+
+[[autodoc]] SiglipProcessor
+
+## SiglipModel
+
+[[autodoc]] SiglipModel
+    - forward
+    - get_text_features
+    - get_image_features
+
+## SiglipTextModel
+
+[[autodoc]] SiglipTextModel
+    - forward
+
+## SiglipVisionModel
+
+[[autodoc]] SiglipVisionModel
+    - forward

docs/source/en/model_doc/vipllava.md

@@ -37,13 +37,13 @@ Tips:
 - For better results, we recommend users to prompt the model with the correct prompt format: 
 
 ```bash
-"USER: <image>\n<prompt>ASSISTANT:"
+A chat between a curious human and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the human's questions.###Human: <image>\n<prompt>###Assistant:
 ```
 
 For multiple turns conversation:
 
 ```bash
-"USER: <image>\n<prompt1>ASSISTANT: <answer1>USER: <prompt2>ASSISTANT: <answer2>USER: <prompt3>ASSISTANT:"
+A chat between a curious human and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the human's questions.###Human: <image>\n<prompt1>###Assistant: <answer1>###Human: <prompt2>###Assistant:
 ```
 
 The original code can be found [here](https://github.com/mu-cai/ViP-LLaVA).

docs/source/en/model_doc/wav2vec2-bert.md

@@ -0,0 +1,90 @@
+<!--Copyright 2024 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+
+⚠️ Note that this file is in Markdown but contain specific syntax for our doc-builder (similar to MDX) that may not be
+rendered properly in your Markdown viewer.
+
+-->
+
+# Wav2Vec2-BERT
+
+## Overview
+
+The Wav2Vec2-BERT model was proposed in [Seamless: Multilingual Expressive and Streaming Speech Translation](https://ai.meta.com/research/publications/seamless-multilingual-expressive-and-streaming-speech-translation/) by the Seamless Communication team from Meta AI.
+
+This model was pre-trained on 4.5M hours of unlabeled audio data covering more than 143 languages. It requires finetuning to be used for downstream tasks such as Automatic Speech Recognition (ASR), or Audio Classification.
+
+The official results of the model can be found in Section 3.2.1 of the paper.
+
+The abstract from the paper is the following:
+
+*Recent advancements in automatic speech translation have dramatically expanded language coverage, improved multimodal capabilities, and enabled a wide range of tasks and functionalities. That said, large-scale automatic speech translation systems today lack key features that help machine-mediated communication feel seamless when compared to human-to-human dialogue. In this work, we introduce a family of models that enable end-to-end expressive and multilingual translations in a streaming fashion. First, we contribute an improved version of the massively multilingual and multimodal SeamlessM4T model—SeamlessM4T v2. This newer model, incorporating an updated UnitY2 framework, was trained on more low-resource language data. The expanded version of SeamlessAlign adds 114,800 hours of automatically aligned data for a total of 76 languages. SeamlessM4T v2 provides the foundation on which our two newest models, SeamlessExpressive and SeamlessStreaming, are initiated. SeamlessExpressive enables translation that preserves vocal styles and prosody. Compared to previous efforts in expressive speech research, our work addresses certain underexplored aspects of prosody, such as speech rate and pauses, while also preserving the style of one’s voice. As for SeamlessStreaming, our model leverages the Efficient Monotonic Multihead Attention (EMMA) mechanism to generate low-latency target translations without waiting for complete source utterances. As the first of its kind, SeamlessStreaming enables simultaneous speech-to-speech/text translation for multiple source and target languages. To understand the performance of these models, we combined novel and modified versions of existing automatic metrics to evaluate prosody, latency, and robustness. For human evaluations, we adapted existing protocols tailored for measuring the most relevant attributes in the preservation of meaning, naturalness, and expressivity. To ensure that our models can be used safely and responsibly, we implemented the first known red-teaming effort for multimodal machine translation, a system for the detection and mitigation of added toxicity, a systematic evaluation of gender bias, and an inaudible localized watermarking mechanism designed to dampen the impact of deepfakes. Consequently, we bring major components from SeamlessExpressive and SeamlessStreaming together to form Seamless, the first publicly available system that unlocks expressive cross-lingual communication in real-time. In sum, Seamless gives us a pivotal look at the technical foundation needed to turn the Universal Speech Translator from a science fiction concept into a real-world technology. Finally, contributions in this work—including models, code, and a watermark detector—are publicly released and accessible at the link below.*
+
+This model was contributed by [ylacombe](https://huggingface.co/ylacombe). The original code can be found [here](https://github.com/facebookresearch/seamless_communication).
+
+## Usage tips
+
+- Wav2Vec2-BERT follows the same architecture as Wav2Vec2-Conformer, but employs a causal depthwise convolutional layer and uses as input a mel-spectrogram representation of the audio instead of the raw waveform.
+- Wav2Vec2-BERT can use either no relative position embeddings, Shaw-like position embeddings, Transformer-XL-like position embeddings, or
+  rotary position embeddings by setting the correct `config.position_embeddings_type`.
+- Wav2Vec2-BERT also introduces a Conformer-based adapter network instead of a simple convolutional network.
+
+## Resources
+
+<PipelineTag pipeline="automatic-speech-recognition"/>
+
+- [`Wav2Vec2BertForCTC`] is supported by this [example script](https://github.com/huggingface/transformers/tree/main/examples/pytorch/speech-recognition).
+- You can also adapt these notebooks on [how to finetune a speech recognition model in English](https://colab.research.google.com/github/huggingface/notebooks/blob/main/examples/speech_recognition.ipynb), and [how to finetune a speech recognition model in any language](https://colab.research.google.com/github/huggingface/notebooks/blob/main/examples/multi_lingual_speech_recognition.ipynb).
+
+<PipelineTag pipeline="audio-classification"/>
+
+- [`Wav2Vec2BertForSequenceClassification`] can be used by adapting this [example script](https://github.com/huggingface/transformers/tree/main/examples/pytorch/audio-classification).
+- See also: [Audio classification task guide](../tasks/audio_classification)
+
+
+## Wav2Vec2BertConfig
+
+[[autodoc]] Wav2Vec2BertConfig
+
+## Wav2Vec2BertProcessor
+
+[[autodoc]] Wav2Vec2BertProcessor
+    - __call__
+    - pad
+    - from_pretrained
+    - save_pretrained
+    - batch_decode
+    - decode
+
+## Wav2Vec2BertModel
+
+[[autodoc]] Wav2Vec2BertModel
+    - forward
+
+## Wav2Vec2BertForCTC
+
+[[autodoc]] Wav2Vec2BertForCTC
+    - forward
+
+## Wav2Vec2BertForSequenceClassification
+
+[[autodoc]] Wav2Vec2BertForSequenceClassification
+    - forward
+
+## Wav2Vec2BertForAudioFrameClassification
+
+[[autodoc]] Wav2Vec2BertForAudioFrameClassification
+    - forward
+
+## Wav2Vec2BertForXVector
+
+[[autodoc]] Wav2Vec2BertForXVector
+    - forward

docs/source/en/perf_infer_gpu_one.md

@@ -52,13 +52,29 @@ FlashAttention-2 is currently supported for the following architectures:
 * [Mixtral](https://huggingface.co/docs/transformers/model_doc/mixtral#transformers.MixtralModel)
 * [OPT](https://huggingface.co/docs/transformers/model_doc/opt#transformers.OPTModel)
 * [Phi](https://huggingface.co/docs/transformers/model_doc/phi#transformers.PhiModel)
+* [Qwen2](https://huggingface.co/docs/transformers/model_doc/qwen2#transformers.Qwen2Model)
 * [Whisper](https://huggingface.co/docs/transformers/model_doc/whisper#transformers.WhisperModel)
 
 You can request to add FlashAttention-2 support for another model by opening a GitHub Issue or Pull Request.
 
-Before you begin, make sure you have FlashAttention-2 installed. For NVIDIA GPUs, the library is installable through pip: `pip install flash-attn --no-build-isolation`. We strongly suggest to refer to the [detailed installation instructions](https://github.com/Dao-AILab/flash-attention?tab=readme-ov-file#installation-and-features).
+Before you begin, make sure you have FlashAttention-2 installed.
 
-FlashAttention-2 is also supported on AMD GPUs, with the current support limited to **Instinct MI210 and Instinct MI250**. We strongly suggest to use the following [Dockerfile](https://github.com/huggingface/optimum-amd/tree/main/docker/transformers-pytorch-amd-gpu-flash/Dockerfile) to use FlashAttention-2 on AMD GPUs.
+<hfoptions id="install">
+<hfoption id="NVIDIA">
+
+```bash
+pip install flash-attn --no-build-isolation
+```
+
+We strongly suggest referring to the detailed [installation instructions](https://github.com/Dao-AILab/flash-attention?tab=readme-ov-file#installation-and-features) to learn more about supported hardware and data types!
+
+</hfoption>
+<hfoption id="AMD">
+
+FlashAttention-2 is also supported on AMD GPUs and current support is limited to **Instinct MI210** and **Instinct MI250**. We strongly suggest using this [Dockerfile](https://github.com/huggingface/optimum-amd/tree/main/docker/transformers-pytorch-amd-gpu-flash/Dockerfile) to use FlashAttention-2 on AMD GPUs.
+
+</hfoption>
+</hfoptions>
 
 To enable FlashAttention-2, pass the argument `attn_implementation="flash_attention_2"` to [`~AutoModelForCausalLM.from_pretrained`]:
 
@@ -80,7 +96,9 @@ model = AutoModelForCausalLM.from_pretrained(
 
 FlashAttention-2 can only be used when the model's dtype is `fp16` or `bf16`. Make sure to cast your model to the appropriate dtype and load them on a supported device before using FlashAttention-2.
 
-Note that `use_flash_attention_2=True` can also be used to enable Flash Attention 2, but is deprecated in favor of `attn_implementation="flash_attention_2"`.
+<br>
+
+You can also set `use_flash_attention_2=True` to enable FlashAttention-2 but it is deprecated in favor of `attn_implementation="flash_attention_2"`.
   
 </Tip>
 
@@ -144,21 +162,28 @@ FlashAttention is more memory efficient, meaning you can train on much larger se
 <img src="https://huggingface.co/datasets/ybelkada/documentation-images/resolve/main/llama-2-large-seqlen-padding.png">
 </div>
 
-## FlashAttention and memory-efficient attention through PyTorch's scaled_dot_product_attention 
+## PyTorch scaled dot product attention
 
-PyTorch's [`torch.nn.functional.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html) (SDPA) can also call FlashAttention and memory-efficient attention kernels under the hood. SDPA support is currently being added natively in Transformers, and is used by default for `torch>=2.1.1` when an implementation is available.
+PyTorch's [`torch.nn.functional.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html) (SDPA) can also call FlashAttention and memory-efficient attention kernels under the hood. SDPA support is currently being added natively in Transformers and is used by default for `torch>=2.1.1` when an implementation is available.
 
-For now, Transformers supports inference and training through SDPA for the following architectures:
+For now, Transformers supports SDPA inference and training for the following architectures:
 * [Bart](https://huggingface.co/docs/transformers/model_doc/bart#transformers.BartModel)
 * [GPTBigCode](https://huggingface.co/docs/transformers/model_doc/gpt_bigcode#transformers.GPTBigCodeModel)
 * [Falcon](https://huggingface.co/docs/transformers/model_doc/falcon#transformers.FalconModel)
 * [Llama](https://huggingface.co/docs/transformers/model_doc/llama#transformers.LlamaModel)
 * [Idefics](https://huggingface.co/docs/transformers/model_doc/idefics#transformers.IdeficsModel)
 * [Whisper](https://huggingface.co/docs/transformers/model_doc/whisper#transformers.WhisperModel)
+* [Mistral](https://huggingface.co/docs/transformers/model_doc/mistral#transformers.MistralModel)
+* [Mixtral](https://huggingface.co/docs/transformers/model_doc/mixtral#transformers.MixtralModel)
+* [Qwen2](https://huggingface.co/docs/transformers/model_doc/qwen2#transformers.Qwen2Model)
 
-Note that FlashAttention can only be used for models with the `fp16` or `bf16` torch type, so make sure to cast your model to the appropriate type before using it.
+<Tip>
 
-By default, `torch.nn.functional.scaled_dot_product_attention` selects the most performant kernel available, but to check whether a backend is available in a given setting (hardware, problem size), you can use [`torch.backends.cuda.sdp_kernel`](https://pytorch.org/docs/master/backends.html#torch.backends.cuda.sdp_kernel) as a context manager:
+FlashAttention can only be used for models with the `fp16` or `bf16` torch type, so make sure to cast your model to the appropriate type first.
+
+</Tip>
+
+By default, SDPA selects the most performant kernel available but you can check whether a backend is available in a given setting (hardware, problem size) with [`torch.backends.cuda.sdp_kernel`](https://pytorch.org/docs/master/backends.html#torch.backends.cuda.sdp_kernel) as a context manager:
 
 ```diff
 import torch
@@ -178,7 +203,7 @@ inputs = tokenizer(input_text, return_tensors="pt").to("cuda")
 print(tokenizer.decode(outputs[0], skip_special_tokens=True))
 ```
 
-If you see a bug with the traceback below, try using nightly version of PyTorch which may have broader coverage for FlashAttention:
+If you see a bug with the traceback below, try using the nightly version of PyTorch which may have broader coverage for FlashAttention:
 
 ```bash
 RuntimeError: No available kernel. Aborting execution.
@@ -191,11 +216,10 @@ pip3 install -U --pre torch torchvision torchaudio --index-url https://download.
 
 <Tip warning={true}>
 
-Part of BetterTransformer features are being upstreamed in Transformers, with native `torch.nn.scaled_dot_product_attention` default support. BetterTransformer still has a wider coverage than the Transformers SDPA integration, but you can expect more and more architectures to support natively SDPA in Transformers.
+Some BetterTransformer features are being upstreamed to Transformers with default support for native `torch.nn.scaled_dot_product_attention`. BetterTransformer still has a wider coverage than the Transformers SDPA integration, but you can expect more and more architectures to natively support SDPA in Transformers.
 
 </Tip>
 
-
 <Tip>
 
 Check out our benchmarks with BetterTransformer and scaled dot product attention in the [Out of the box acceleration and memory savings of 🤗 decoder models with PyTorch 2.0](https://pytorch.org/blog/out-of-the-box-acceleration/) and learn more about the fastpath execution in the [BetterTransformer](https://medium.com/pytorch/bettertransformer-out-of-the-box-performance-for-huggingface-transformers-3fbe27d50ab2) blog post.

docs/source/en/perf_train_cpu.md

@@ -31,14 +31,16 @@ IPEX release is following PyTorch, to install via pip:
 
 | PyTorch Version   | IPEX version   |
 | :---------------: | :----------:   |
+| 2.1.x             |  2.1.100+cpu   |
+| 2.0.x             |  2.0.100+cpu   |
 | 1.13              |  1.13.0+cpu    |
 | 1.12              |  1.12.300+cpu  |
-| 1.11              |  1.11.200+cpu  |
-| 1.10              |  1.10.100+cpu  |
 
+Please run `pip list | grep torch` to get your `pytorch_version`, so you can get the `IPEX version_name`.
 ```
 pip install intel_extension_for_pytorch==<version_name> -f https://developer.intel.com/ipex-whl-stable-cpu
 ```
+You can check the latest versions in [ipex-whl-stable-cpu](https://developer.intel.com/ipex-whl-stable-cpu) if needed.
 
 Check more approaches for [IPEX installation](https://intel.github.io/intel-extension-for-pytorch/cpu/latest/tutorials/installation.html).
 
@@ -59,8 +61,20 @@ Take an example of the use cases on [Transformers question-answering](https://gi
 --max_seq_length 384 \
 --doc_stride 128 \
 --output_dir /tmp/debug_squad/ \
-<b>--use_ipex \</b>
-<b>--bf16 --no_cuda</b></pre> 
+<b>--use_ipex</b> \
+<b>--bf16</b> \
+<b>--use_cpu</b></pre> 
+
+If you want to enable `use_ipex` and `bf16` in your script, add these parameters to `TrainingArguments` like this:
+```diff
+training_args = TrainingArguments(
+    output_dir=args.output_path,
++   bf16=True,
++   use_ipex=True,
++   use_cpu=True,
+    **kwargs
+)
+```
 
 ### Practice example
 

docs/source/en/perf_train_cpu_many.md

@@ -32,16 +32,17 @@ Wheel files are available for the following Python versions:
 
 | Extension Version | Python 3.6 | Python 3.7 | Python 3.8 | Python 3.9 | Python 3.10 |
 | :---------------: | :--------: | :--------: | :--------: | :--------: | :---------: |
+| 2.1.0             |            | √          | √          | √          | √           |
+| 2.0.0             |            | √          | √          | √          | √           |
 | 1.13.0            |            | √          | √          | √          | √           |
 | 1.12.100          |            | √          | √          | √          | √           |
 | 1.12.0            |            | √          | √          | √          | √           |
-| 1.11.0            |            | √          | √          | √          | √           |
-| 1.10.0            | √          | √          | √          | √          |             |
 
+Please run `pip list | grep torch` to get your `pytorch_version`.
 ```
 pip install oneccl_bind_pt=={pytorch_version} -f https://developer.intel.com/ipex-whl-stable-cpu
 ```
-where `{pytorch_version}` should be your PyTorch version, for instance 1.13.0.
+where `{pytorch_version}` should be your PyTorch version, for instance 2.1.0.
 Check more approaches for [oneccl_bind_pt installation](https://github.com/intel/torch-ccl).
 Versions of oneCCL and PyTorch must match.
 

docs/source/en/perf_train_gpu_many.md

@@ -577,3 +577,83 @@ for that. And then you can train. A different setup will have its own custom opt
 
 🤗 Transformers status: Transformers models are FX-trace-able via [transformers.utils.fx](https://github.com/huggingface/transformers/blob/master/src/transformers/utils/fx.py), 
 which is a prerequisite for FlexFlow, however, changes are required on the FlexFlow side to make it work with Transformers models.
+
+## GPU selection
+
+When training on multiple GPUs, you can specify the number of GPUs to use and in what order. This can be useful for instance when you have GPUs with different computing power and want to use the faster GPU first. The selection process works for both [DistributedDataParallel](https://pytorch.org/docs/stable/generated/torch.nn.parallel.DistributedDataParallel.html) and [DataParallel](https://pytorch.org/docs/stable/generated/torch.nn.DataParallel.html) to use only a subset of the available GPUs, and you don't need Accelerate or the [DeepSpeed integration](./main_classes/deepspeed).
+
+### Number of GPUs
+
+For example, if you have 4 GPUs and you only want to use the first 2:
+
+<hfoptions id="select-gpu">
+<hfoption id="torchrun">
+
+Use the `--nproc_per_node` to select how many GPUs to use.
+
+```bash
+torchrun --nproc_per_node=2  trainer-program.py ...
+```
+
+</hfoption>
+<hfoption id="Accelerate">
+
+Use `--num_processes` to select how many GPUs to use.
+
+```bash
+accelerate launch --num_processes 2 trainer-program.py ...
+```
+
+</hfoption>
+<hfoption id="DeepSpeed">
+
+Use `--num_gpus` to select how many GPUs to use.
+
+```bash
+deepspeed --num_gpus 2 trainer-program.py ...
+```
+
+</hfoption>
+</hfoptions>
+
+### Order of GPUs
+
+Now, to select which GPUs to use and their order, you'll use the `CUDA_VISIBLE_DEVICES` environment variable. It is easiest to set the environment variable in a `~/bashrc` or another startup config file. `CUDA_VISIBLE_DEVICES` is used to map which GPUs are used. For example, if you have 4 GPUs (0, 1, 2, 3) and you only want to run GPUs 0 and 2:
+
+```bash
+CUDA_VISIBLE_DEVICES=0,2 torchrun trainer-program.py ...
+```
+
+Only the 2 physical GPUs (0 and 2) are "visible" to PyTorch and these are mapped to `cuda:0` and `cuda:1` respectively. You can also reverse the order of the GPUs to use 2 first. Now, the mapping is `cuda:1` for GPU 0 and `cuda:0` for GPU 2.
+
+```bash
+CUDA_VISIBLE_DEVICES=2,0 torchrun trainer-program.py ...
+```
+
+You can also set the `CUDA_VISIBLE_DEVICES` environment variable to an empty value to create an environment without GPUs.
+
+```bash
+CUDA_VISIBLE_DEVICES= python trainer-program.py ...
+```
+
+<Tip warning={true}>
+
+As with any environment variable, they can be exported instead of being added to the command line. However, this is not recommended because it can be confusing if you forget how the environment variable was setup and you end up using the wrong GPUs. Instead, it is common practice to set the environment variable for a specific training run on the same command line.
+
+</Tip>
+
+`CUDA_DEVICE_ORDER` is an alternative environment variable you can use to control how the GPUs are ordered. You can either order them by:
+
+1. PCIe bus ID's that matches the order of [`nvidia-smi`](https://developer.nvidia.com/nvidia-system-management-interface) and [`rocm-smi`](https://rocm.docs.amd.com/projects/rocm_smi_lib/en/latest/.doxygen/docBin/html/index.html) for NVIDIA and AMD GPUs respectively
+
+```bash
+export CUDA_DEVICE_ORDER=PCI_BUS_ID
+```
+
+2. GPU compute ability
+
+```bash
+export CUDA_DEVICE_ORDER=FASTEST_FIRST
+```
+
+The `CUDA_DEVICE_ORDER` is especially useful if your training setup consists of an older and newer GPU, where the older GPU appears first, but you cannot physically swap the cards to make the newer GPU appear first. In this case, set `CUDA_DEVICE_ORDER=FASTEST_FIRST` to always use the newer and faster GPU first (`nvidia-smi` or `rocm-smi` still reports the GPUs in their PCIe order). Or you could also set `export CUDA_VISIBLE_DEVICES=1,0`.

docs/source/en/perf_train_gpu_one.md

@@ -51,7 +51,8 @@ The methods and tools covered in this guide can be classified based on the effec
 | [Data preloading](#data-preloading)                        | Yes                     | No                           |
 | [DeepSpeed Zero](#deepspeed-zero)                          | No                      | Yes                          |
 | [torch.compile](#using-torchcompile)                       | Yes                     | No                           |
-
+| [Parameter-Efficient Fine Tuning (PEFT)](#peft)            | No                      | Yes                          |
+ 
 <Tip>
 
 Note: when using mixed precision with a small model and a large batch size, there will be some memory savings but with a 
@@ -400,6 +401,25 @@ Choose which backend to use by specifying it via `torch_compile_backend` in the
 
 For an example of using `torch.compile` with 🤗 Transformers, check out this [blog post on fine-tuning a BERT model for Text Classification using the newest PyTorch 2.0 features](https://www.philschmid.de/getting-started-pytorch-2-0-transformers)
 
+## Using 🤗 PEFT
+
+[Parameter-Efficient Fine Tuning (PEFT)](https://huggingface.co/blog/peft) methods freeze the pretrained model parameters during fine-tuning and add a small number of trainable parameters (the adapters) on top of it.
+
+As a result the [memory associated to the optimizer states and gradients](https://huggingface.co/docs/transformers/model_memory_anatomy#anatomy-of-models-memory) are greatly reduced.
+
+For example with a vanilla AdamW, the memory requirement for the optimizer state would be:
+* fp32 copy of parameters: 4 bytes/param
+* Momentum: 4 bytes/param
+* Variance: 4 bytes/param
+
+Suppose a model with 7B parameters and 200 millions parameters injected with [Low Rank Adapters](https://huggingface.co/docs/peft/conceptual_guides/lora).
+
+The memory requirement for the optimizer state of the plain model would be 12 * 7 = 84 GB (assuming 7B trainable parameters).
+
+Adding Lora increases slightly the memory associated to the model weights and substantially decreases memory requirement for the optimizer state to 12 * 0.2 = 2.4GB.
+
+Read more about PEFT and its detailed usage in [the PEFT documentation](https://huggingface.co/docs/peft/) or [PEFT repository](https://github.com/huggingface/peft).
+
 ## Using 🤗 Accelerate
 
 With [🤗 Accelerate](https://huggingface.co/docs/accelerate/index) you can use the above methods while gaining full 

docs/source/en/perf_train_special.md

@@ -13,12 +13,51 @@ rendered properly in your Markdown viewer.
 
 -->
 
-# Training on Specialized Hardware
+# PyTorch training on Apple silicon
 
-<Tip>
+Previously, training models on a Mac was limited to the CPU only. With the release of PyTorch v1.12, you can take advantage of training models with Apple's silicon GPUs for significantly faster performance and training. This is powered in PyTorch by integrating Apple's Metal Performance Shaders (MPS) as a backend. The [MPS backend](https://pytorch.org/docs/stable/notes/mps.html) implements PyTorch operations as custom Metal shaders and places these modules on a `mps` device.
 
- Note: Most of the strategies introduced in the [single GPU section](perf_train_gpu_one) (such as mixed precision training or gradient accumulation) and [multi-GPU section](perf_train_gpu_many) are generic and apply to training models in general so make sure to have a look at it before diving into this section.
+<Tip warning={true}>
+
+Some PyTorch operations are not implemented in MPS yet and will throw an error. To avoid this, you should set the environment variable `PYTORCH_ENABLE_MPS_FALLBACK=1` to use the CPU kernels instead (you'll still see a `UserWarning`).
+
+<br>
+
+If you run into any other errors, please open an issue in the [PyTorch](https://github.com/pytorch/pytorch/issues) repository because the [`Trainer`] only integrates the MPS backend.
 
 </Tip>
 
-This document will be completed soon with information on how to train on specialized hardware.
+With the `mps` device set, you can:
+
+* train larger networks or batch sizes locally
+* reduce data retrieval latency because the GPU's unified memory architecture allows direct access to the full memory store
+* reduce costs because you don't need to train on cloud-based GPUs or add additional local GPUs
+
+Get started by making sure you have PyTorch installed. MPS acceleration is supported on macOS 12.3+.
+
+```bash
+pip install torch torchvision torchaudio
+```
+
+[`TrainingArguments`] uses the `mps` device by default if it's available which means you don't need to explicitly set the device. For example, you can run the [run_glue.py](https://github.com/huggingface/transformers/blob/main/examples/pytorch/text-classification/run_glue.py) script with the MPS backend automatically enabled without making any changes.
+
+```diff
+export TASK_NAME=mrpc
+
+python examples/pytorch/text-classification/run_glue.py \
+  --model_name_or_path bert-base-cased \
+  --task_name $TASK_NAME \
+- --use_mps_device \
+  --do_train \
+  --do_eval \
+  --max_seq_length 128 \
+  --per_device_train_batch_size 32 \
+  --learning_rate 2e-5 \
+  --num_train_epochs 3 \
+  --output_dir /tmp/$TASK_NAME/ \
+  --overwrite_output_dir
+```
+
+Backends for [distributed setups](https://pytorch.org/docs/stable/distributed.html#backends) like `gloo` and `nccl` are not supported by the `mps` device which means you can only train on a single GPU with the MPS backend.
+
+You can learn more about the MPS backend in the [Introducing Accelerated PyTorch Training on Mac](https://pytorch.org/blog/introducing-accelerated-pytorch-training-on-mac/) blog post.

docs/source/en/perf_train_tpu.md

@@ -1,24 +0,0 @@
-<!--Copyright 2022 The HuggingFace Team. All rights reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
-the License. You may obtain a copy of the License at
-
-http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
-an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
-
-⚠️ Note that this file is in Markdown but contain specific syntax for our doc-builder (similar to MDX) that may not be
-rendered properly in your Markdown viewer.
-
--->
-
-# Training on TPUs
-
-<Tip>
-
- Note: Most of the strategies introduced in the [single GPU section](perf_train_gpu_one) (such as mixed precision training or gradient accumulation) and [multi-GPU section](perf_train_gpu_many) are generic and apply to training models in general so make sure to have a look at it before diving into this section.
-
-</Tip>
-
-This document will be completed soon with information on how to train on TPUs.

docs/source/en/preprocessing.md

@@ -216,6 +216,12 @@ array([[1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0],
 </tf>
 </frameworkcontent>
 
+<Tip>
+Different pipelines support tokenizer arguments in their `__call__()` differently. `text-2-text-generation` pipelines support (i.e. pass on)
+only `truncation`. `text-generation` pipelines support `max_length`, `truncation`, `padding` and `add_special_tokens`. 
+In `fill-mask` pipelines, tokenizer arguments can be passed in the `tokenizer_kwargs` argument (dictionary).
+</Tip>
+
 ## Audio
 
 For audio tasks, you'll need a [feature extractor](main_classes/feature_extractor) to prepare your dataset for the model. The feature extractor is designed to extract features from raw audio data, and convert them into tensors.

docs/source/en/quantization.md

@@ -345,7 +345,7 @@ model_4bit = AutoModelForCausalLM.from_pretrained("facebook/opt-350m", load_in_4
 model_4bit.model.decoder.layers[-1].final_layer_norm.weight.dtype
 ```
 
-Once a model is quantized to 4-bit, you can't push the quantized weights to the Hub.
+If you have `bitsandbytes>=0.41.3`, you can serialize 4-bit models and push them on Hugging Face Hub. Simply call `model.push_to_hub()` after loading it in 4-bit precision. You can also save the serialized 4-bit models locally with `model.save_pretrained()` command.  
 
 </hfoption>
 </hfoptions>
@@ -468,6 +468,7 @@ Try 4-bit quantization in this [notebook](https://colab.research.google.com/driv
 
 This section explores some of the specific features of 4-bit models, such as changing the compute data type, using the Normal Float 4 (NF4) data type, and using nested quantization.
 
+
 #### Compute data type
 
 To speedup computation, you can change the data type from float32 (the default value) to bf16 using the `bnb_4bit_compute_dtype` parameter in [`BitsAndBytesConfig`]:
@@ -548,6 +549,7 @@ The benchmarks indicate AWQ quantization is the fastest for inference, text gene
 The [TheBloke/Mistral-7B-OpenOrca-AWQ](https://huggingface.co/TheBloke/Mistral-7B-OpenOrca-AWQ) model was benchmarked with `batch_size=1` with and without fused modules.
 
 <figcaption class="text-center text-gray-500 text-lg">Unfused module</figcaption>
+
 |   Batch Size |   Prefill Length |   Decode Length |   Prefill tokens/s |   Decode tokens/s | Memory (VRAM)   |
 |-------------:|-----------------:|----------------:|-------------------:|------------------:|:----------------|
 |            1 |               32 |              32 |            60.0984 |           38.4537 | 4.50 GB (5.68%) |
@@ -559,6 +561,7 @@ The [TheBloke/Mistral-7B-OpenOrca-AWQ](https://huggingface.co/TheBloke/Mistral-7
 |            1 |             2048 |            2048 |          2927.33   |           35.2676 | 5.73 GB (7.23%) |
 
 <figcaption class="text-center text-gray-500 text-lg">Fused module</figcaption>
+
 |   Batch Size |   Prefill Length |   Decode Length |   Prefill tokens/s |   Decode tokens/s | Memory (VRAM)   |
 |-------------:|-----------------:|----------------:|-------------------:|------------------:|:----------------|
 |            1 |               32 |              32 |            81.4899 |           80.2569 | 4.00 GB (5.05%) |

docs/source/en/task_summary.md

@@ -182,7 +182,7 @@ Like classification tasks in any modality, text classification labels a sequence
 
 ### Token classification
 
-In any NLP task, text is preprocessed by separating the sequence of text into individual words or subwords. These are known as [tokens](/glossary#token). Token classification assigns each token a label from a predefined set of classes. 
+In any NLP task, text is preprocessed by separating the sequence of text into individual words or subwords. These are known as [tokens](glossary#token). Token classification assigns each token a label from a predefined set of classes. 
 
 Two common types of token classification are:
 

docs/source/en/tasks/asr.md

@@ -32,7 +32,7 @@ The task illustrated in this tutorial is supported by the following model archit
 
 <!--This tip is automatically generated by `make fix-copies`, do not fill manually!-->
 
-[Data2VecAudio](../model_doc/data2vec-audio), [Hubert](../model_doc/hubert), [M-CTC-T](../model_doc/mctct), [SEW](../model_doc/sew), [SEW-D](../model_doc/sew-d), [UniSpeech](../model_doc/unispeech), [UniSpeechSat](../model_doc/unispeech-sat), [Wav2Vec2](../model_doc/wav2vec2), [Wav2Vec2-Conformer](../model_doc/wav2vec2-conformer), [WavLM](../model_doc/wavlm)
+[Data2VecAudio](../model_doc/data2vec-audio), [Hubert](../model_doc/hubert), [M-CTC-T](../model_doc/mctct), [SEW](../model_doc/sew), [SEW-D](../model_doc/sew-d), [UniSpeech](../model_doc/unispeech), [UniSpeechSat](../model_doc/unispeech-sat), [Wav2Vec2](../model_doc/wav2vec2), [Wav2Vec2-BERT](../model_doc/wav2vec2-bert), [Wav2Vec2-Conformer](../model_doc/wav2vec2-conformer), [WavLM](../model_doc/wavlm)
 
 <!--End of the generated tip-->
 

docs/source/en/tasks/audio_classification.md

@@ -32,7 +32,7 @@ The task illustrated in this tutorial is supported by the following model archit
 
 <!--This tip is automatically generated by `make fix-copies`, do not fill manually!-->
 
-[Audio Spectrogram Transformer](../model_doc/audio-spectrogram-transformer), [Data2VecAudio](../model_doc/data2vec-audio), [Hubert](../model_doc/hubert), [SEW](../model_doc/sew), [SEW-D](../model_doc/sew-d), [UniSpeech](../model_doc/unispeech), [UniSpeechSat](../model_doc/unispeech-sat), [Wav2Vec2](../model_doc/wav2vec2), [Wav2Vec2-Conformer](../model_doc/wav2vec2-conformer), [WavLM](../model_doc/wavlm), [Whisper](../model_doc/whisper)
+[Audio Spectrogram Transformer](../model_doc/audio-spectrogram-transformer), [Data2VecAudio](../model_doc/data2vec-audio), [Hubert](../model_doc/hubert), [SEW](../model_doc/sew), [SEW-D](../model_doc/sew-d), [UniSpeech](../model_doc/unispeech), [UniSpeechSat](../model_doc/unispeech-sat), [Wav2Vec2](../model_doc/wav2vec2), [Wav2Vec2-BERT](../model_doc/wav2vec2-bert), [Wav2Vec2-Conformer](../model_doc/wav2vec2-conformer), [WavLM](../model_doc/wavlm), [Whisper](../model_doc/whisper)
 
 <!--End of the generated tip-->
 

docs/source/en/tasks/language_modeling.md

@@ -37,7 +37,7 @@ You can finetune other architectures for causal language modeling following the
 Choose one of the following architectures:
 
 <!--This tip is automatically generated by `make fix-copies`, do not fill manually!-->
-[BART](../model_doc/bart), [BERT](../model_doc/bert), [Bert Generation](../model_doc/bert-generation), [BigBird](../model_doc/big_bird), [BigBird-Pegasus](../model_doc/bigbird_pegasus), [BioGpt](../model_doc/biogpt), [Blenderbot](../model_doc/blenderbot), [BlenderbotSmall](../model_doc/blenderbot-small), [BLOOM](../model_doc/bloom), [CamemBERT](../model_doc/camembert), [CodeLlama](../model_doc/code_llama), [CodeGen](../model_doc/codegen), [CPM-Ant](../model_doc/cpmant), [CTRL](../model_doc/ctrl), [Data2VecText](../model_doc/data2vec-text), [ELECTRA](../model_doc/electra), [ERNIE](../model_doc/ernie), [Falcon](../model_doc/falcon), [Fuyu](../model_doc/fuyu), [GIT](../model_doc/git), [GPT-Sw3](../model_doc/gpt-sw3), [OpenAI GPT-2](../model_doc/gpt2), [GPTBigCode](../model_doc/gpt_bigcode), [GPT Neo](../model_doc/gpt_neo), [GPT NeoX](../model_doc/gpt_neox), [GPT NeoX Japanese](../model_doc/gpt_neox_japanese), [GPT-J](../model_doc/gptj), [LLaMA](../model_doc/llama), [Marian](../model_doc/marian), [mBART](../model_doc/mbart), [MEGA](../model_doc/mega), [Megatron-BERT](../model_doc/megatron-bert), [Mistral](../model_doc/mistral), [Mixtral](../model_doc/mixtral), [MPT](../model_doc/mpt), [MusicGen](../model_doc/musicgen), [MVP](../model_doc/mvp), [OpenLlama](../model_doc/open-llama), [OpenAI GPT](../model_doc/openai-gpt), [OPT](../model_doc/opt), [Pegasus](../model_doc/pegasus), [Persimmon](../model_doc/persimmon), [Phi](../model_doc/phi), [PLBart](../model_doc/plbart), [ProphetNet](../model_doc/prophetnet), [QDQBert](../model_doc/qdqbert), [Reformer](../model_doc/reformer), [RemBERT](../model_doc/rembert), [RoBERTa](../model_doc/roberta), [RoBERTa-PreLayerNorm](../model_doc/roberta-prelayernorm), [RoCBert](../model_doc/roc_bert), [RoFormer](../model_doc/roformer), [RWKV](../model_doc/rwkv), [Speech2Text2](../model_doc/speech_to_text_2), [Transformer-XL](../model_doc/transfo-xl), [TrOCR](../model_doc/trocr), [Whisper](../model_doc/whisper), [XGLM](../model_doc/xglm), [XLM](../model_doc/xlm), [XLM-ProphetNet](../model_doc/xlm-prophetnet), [XLM-RoBERTa](../model_doc/xlm-roberta), [XLM-RoBERTa-XL](../model_doc/xlm-roberta-xl), [XLNet](../model_doc/xlnet), [X-MOD](../model_doc/xmod)
+[BART](../model_doc/bart), [BERT](../model_doc/bert), [Bert Generation](../model_doc/bert-generation), [BigBird](../model_doc/big_bird), [BigBird-Pegasus](../model_doc/bigbird_pegasus), [BioGpt](../model_doc/biogpt), [Blenderbot](../model_doc/blenderbot), [BlenderbotSmall](../model_doc/blenderbot-small), [BLOOM](../model_doc/bloom), [CamemBERT](../model_doc/camembert), [CodeLlama](../model_doc/code_llama), [CodeGen](../model_doc/codegen), [CPM-Ant](../model_doc/cpmant), [CTRL](../model_doc/ctrl), [Data2VecText](../model_doc/data2vec-text), [ELECTRA](../model_doc/electra), [ERNIE](../model_doc/ernie), [Falcon](../model_doc/falcon), [Fuyu](../model_doc/fuyu), [GIT](../model_doc/git), [GPT-Sw3](../model_doc/gpt-sw3), [OpenAI GPT-2](../model_doc/gpt2), [GPTBigCode](../model_doc/gpt_bigcode), [GPT Neo](../model_doc/gpt_neo), [GPT NeoX](../model_doc/gpt_neox), [GPT NeoX Japanese](../model_doc/gpt_neox_japanese), [GPT-J](../model_doc/gptj), [LLaMA](../model_doc/llama), [Marian](../model_doc/marian), [mBART](../model_doc/mbart), [MEGA](../model_doc/mega), [Megatron-BERT](../model_doc/megatron-bert), [Mistral](../model_doc/mistral), [Mixtral](../model_doc/mixtral), [MPT](../model_doc/mpt), [MusicGen](../model_doc/musicgen), [MVP](../model_doc/mvp), [OpenLlama](../model_doc/open-llama), [OpenAI GPT](../model_doc/openai-gpt), [OPT](../model_doc/opt), [Pegasus](../model_doc/pegasus), [Persimmon](../model_doc/persimmon), [Phi](../model_doc/phi), [PLBart](../model_doc/plbart), [ProphetNet](../model_doc/prophetnet), [QDQBert](../model_doc/qdqbert), [Qwen2](../model_doc/qwen2), [Reformer](../model_doc/reformer), [RemBERT](../model_doc/rembert), [RoBERTa](../model_doc/roberta), [RoBERTa-PreLayerNorm](../model_doc/roberta-prelayernorm), [RoCBert](../model_doc/roc_bert), [RoFormer](../model_doc/roformer), [RWKV](../model_doc/rwkv), [Speech2Text2](../model_doc/speech_to_text_2), [Transformer-XL](../model_doc/transfo-xl), [TrOCR](../model_doc/trocr), [Whisper](../model_doc/whisper), [XGLM](../model_doc/xglm), [XLM](../model_doc/xlm), [XLM-ProphetNet](../model_doc/xlm-prophetnet), [XLM-RoBERTa](../model_doc/xlm-roberta), [XLM-RoBERTa-XL](../model_doc/xlm-roberta-xl), [XLNet](../model_doc/xlnet), [X-MOD](../model_doc/xmod)
 
 
 

docs/source/en/tasks/object_detection.md

@@ -139,6 +139,17 @@ To get an even better understanding of the data, visualize an example in the dat
 ...     box = annotations["bbox"][i]
 ...     class_idx = annotations["category"][i]
 ...     x, y, w, h = tuple(box)
+...     # Check if coordinates are normalized or not
+...     if max(box) > 1.0:
+...         # Coordinates are un-normalized, no need to re-scale them
+...         x1, y1 = int(x), int(y)
+...         x2, y2 = int(x + w), int(y + h)
+...     else:
+...         # Coordinates are normalized, re-scale them
+...         x1 = int(x * width)
+...         y1 = int(y * height)
+...         x2 = int((x + w) * width)
+...         y2 = int((y + h) * height)
 ...     draw.rectangle((x, y, x + w, y + h), outline="red", width=1)
 ...     draw.text((x, y), id2label[class_idx], fill="white")
 
@@ -153,7 +164,7 @@ To visualize the bounding boxes with associated labels, you can get the labels f
 the `category` field.
 You'll also want to create dictionaries that map a label id to a label class (`id2label`) and the other way around (`label2id`).
 You can use them later when setting up the model. Including these maps will make your model reusable by others if you share
-it on the Hugging Face Hub.
+it on the Hugging Face Hub. Please note that, the part of above code that draws the bounding boxes assume that it is in `XYWH` (x,y co-ordinates and width and height of the box) format. It might not work for other formats like `(x1, y1, x2, y2)`.  
 
 As a final step of getting familiar with the data, explore it for potential issues. One common problem with datasets for
 object detection is bounding boxes that "stretch" beyond the edge of the image. Such "runaway" bounding boxes can raise

docs/source/en/tasks/semantic_segmentation.md

@@ -276,8 +276,7 @@ You could also create and use your own dataset if you prefer to train with the [
                                      "label": sorted(label_paths)})
          dataset = dataset.cast_column("image", Image())
          dataset = dataset.cast_column("label", Image())
-
-     return dataset
+         return dataset
 
      # step 1: create Dataset objects
      train_dataset = create_dataset(image_paths_train, label_paths_train)

docs/source/en/tasks/sequence_classification.md

@@ -33,7 +33,7 @@ The task illustrated in this tutorial is supported by the following model archit
 <!--This tip is automatically generated by `make fix-copies`, do not fill manually!-->
 
 
-[ALBERT](../model_doc/albert), [BART](../model_doc/bart), [BERT](../model_doc/bert), [BigBird](../model_doc/big_bird), [BigBird-Pegasus](../model_doc/bigbird_pegasus), [BioGpt](../model_doc/biogpt), [BLOOM](../model_doc/bloom), [CamemBERT](../model_doc/camembert), [CANINE](../model_doc/canine), [CodeLlama](../model_doc/code_llama), [ConvBERT](../model_doc/convbert), [CTRL](../model_doc/ctrl), [Data2VecText](../model_doc/data2vec-text), [DeBERTa](../model_doc/deberta), [DeBERTa-v2](../model_doc/deberta-v2), [DistilBERT](../model_doc/distilbert), [ELECTRA](../model_doc/electra), [ERNIE](../model_doc/ernie), [ErnieM](../model_doc/ernie_m), [ESM](../model_doc/esm), [Falcon](../model_doc/falcon), [FlauBERT](../model_doc/flaubert), [FNet](../model_doc/fnet), [Funnel Transformer](../model_doc/funnel), [GPT-Sw3](../model_doc/gpt-sw3), [OpenAI GPT-2](../model_doc/gpt2), [GPTBigCode](../model_doc/gpt_bigcode), [GPT Neo](../model_doc/gpt_neo), [GPT NeoX](../model_doc/gpt_neox), [GPT-J](../model_doc/gptj), [I-BERT](../model_doc/ibert), [LayoutLM](../model_doc/layoutlm), [LayoutLMv2](../model_doc/layoutlmv2), [LayoutLMv3](../model_doc/layoutlmv3), [LED](../model_doc/led), [LiLT](../model_doc/lilt), [LLaMA](../model_doc/llama), [Longformer](../model_doc/longformer), [LUKE](../model_doc/luke), [MarkupLM](../model_doc/markuplm), [mBART](../model_doc/mbart), [MEGA](../model_doc/mega), [Megatron-BERT](../model_doc/megatron-bert), [Mistral](../model_doc/mistral), [Mixtral](../model_doc/mixtral), [MobileBERT](../model_doc/mobilebert), [MPNet](../model_doc/mpnet), [MPT](../model_doc/mpt), [MRA](../model_doc/mra), [MT5](../model_doc/mt5), [MVP](../model_doc/mvp), [Nezha](../model_doc/nezha), [Nyströmformer](../model_doc/nystromformer), [OpenLlama](../model_doc/open-llama), [OpenAI GPT](../model_doc/openai-gpt), [OPT](../model_doc/opt), [Perceiver](../model_doc/perceiver), [Persimmon](../model_doc/persimmon), [Phi](../model_doc/phi), [PLBart](../model_doc/plbart), [QDQBert](../model_doc/qdqbert), [Reformer](../model_doc/reformer), [RemBERT](../model_doc/rembert), [RoBERTa](../model_doc/roberta), [RoBERTa-PreLayerNorm](../model_doc/roberta-prelayernorm), [RoCBert](../model_doc/roc_bert), [RoFormer](../model_doc/roformer), [SqueezeBERT](../model_doc/squeezebert), [T5](../model_doc/t5), [TAPAS](../model_doc/tapas), [Transformer-XL](../model_doc/transfo-xl), [UMT5](../model_doc/umt5), [XLM](../model_doc/xlm), [XLM-RoBERTa](../model_doc/xlm-roberta), [XLM-RoBERTa-XL](../model_doc/xlm-roberta-xl), [XLNet](../model_doc/xlnet), [X-MOD](../model_doc/xmod), [YOSO](../model_doc/yoso)
+[ALBERT](../model_doc/albert), [BART](../model_doc/bart), [BERT](../model_doc/bert), [BigBird](../model_doc/big_bird), [BigBird-Pegasus](../model_doc/bigbird_pegasus), [BioGpt](../model_doc/biogpt), [BLOOM](../model_doc/bloom), [CamemBERT](../model_doc/camembert), [CANINE](../model_doc/canine), [CodeLlama](../model_doc/code_llama), [ConvBERT](../model_doc/convbert), [CTRL](../model_doc/ctrl), [Data2VecText](../model_doc/data2vec-text), [DeBERTa](../model_doc/deberta), [DeBERTa-v2](../model_doc/deberta-v2), [DistilBERT](../model_doc/distilbert), [ELECTRA](../model_doc/electra), [ERNIE](../model_doc/ernie), [ErnieM](../model_doc/ernie_m), [ESM](../model_doc/esm), [Falcon](../model_doc/falcon), [FlauBERT](../model_doc/flaubert), [FNet](../model_doc/fnet), [Funnel Transformer](../model_doc/funnel), [GPT-Sw3](../model_doc/gpt-sw3), [OpenAI GPT-2](../model_doc/gpt2), [GPTBigCode](../model_doc/gpt_bigcode), [GPT Neo](../model_doc/gpt_neo), [GPT NeoX](../model_doc/gpt_neox), [GPT-J](../model_doc/gptj), [I-BERT](../model_doc/ibert), [LayoutLM](../model_doc/layoutlm), [LayoutLMv2](../model_doc/layoutlmv2), [LayoutLMv3](../model_doc/layoutlmv3), [LED](../model_doc/led), [LiLT](../model_doc/lilt), [LLaMA](../model_doc/llama), [Longformer](../model_doc/longformer), [LUKE](../model_doc/luke), [MarkupLM](../model_doc/markuplm), [mBART](../model_doc/mbart), [MEGA](../model_doc/mega), [Megatron-BERT](../model_doc/megatron-bert), [Mistral](../model_doc/mistral), [Mixtral](../model_doc/mixtral), [MobileBERT](../model_doc/mobilebert), [MPNet](../model_doc/mpnet), [MPT](../model_doc/mpt), [MRA](../model_doc/mra), [MT5](../model_doc/mt5), [MVP](../model_doc/mvp), [Nezha](../model_doc/nezha), [Nyströmformer](../model_doc/nystromformer), [OpenLlama](../model_doc/open-llama), [OpenAI GPT](../model_doc/openai-gpt), [OPT](../model_doc/opt), [Perceiver](../model_doc/perceiver), [Persimmon](../model_doc/persimmon), [Phi](../model_doc/phi), [PLBart](../model_doc/plbart), [QDQBert](../model_doc/qdqbert), [Qwen2](../model_doc/qwen2), [Reformer](../model_doc/reformer), [RemBERT](../model_doc/rembert), [RoBERTa](../model_doc/roberta), [RoBERTa-PreLayerNorm](../model_doc/roberta-prelayernorm), [RoCBert](../model_doc/roc_bert), [RoFormer](../model_doc/roformer), [SqueezeBERT](../model_doc/squeezebert), [T5](../model_doc/t5), [TAPAS](../model_doc/tapas), [Transformer-XL](../model_doc/transfo-xl), [UMT5](../model_doc/umt5), [XLM](../model_doc/xlm), [XLM-RoBERTa](../model_doc/xlm-roberta), [XLM-RoBERTa-XL](../model_doc/xlm-roberta-xl), [XLNet](../model_doc/xlnet), [X-MOD](../model_doc/xmod), [YOSO](../model_doc/yoso)
 
 
 

docs/source/en/tasks/text-to-speech.md

@@ -44,10 +44,8 @@ Here's a code snippet you can use to listen to the resulting audio in a notebook
 For more examples on what Bark and other pretrained TTS models can do, refer to our 
 [Audio course](https://huggingface.co/learn/audio-course/chapter6/pre-trained_models). 
 
-If you are looking to fine-tune a TTS model, you can currently fine-tune SpeechT5 only. SpeechT5 is pre-trained on a combination of 
-speech-to-text and text-to-speech data, allowing it to learn a unified space of hidden representations shared by both text 
-and speech. This means that the same pre-trained model can be fine-tuned for different tasks. Furthermore, SpeechT5 
-supports multiple speakers through x-vector speaker embeddings. 
+If you are looking to fine-tune a TTS model, the only text-to-speech models currently available in 🤗 Transformers 
+are [SpeechT5](model_doc/speecht5) and [FastSpeech2Conformer](model_doc/fastspeech2_conformer), though more will be added in the future. SpeechT5 is pre-trained on a combination of speech-to-text and text-to-speech data, allowing it to learn a unified space of hidden representations shared by both text and speech. This means that the same pre-trained model can be fine-tuned for different tasks. Furthermore, SpeechT5 supports multiple speakers through x-vector speaker embeddings. 
 
 The remainder of this guide illustrates how to:
 

docs/source/en/tasks/zero_shot_object_detection.md

@@ -299,11 +299,3 @@ as before except now there are no labels.
      <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers/tasks/zero-sh-obj-detection_6.png" alt="Cats with bounding boxes"/>
 </div>
 
-If you'd like to interactively try out inference with OWL-ViT, check out this demo:
-
-<iframe
-	src="https://adirik-owl-vit.hf.space"
-	frameborder="0"
-	width="850"
-	height="450"
-></iframe>

docs/source/en/trainer.md

@@ -0,0 +1,414 @@
+<!--Copyright 2023 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+
+⚠️ Note that this file is in Markdown but contain specific syntax for our doc-builder (similar to MDX) that may not be
+rendered properly in your Markdown viewer.
+
+-->
+
+# Trainer
+
+The [`Trainer`] is a complete training and evaluation loop for PyTorch models implemented in the Transformers library. You only need to pass it the necessary pieces for training (model, tokenizer, dataset, evaluation function, training hyperparameters, etc.), and the [`Trainer`] class takes care of the rest. This makes it easier to start training faster without manually writing your own training loop. But at the same time, [`Trainer`] is very customizable and offers a ton of training options so you can tailor it to your exact training needs.
+
+<Tip>
+
+In addition to the [`Trainer`] class, Transformers also provides a [`Seq2SeqTrainer`] class for sequence-to-sequence tasks like translation or summarization. There is also the [`~trl.SFTTrainer`] class from the [TRL](https://hf.co/docs/trl) library which wraps the [`Trainer`] class and is optimized for training language models like Llama-2 and Mistral with autoregressive techniques. [`~trl.SFTTrainer`] also supports features like sequence packing, LoRA, quantization, and DeepSpeed for efficiently scaling to any model size.
+
+<br>
+
+Feel free to check out the [API reference](./main_classes/trainer) for these other [`Trainer`]-type classes to learn more about when to use which one. In general, [`Trainer`] is the most versatile option and is appropriate for a broad spectrum of tasks. [`Seq2SeqTrainer`] is designed for sequence-to-sequence tasks and [`~trl.SFTTrainer`] is designed for training language models.
+
+</Tip>
+
+Before you start, make sure [Accelerate](https://hf.co/docs/accelerate) - a library for enabling and running PyTorch training across distributed environments - is installed.
+
+```bash
+pip install accelerate
+
+# upgrade
+pip install accelerate --upgrade
+```
+
+This guide provides an overview of the [`Trainer`] class.
+
+## Basic usage
+
+[`Trainer`] includes all the code you'll find in a basic training loop:
+
+1. perform a training step to calculate the loss
+2. calculate the gradients with the [`~accelerate.Accelerator.backward`] method
+3. update the weights based on the gradients
+4. repeat this process until you've reached a predetermined number of epochs
+
+The [`Trainer`] class abstracts all of this code away so you don't have to worry about manually writing a training loop every time or if you're just getting started with PyTorch and training. You only need to provide the essential components required for training, such as a model and a dataset, and the [`Trainer`] class handles everything else.
+
+If you want to specify any training options or hyperparameters, you can find them in the [`TrainingArguments`] class. For example, let's define where to save the model in `output_dir` and push the model to the Hub after training with `push_to_hub=True`.
+
+```py
+from transformers import TrainingArguments
+
+training_args = TrainingArguments(
+    output_dir="your-model",
+    learning_rate=2e-5,
+    per_device_train_batch_size=16,
+    per_device_eval_batch_size=16,
+    num_train_epochs=2,
+    weight_decay=0.01,
+    evaluation_strategy="epoch",
+    save_strategy="epoch",
+    load_best_model_at_end=True,
+    push_to_hub=True,
+)
+```
+
+Pass `training_args` to the [`Trainer`] along with a model, dataset, something to preprocess the dataset with (depending on your data type it could be a tokenizer, feature extractor or image processor), a data collator, and a function to compute the metrics you want to track during training.
+
+Finally, call [`~Trainer.train`] to start training!
+
+```py
+from transformers import Trainer
+
+trainer = Trainer(
+    model=model,
+    args=training_args,
+    train_dataset=dataset["train"],
+    eval_dataset=dataset["test"],
+    tokenizer=tokenizer,
+    data_collator=data_collator,
+    compute_metrics=compute_metrics,
+)
+
+trainer.train()
+```
+
+### Checkpoints
+
+The [`Trainer`] class saves your model checkpoints to the directory specified in the `output_dir` parameter of [`TrainingArguments`]. You'll find the checkpoints saved in a `checkpoint-000` subfolder where the numbers at the end correspond to the training step. Saving checkpoints are useful for resuming training later.
+
+```py
+# resume from latest checkpoint
+trainer.train(resume_from_checkpoint=True)
+
+# resume from specific checkpoint saved in output directory
+trainer.train(resume_from_checkpoint="your-model/checkpoint-1000")
+```
+
+You can save your checkpoints (the optimizer state is not saved by default) to the Hub by setting `push_to_hub=True` in [`TrainingArguments`] to commit and push them. Other options for deciding how your checkpoints are saved are set up in the [`hub_strategy`](https://huggingface.co/docs/transformers/main_classes/trainer#transformers.TrainingArguments.hub_strategy) parameter:
+
+* `hub_strategy="checkpoint"` pushes the latest checkpoint to a subfolder named "last-checkpoint" from which you can resume training
+* `hug_strategy="all_checkpoints"` pushes all checkpoints to the directory defined in `output_dir` (you'll see one checkpoint per folder in your model repository)
+
+When you resume training from a checkpoint, the [`Trainer`] tries to keep the Python, NumPy, and PyTorch RNG states the same as they were when the checkpoint was saved. But because PyTorch has various non-deterministic default settings, the RNG states aren't guaranteed to be the same. If you want to enable full determinism, take a look at the [Controlling sources of randomness](https://pytorch.org/docs/stable/notes/randomness#controlling-sources-of-randomness) guide to learn what you can enable to make your training fully deterministic. Keep in mind though that by making certain settings deterministic, training may be slower.
+
+## Customize the Trainer
+
+While the [`Trainer`] class is designed to be accessible and easy-to-use, it also offers a lot of customizability for more adventurous users. Many of the [`Trainer`]'s method can be subclassed and overridden to support the functionality you want, without having to rewrite the entire training loop from scratch to accommodate it. These methods include:
+
+* [`~Trainer.get_train_dataloader`] creates a training DataLoader
+* [`~Trainer.get_eval_dataloader`] creates an evaluation DataLoader
+* [`~Trainer.get_test_dataloader`] creates a test DataLoader
+* [`~Trainer.log`] logs information on the various objects that watch training
+* [`~Trainer.create_optimizer_and_scheduler`] creates an optimizer and learning rate scheduler if they weren't passed in the `__init__`; these can also be separately customized with [`~Trainer.create_optimizer`] and [`~Trainer.create_scheduler`] respectively
+* [`~Trainer.compute_loss`] computes the loss on a batch of training inputs
+* [`~Trainer.training_step`] performs the training step
+* [`~Trainer.prediction_step`] performs the prediction and test step
+* [`~Trainer.evaluate`] evaluates the model and returns the evaluation metrics
+* [`~Trainer.predict`] makes predictions (with metrics if labels are available) on the test set
+
+For example, if you want to customize the [`~Trainer.compute_loss`] method to use a weighted loss instead.
+
+```py
+from torch import nn
+from transformers import Trainer
+
+class CustomTrainer(Trainer):
+    def compute_loss(self, model, inputs, return_outputs=False):
+        labels = inputs.pop("labels")
+        # forward pass
+        outputs = model(**inputs)
+        logits = outputs.get("logits")
+        # compute custom loss for 3 labels with different weights
+        loss_fct = nn.CrossEntropyLoss(weight=torch.tensor([1.0, 2.0, 3.0], device=model.device))
+        loss = loss_fct(logits.view(-1, self.model.config.num_labels), labels.view(-1))
+        return (loss, outputs) if return_outputs else loss
+```
+
+### Callbacks
+
+Another option for customizing the [`Trainer`] is to use [callbacks](callbacks). Callbacks *don't change* anything in the training loop. They inspect the training loop state and then execute some action (early stopping, logging results, etc.) depending on the state. In other words, a callback can't be used to implement something like a custom loss function and you'll need to subclass and override the [`~Trainer.compute_loss`] method for that.
+
+For example, if you want to add an early stopping callback to the training loop after 10 steps.
+
+```py
+from transformers import TrainerCallback
+
+class EarlyStoppingCallback(TrainerCallback):
+    def __init__(self, num_steps=10):
+        self.num_steps = num_steps
+    
+    def on_step_end(self, args, state, control, **kwargs):
+        if state.global_step >= self.num_steps:
+            return {"should_training_stop": True}
+        else:
+            return {}
+```
+
+Then pass it to the [`Trainer`]'s `callback` parameter.
+
+```py
+from transformers import Trainer
+
+trainer = Trainer(
+    model=model,
+    args=training_args,
+    train_dataset=dataset["train"],
+    eval_dataset=dataset["test"],
+    tokenizer=tokenizer,
+    data_collator=data_collator,
+    compute_metrics=compute_metrics,
+    callback=[EarlyStoppingCallback()],
+)
+```
+
+## Logging
+
+<Tip>
+
+Check out the [logging](./main_classes/logging) API reference for more information about the different logging levels.
+
+</Tip>
+
+The [`Trainer`] is set to `logging.INFO` by default which reports errors, warnings, and other basic information. A [`Trainer`] replica - in distributed environments - is set to `logging.WARNING` which only reports errors and warnings. You can change the logging level with the [`log_level`](https://huggingface.co/docs/transformers/main_classes/trainer#transformers.TrainingArguments.log_level) and [`log_level_replica`](https://huggingface.co/docs/transformers/main_classes/trainer#transformers.TrainingArguments.log_level_replica) parameters in [`TrainingArguments`].
+
+To configure the log level setting for each node, use the [`log_on_each_node`](https://huggingface.co/docs/transformers/main/en/main_classes/trainer#transformers.TrainingArguments.log_on_each_node) parameter to determine whether to use the log level on each node or only on the main node.
+
+<Tip>
+
+[`Trainer`] sets the log level separately for each node in the [`Trainer.__init__`] method, so you may want to consider setting this sooner if you're using other Transformers functionalities before creating the [`Trainer`] object.
+
+</Tip>
+
+For example, to set your main code and modules to use the same log level according to each node:
+
+```py
+logger = logging.getLogger(__name__)
+
+logging.basicConfig(
+    format="%(asctime)s - %(levelname)s - %(name)s - %(message)s",
+    datefmt="%m/%d/%Y %H:%M:%S",
+    handlers=[logging.StreamHandler(sys.stdout)],
+)
+
+log_level = training_args.get_process_log_level()
+logger.setLevel(log_level)
+datasets.utils.logging.set_verbosity(log_level)
+transformers.utils.logging.set_verbosity(log_level)
+
+trainer = Trainer(...)
+```
+
+Use different combinations of `log_level` and `log_level_replica` to configure what gets logged on each of the nodes.
+
+<hfoptions id="logging">
+<hfoption id="single node">
+
+```bash
+my_app.py ... --log_level warning --log_level_replica error
+```
+
+</hfoption>
+<hfoption id="multi-node">
+
+Add the `log_on_each_node 0` parameter for multi-node environments.
+
+```bash
+my_app.py ... --log_level warning --log_level_replica error --log_on_each_node 0
+
+# set to only report errors
+my_app.py ... --log_level error --log_level_replica error --log_on_each_node 0
+```
+
+</hfoption>
+</hfoptions>
+
+## NEFTune
+
+[NEFTune](https://hf.co/papers/2310.05914) is a technique that can improve performance by adding noise to the embedding vectors during training. To enable it in [`Trainer`], set the `neftune_noise_alpha` parameter in [`TrainingArguments`] to control how much noise is added.
+
+```py
+from transformers import TrainingArguments, Trainer
+
+training_args = TrainingArguments(..., neftune_noise_alpha=0.1)
+trainer = Trainer(..., args=training_args)
+```
+
+NEFTune is disabled after training to restore the original embedding layer to avoid any unexpected behavior.
+
+## Accelerate and Trainer
+
+The [`Trainer`] class is powered by [Accelerate](https://hf.co/docs/accelerate), a library for easily training PyTorch models in distributed environments with support for integrations such as [FullyShardedDataParallel (FSDP)](https://pytorch.org/blog/introducing-pytorch-fully-sharded-data-parallel-api/) and [DeepSpeed](https://www.deepspeed.ai/).
+
+<Tip>
+
+Learn more about FSDP sharding strategies, CPU offloading, and more with the [`Trainer`] in the [Fully Sharded Data Parallel](fsdp) guide.
+
+</Tip>
+
+To use Accelerate with [`Trainer`], run the [`accelerate.config`](https://huggingface.co/docs/accelerate/package_reference/cli#accelerate-config) command to set up training for your training environment. This command creates a `config_file.yaml` that'll be used when you launch your training script. For example, some example configurations you can setup are:
+
+<hfoptions id="config">
+<hfoption id="DistributedDataParallel">
+
+```yml
+compute_environment: LOCAL_MACHINE                                                                                             
+distributed_type: MULTI_GPU                                                                                                    
+downcast_bf16: 'no'
+gpu_ids: all
+machine_rank: 0 #change rank as per the node
+main_process_ip: 192.168.20.1
+main_process_port: 9898
+main_training_function: main
+mixed_precision: fp16
+num_machines: 2
+num_processes: 8
+rdzv_backend: static
+same_network: true
+tpu_env: []
+tpu_use_cluster: false
+tpu_use_sudo: false
+use_cpu: false
+```
+
+</hfoption>
+<hfoption id="FSDP">
+
+```yml
+compute_environment: LOCAL_MACHINE
+distributed_type: FSDP
+downcast_bf16: 'no'
+fsdp_config:
+  fsdp_auto_wrap_policy: TRANSFORMER_BASED_WRAP
+  fsdp_backward_prefetch_policy: BACKWARD_PRE
+  fsdp_forward_prefetch: true
+  fsdp_offload_params: false
+  fsdp_sharding_strategy: 1
+  fsdp_state_dict_type: FULL_STATE_DICT
+  fsdp_sync_module_states: true
+  fsdp_transformer_layer_cls_to_wrap: BertLayer
+  fsdp_use_orig_params: true
+machine_rank: 0
+main_training_function: main
+mixed_precision: bf16
+num_machines: 1
+num_processes: 2
+rdzv_backend: static
+same_network: true
+tpu_env: []
+tpu_use_cluster: false
+tpu_use_sudo: false
+use_cpu: false
+```
+
+</hfoption>
+<hfoption id="DeepSpeed">
+
+```yml
+compute_environment: LOCAL_MACHINE
+deepspeed_config:
+  deepspeed_config_file: /home/user/configs/ds_zero3_config.json
+  zero3_init_flag: true
+distributed_type: DEEPSPEED
+downcast_bf16: 'no'
+machine_rank: 0
+main_training_function: main
+num_machines: 1
+num_processes: 4
+rdzv_backend: static
+same_network: true
+tpu_env: []
+tpu_use_cluster: false
+tpu_use_sudo: false
+use_cpu: false
+```
+
+</hfoption>
+<hfoption id="DeepSpeed with Accelerate plugin">
+
+```yml
+compute_environment: LOCAL_MACHINE                                                                                             
+deepspeed_config:                                                                                                              
+  gradient_accumulation_steps: 1
+  gradient_clipping: 0.7
+  offload_optimizer_device: cpu
+  offload_param_device: cpu
+  zero3_init_flag: true
+  zero_stage: 2
+distributed_type: DEEPSPEED
+downcast_bf16: 'no'
+machine_rank: 0
+main_training_function: main
+mixed_precision: bf16
+num_machines: 1
+num_processes: 4
+rdzv_backend: static
+same_network: true
+tpu_env: []
+tpu_use_cluster: false
+tpu_use_sudo: false
+use_cpu: false
+```
+
+</hfoption>
+</hfoptions>
+
+The [`accelerate_launch`](https://huggingface.co/docs/accelerate/package_reference/cli#accelerate-launch) command is the recommended way to launch your training script on a distributed system with Accelerate and [`Trainer`] with the parameters specified in `config_file.yaml`. This file is saved to the Accelerate cache folder and automatically loaded when you run `accelerate_launch`.
+
+For example, to run the [run_glue.py](https://github.com/huggingface/transformers/blob/f4db565b695582891e43a5e042e5d318e28f20b8/examples/pytorch/text-classification/run_glue.py#L4) training script with the FSDP configuration:
+
+```bash
+accelerate launch \
+    ./examples/pytorch/text-classification/run_glue.py \
+    --model_name_or_path bert-base-cased \
+    --task_name $TASK_NAME \
+    --do_train \
+    --do_eval \
+    --max_seq_length 128 \
+    --per_device_train_batch_size 16 \
+    --learning_rate 5e-5 \
+    --num_train_epochs 3 \
+    --output_dir /tmp/$TASK_NAME/ \
+    --overwrite_output_dir
+```
+
+You could also specify the parameters from the `config_file.yaml` file directly in the command line:
+
+```bash
+accelerate launch --num_processes=2 \
+    --use_fsdp \
+    --mixed_precision=bf16 \
+    --fsdp_auto_wrap_policy=TRANSFORMER_BASED_WRAP  \
+    --fsdp_transformer_layer_cls_to_wrap="BertLayer" \
+    --fsdp_sharding_strategy=1 \
+    --fsdp_state_dict_type=FULL_STATE_DICT \
+    ./examples/pytorch/text-classification/run_glue.py
+    --model_name_or_path bert-base-cased \
+    --task_name $TASK_NAME \
+    --do_train \
+    --do_eval \
+    --max_seq_length 128 \
+    --per_device_train_batch_size 16 \
+    --learning_rate 5e-5 \
+    --num_train_epochs 3 \
+    --output_dir /tmp/$TASK_NAME/ \
+    --overwrite_output_dir
+```
+
+Check out the [Launching your Accelerate scripts](https://huggingface.co/docs/accelerate/basic_tutorials/launch) tutorial to learn more about `accelerate_launch` and custom configurations.

docs/source/es/_toctree.yml

@@ -21,57 +21,60 @@
     title: Compartir un modelo
   title: Tutoriales
 - sections:
-  - sections:
-    - local: create_a_model
-      title: Crea una arquitectura personalizada
-    - local: custom_models
-      title: Compartir modelos personalizados
-    - local: run_scripts
-      title: Entrenamiento con scripts
-    - local: sagemaker
-      title: Ejecutar el entrenamiento en Amazon SageMaker
-    - local: converting_tensorflow_models
-      title: Convertir checkpoints de TensorFlow
-    - local: serialization
-      title: Exportar a ONNX
-    title: Uso general
-  - sections:
-    - local: fast_tokenizers
-      title: Usa tokenizadores de 🤗 Tokenizers
-    - local: multilingual
-      title: Modelos multilingües para inferencia
-    - sections:
-      - local: tasks/question_answering
-        title: Respuesta a preguntas
-      - local: tasks/language_modeling
-        title: Modelado de lenguaje
-      - local: tasks/summarization
-        title: Generación de resúmenes
-      - local: tasks/multiple_choice
-        title: Selección múltiple
-      title: Guías de tareas
+  - isExpanded: false
+    sections:
+    - local: tasks/question_answering
+      title: Respuesta a preguntas
+    - local: tasks/language_modeling
+      title: Modelado de lenguaje
+    - local: tasks/summarization
+      title: Generación de resúmenes
+    - local: tasks/multiple_choice
+      title: Selección múltiple
     title: Procesamiento del Lenguaje Natural
-  - sections:
+  - isExpanded: false
+    sections:
     - local: tasks/asr
       title: Reconocimiento automático del habla
     title: Audio
-  - sections:
+  - isExpanded: false
+    sections:
     - local: tasks/image_classification
       title: Clasificación de imágenes
     title: Visión Artificial
-  - sections:
-    - local: debugging
-      title: Debugging
-    title: Rendimiento y escalabilidad
-  - sections:
-    - local: add_new_pipeline
-      title: ¿Cómo puedo añadir un pipeline a 🤗 Transformers?
-    - local: pr_checks
-      title: Verificaciones en un Pull Request
-    title: Contribuir
+  title: Guías prácticas
+- sections:
+  - local: fast_tokenizers
+    title: Usa tokenizadores de 🤗 Tokenizers
+  - local: multilingual
+    title: Modelos multilingües para inferencia
+  - local: create_a_model
+    title: Crea una arquitectura personalizada
+  - local: custom_models
+    title: Compartir modelos personalizados
+  - local: run_scripts
+    title: Entrenamiento con scripts
+  - local: sagemaker
+    title: Ejecutar el entrenamiento en Amazon SageMaker
+  - local: converting_tensorflow_models
+    title: Convertir checkpoints de TensorFlow
+  - local: serialization
+    title: Exportar a ONNX
   - local: community
     title: Los recursos de la comunidad
-  title: Guías prácticas
+  title: Guías para desarrolladores
+- sections:
+  - local: performance
+    title: Descripción general
+  - local: debugging
+    title: Debugging
+  title: Rendimiento y escalabilidad
+- sections:
+  - local: add_new_pipeline
+    title: ¿Cómo puedo añadir un pipeline a 🤗 Transformers?
+  - local: pr_checks
+    title: Verificaciones en un Pull Request
+  title: Contribuir
 - sections:
   - local: philosophy
     title: Filosofía

docs/source/es/add_new_pipeline.md

@@ -15,7 +15,7 @@ rendered properly in your Markdown viewer.
 
 # ¿Cómo puedo crear un pipeline personalizado?
 
-En esta guía, veremos cómo crear un pipeline personalizado y cómo compartirlo en el [Hub](hf.co/models) o añadirlo
+En esta guía, veremos cómo crear un pipeline personalizado y cómo compartirlo en el [Hub](https://hf.co/models) o añadirlo
 a la biblioteca 🤗 Transformers.
 
 En primer lugar, debes decidir las entradas que tu pipeline podrá recibir. Pueden ser strings, bytes,

docs/source/es/installation.md

@@ -131,10 +131,10 @@ El entorno de Python que creaste para la instalación de 🤗 Transformers encon
 
 ## Instalación con conda
 
-Puedes instalar 🤗 Transformers desde el canal de conda `huggingface` con el siguiente comando:
+Puedes instalar 🤗 Transformers desde el canal de conda `conda-forge` con el siguiente comando:
 
 ```bash
-conda install -c huggingface transformers
+conda install conda-forge::transformers
 ```
 
 ## Configuración de Caché
@@ -148,7 +148,7 @@ Los modelos preentrenados se descargan y almacenan en caché localmente en: `~/.
 <Tip>
 
 🤗 Transformers usará las variables de entorno de shell `PYTORCH_TRANSFORMERS_CACHE` o `PYTORCH_PRETRAINED_BERT_CACHE` si viene de una iteración anterior de la biblioteca y ha configurado esas variables de entorno, a menos que especifiques la variable de entorno de shell `TRANSFORMERS_CACHE`.
-    
+
 </Tip>
 
 
@@ -204,7 +204,7 @@ Otra opción para usar 🤗 Transformers offline es descargando previamente los
     >>> model.save_pretrained("./your/path/bigscience_t0")
     ```
 
-    3. Cuando te encuentres offline, recarga los archivos con [`PreTrainedModel.from_pretrained`] desde el directorio especificado: 
+    3. Cuando te encuentres offline, recarga los archivos con [`PreTrainedModel.from_pretrained`] desde el directorio especificado:
 
     ```py
     >>> tokenizer = AutoTokenizer.from_pretrained("./your/path/bigscience_t0")
@@ -213,7 +213,7 @@ Otra opción para usar 🤗 Transformers offline es descargando previamente los
 
 * Descarga de manera programática los archivos con la biblioteca [huggingface_hub](https://github.com/huggingface/huggingface_hub/tree/main/src/huggingface_hub):
 
-    1. Instala la biblioteca [huggingface_hub](https://github.com/huggingface/huggingface_hub/tree/main/src/huggingface_hub) en tu entorno virtual: 
+    1. Instala la biblioteca [huggingface_hub](https://github.com/huggingface/huggingface_hub/tree/main/src/huggingface_hub) en tu entorno virtual:
 
     ```bash
     python -m pip install huggingface_hub

docs/source/es/performance.md

@@ -0,0 +1,61 @@
+<!---
+Copyright 2021 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+
+⚠️ Note that this file is in Markdown but contain specific syntax for our doc-builder (similar to MDX) that may not be
+rendered properly in your Markdown viewer.
+
+-->
+
+# Rendimiento y Escalabilidad
+
+Entrenar modelos grandes de transformadores y desplegarlos en producción presenta varios desafíos. Durante el entrenamiento, el modelo puede requerir más memoria de GPU de la disponible o mostrar una velocidad de entrenamiento lenta. En la fase de implementación, el modelo puede tener dificultades para manejar el rendimiento necesario en un entorno de producción.
+
+Esta documentación tiene como objetivo ayudarte a superar estos desafíos y encontrar la configuración óptima para tu caso de uso. Las guías están divididas en secciones de entrenamiento e inferencia, ya que cada una presenta diferentes desafíos y soluciones. Dentro de cada sección, encontrarás guías separadas para diferentes configuraciones de hardware, como GPU única vs. multi-GPU para el entrenamiento o CPU vs. GPU para la inferencia.
+
+Utiliza este documento como punto de partida para navegar hacia los métodos que se ajusten a tu escenario.
+
+## Entrenamiento
+
+Entrenar modelos grandes de transformadores de manera eficiente requiere un acelerador como una GPU o TPU. El caso más común es cuando tienes una GPU única. Los métodos que puedes aplicar para mejorar la eficiencia de entrenamiento en una GPU única también se aplican a otras configuraciones, como múltiples GPU. Sin embargo, también existen técnicas específicas para entrenamiento con múltiples GPU o CPU, las cuales cubrimos en secciones separadas.
+
+* [Métodos y herramientas para un entrenamiento eficiente en una sola GPU](https://huggingface.co/docs/transformers/perf_train_gpu_one): comienza aquí para aprender enfoques comunes que pueden ayudar a optimizar la utilización de memoria de la GPU, acelerar el entrenamiento o ambas cosas.
+* [Sección de entrenamiento con varias GPU](https://huggingface.co/docs/transformers/perf_train_gpu_many): explora esta sección para conocer métodos de optimización adicionales que se aplican a configuraciones con varias GPU, como paralelismo de datos, tensores y canalizaciones.
+* [Sección de entrenamiento en CPU](https://huggingface.co/docs/transformers/perf_train_cpu): aprende sobre entrenamiento de precisión mixta en CPU.
+* [Entrenamiento eficiente en múltiples CPUs](https://huggingface.co/docs/transformers/perf_train_cpu_many): aprende sobre el entrenamiento distribuido en CPU.
+* [Entrenamiento en TPU con TensorFlow](https://huggingface.co/docs/transformers/perf_train_tpu_tf): si eres nuevo en TPUs, consulta esta sección para obtener una introducción basada en opiniones sobre el entrenamiento en TPUs y el uso de XLA.
+* [Hardware personalizado para el entrenamiento](https://huggingface.co/docs/transformers/perf_hardware): encuentra consejos y trucos al construir tu propia plataforma de aprendizaje profundo.
+* [Búsqueda de hiperparámetros utilizando la API del Entrenador](https://huggingface.co/docs/transformers/hpo_train)
+
+## Inferencia
+
+Realizar inferencias eficientes con modelos grandes en un entorno de producción puede ser tan desafiante como entrenarlos. En las siguientes secciones, describimos los pasos para ejecutar inferencias en CPU y configuraciones con GPU única/múltiple.
+
+* [Inferencia en una sola CPU](https://huggingface.co/docs/transformers/perf_infer_cpu)
+* [Inferencia en una sola GPU](https://huggingface.co/docs/transformers/perf_infer_gpu_one)
+* [Inferencia con múltiples GPU](https://huggingface.co/docs/transformers/perf_infer_gpu_one)
+* [Integración de XLA para modelos de TensorFlow](https://huggingface.co/docs/transformers/tf_xla)
+
+## Entrenamiento e Inferencia
+
+Aquí encontrarás técnicas, consejos y trucos que aplican tanto si estás entrenando un modelo como si estás ejecutando inferencias con él.
+
+* [Instanciar un modelo grande](https://huggingface.co/docs/transformers/big_models)
+* [Solución de problemas de rendimiento](https://huggingface.co/docs/transformers/debugging)
+
+## Contribuir
+
+Este documento está lejos de estar completo y aún se deben agregar muchas cosas, así que si tienes adiciones o correcciones que hacer, no dudes en abrir un PR. Si no estás seguro, inicia un Issue y podemos discutir los detalles allí.
+
+Cuando hagas contribuciones que indiquen que A es mejor que B, intenta incluir un benchmark reproducible y/o un enlace a la fuente de esa información (a menos que provenga directamente de ti).

docs/source/fr/index.md

@@ -108,6 +108,7 @@ La documentation est organisée en 5 parties:
 1. **[EncoderDecoder](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](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.
 1. **[ESM](model_doc/esm)** (from Meta AI) are transformer protein language models.  **ESM-1b** was released with the paper [Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences](https://www.pnas.org/content/118/15/e2016239118) by Alexander Rives, Joshua Meier, Tom Sercu, Siddharth Goyal, Zeming Lin, Jason Liu, Demi Guo, Myle Ott, C. Lawrence Zitnick, Jerry Ma, and Rob Fergus. **ESM-1v** was released with the paper [Language models enable zero-shot prediction of the effects of mutations on protein function](https://doi.org/10.1101/2021.07.09.450648) by Joshua Meier, Roshan Rao, Robert Verkuil, Jason Liu, Tom Sercu and Alexander Rives. **ESM-2 and ESMFold** were released with the paper [Language models of protein sequences at the scale of evolution enable accurate structure prediction](https://doi.org/10.1101/2022.07.20.500902) by Zeming Lin, Halil Akin, Roshan Rao, Brian Hie, Zhongkai Zhu, Wenting Lu, Allan dos Santos Costa, Maryam Fazel-Zarandi, Tom Sercu, Sal Candido, Alexander Rives.
+1. **[FastSpeech2Conformer](model_doc/fastspeech2_conformer)** (from ESPnet) released with the paper [Recent Developments On Espnet Toolkit Boosted By Conformer](https://arxiv.org/abs/2010.13956) by Pengcheng Guo, Florian Boyer, Xuankai Chang, Tomoki Hayashi, Yosuke Higuchi, Hirofumi Inaguma, Naoyuki Kamo, Chenda Li, Daniel Garcia-Romero, Jiatong Shi, Jing Shi, Shinji Watanabe, Kun Wei, Wangyou Zhang, and Yuekai Zhang.
 1. **[FLAN-T5](model_doc/flan-t5)** (from Google AI) released in the repository [google-research/t5x](https://github.com/google-research/t5x/blob/main/docs/models.md#flan-t5-checkpoints) by Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Eric Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Albert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdhery, Sharan Narang, Gaurav Mishra, Adams Yu, Vincent Zhao, Yanping Huang, Andrew Dai, Hongkun Yu, Slav Petrov, Ed H. Chi, Jeff Dean, Jacob Devlin, Adam Roberts, Denny Zhou, Quoc V. Le, and Jason Wei
 1. **[FlauBERT](model_doc/flaubert)** (from CNRS) released with the paper [FlauBERT: Unsupervised Language Model Pre-training for French](https://arxiv.org/abs/1912.05372) by Hang Le, Loïc Vial, Jibril Frej, Vincent Segonne, Maximin Coavoux, Benjamin Lecouteux, Alexandre Allauzen, Benoît Crabbé, Laurent Besacier, Didier Schwab.
 1. **[FLAVA](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.
@@ -290,6 +291,7 @@ Le tableau ci-dessous représente la prise en charge actuelle dans la bibliothè
 |             ERNIE             |       ❌       |       ❌       |       ✅        |         ❌         |      ❌      |
 |              ESM              |       ✅       |       ❌       |       ✅        |         ✅         |      ❌      |
 |  FairSeq Machine-Translation  |       ✅       |       ❌       |       ✅        |         ❌         |      ❌      |
+|      FastSpeech2Conformer     |       ✅       |       ❌       |       ✅        |         ❌         |      ❌      |
 |           FlauBERT            |       ✅       |       ❌       |       ✅        |         ✅         |      ❌      |
 |             FLAVA             |       ❌       |       ❌       |       ✅        |         ❌         |      ❌      |
 |             FNet              |       ✅       |       ✅       |       ✅        |         ❌         |      ❌      |

docs/source/fr/installation.md

@@ -149,10 +149,10 @@ Votre environnement Python utilisera la version de la branche `main` lors de la
 
 ## Installation avec conda
 
-Installation via le canal `huggingface` de conda :
+Installation via le canal `conda-forge` de conda :
 
 ```bash
-conda install -c huggingface transformers
+conda install conda-forge::transformers
 ```
 
 ## Configuration du cache

docs/source/it/add_new_pipeline.md

@@ -15,7 +15,7 @@ rendered properly in your Markdown viewer.
 
 # Come creare una pipeline personalizzata?
 
-In questa guida, scopriremo come creare una pipeline personalizzata e condividerla sull' [Hub](hf.co/models) o aggiungerla nella libreria
+In questa guida, scopriremo come creare una pipeline personalizzata e condividerla sull' [Hub](https://hf.co/models) o aggiungerla nella libreria
 Transformers.
 
 Innanzitutto, è necessario decidere gli input grezzi che la pipeline sarà in grado di accettare. Possono essere strings, raw bytes,

docs/source/it/installation.md

@@ -130,10 +130,10 @@ Il tuo ambiente Python troverà la versione `main` di 🤗 Transformers alla pro
 
 ## Installazione con conda
 
-Installazione dal canale conda `huggingface`:
+Installazione dal canale conda `conda-forge`:
 
 ```bash
-conda install -c huggingface transformers
+conda install conda-forge::transformers
 ```
 
 ## Impostazione della cache
@@ -236,4 +236,4 @@ Una volta che il tuo file è scaricato e salvato in cache localmente, specifica
 
 Fai riferimento alla sezione [How to download files from the Hub](https://huggingface.co/docs/hub/how-to-downstream) per avere maggiori dettagli su come scaricare modelli presenti sull Hub.
 
-</Tip>
+</Tip>

docs/source/it/migration.md

@@ -1,320 +1,313 @@
-<!--- 
-Copyright 2020 The HuggingFace Team. Tutti i diritti riservati. 
+<!---
+Copyright 2020 The HuggingFace Team. Tutti i diritti riservati.
 
-Concesso in licenza in base alla Licenza Apache, Versione 2.0 (la "Licenza"); 
-non è possibile utilizzare questo file se non in conformità con la Licenza. 
-È possibile ottenere una copia della Licenza all'indirizzo 
+Concesso in licenza in base alla Licenza Apache, Versione 2.0 (la "Licenza");
+non è possibile utilizzare questo file se non in conformità con la Licenza.
+È possibile ottenere una copia della Licenza all'indirizzo
 
-http://www.apache.org/licenses/LICENSE-2.0 
+http://www.apache.org/licenses/LICENSE-2.0
 
-A meno che non sia richiesto dalla legge applicabile o concordato per iscritto, il software 
-distribuito con la Licenza è distribuito su BASE "COSÌ COM'È", 
-SENZA GARANZIE O CONDIZIONI DI ALCUN TIPO, espresse o implicite. 
-Per la lingua specifica vedi la Licenza che regola le autorizzazioni e 
-le limitazioni ai sensi della STESSA. 
+A meno che non sia richiesto dalla legge applicabile o concordato per iscritto, il software
+distribuito con la Licenza è distribuito su BASE "COSÌ COM'È",
+SENZA GARANZIE O CONDIZIONI DI ALCUN TIPO, espresse o implicite.
+Per la lingua specifica vedi la Licenza che regola le autorizzazioni e
+le limitazioni ai sensi della STESSA.
 
 ⚠️ Note that this file is in Markdown but contain specific syntax for our doc-builder (similar to MDX) that may not be
 rendered properly in your Markdown viewer.
 
---> 
+-->
 
-# Migrazione da pacchetti precedenti 
+# Migrazione da pacchetti precedenti
 
-## Migrazione da transformers `v3.x` a `v4.x` 
+## Migrazione da transformers `v3.x` a `v4.x`
 
-Un paio di modifiche sono state introdotte nel passaggio dalla versione 3 alla versione 4. Di seguito è riportato un riepilogo delle 
-modifiche previste: 
+Un paio di modifiche sono state introdotte nel passaggio dalla versione 3 alla versione 4. Di seguito è riportato un riepilogo delle
+modifiche previste:
 
-#### 1. AutoTokenizer e pipeline ora utilizzano tokenizer veloci (rust) per impostazione predefinita. 
+#### 1. AutoTokenizer e pipeline ora utilizzano tokenizer veloci (rust) per impostazione predefinita.
 
-I tokenizer python e rust hanno all'incirca le stesse API, ma i tokenizer rust hanno un set di funzionalità più completo. 
+I tokenizer python e rust hanno all'incirca le stesse API, ma i tokenizer rust hanno un set di funzionalità più completo.
 
-Ciò introduce due modifiche sostanziali: 
-- La gestione dei token in overflow tra i tokenizer Python e Rust è diversa. 
-- I tokenizers di rust non accettano numeri interi nei metodi di codifica. 
+Ciò introduce due modifiche sostanziali:
+- La gestione dei token in overflow tra i tokenizer Python e Rust è diversa.
+- I tokenizers di rust non accettano numeri interi nei metodi di codifica.
 
-##### Come ottenere lo stesso comportamento di v3.x in v4.x 
+##### Come ottenere lo stesso comportamento di v3.x in v4.x
 
-- Le pipeline ora contengono funzionalità aggiuntive pronte all'uso. Vedi la [pipeline di classificazione dei token con il flag `grouped_entities`](main_classes/pipelines#transformers.TokenClassificationPipeline). 
-- Gli auto-tokenizer ora restituiscono tokenizer rust. Per ottenere invece i tokenizer python, l'utente deve usare il flag `use_fast` impostandolo `False`: 
+- Le pipeline ora contengono funzionalità aggiuntive pronte all'uso. Vedi la [pipeline di classificazione dei token con il flag `grouped_entities`](main_classes/pipelines#transformers.TokenClassificationPipeline).
+- Gli auto-tokenizer ora restituiscono tokenizer rust. Per ottenere invece i tokenizer python, l'utente deve usare il flag `use_fast` impostandolo `False`:
 
-Nella versione `v3.x`: 
-```py 
-from transformers import AutoTokenizer 
+Nella versione `v3.x`:
+```py
+from transformers import AutoTokenizer
 
-tokenizer = AutoTokenizer.from_pretrained("bert-base-cased") 
-``` 
-per ottenere lo stesso nella versione `v4.x`: 
-```py 
-from transformers import AutoTokenizer 
+tokenizer = AutoTokenizer.from_pretrained("bert-base-cased")
+```
+per ottenere lo stesso nella versione `v4.x`:
+```py
+from transformers import AutoTokenizer
 
-tokenizer = AutoTokenizer.from_pretrained("bert-base-cased", use_fast=False) 
-``` 
+tokenizer = AutoTokenizer.from_pretrained("bert-base-cased", use_fast=False)
+```
 
-#### 2. SentencePiece è stato rimosso dalle dipendenze richieste 
+#### 2. SentencePiece è stato rimosso dalle dipendenze richieste
 
-Il requisito sulla dipendenza SentencePiece è stato rimosso da `setup.py`. È stato fatto per avere un canale su anaconda cloud senza basarsi su `conda-forge`. Ciò significa che i tokenizer che dipendono dalla libreria SentencePiece non saranno disponibili con un'installazione standard di `transformers`. 
+Il requisito sulla dipendenza SentencePiece è stato rimosso da `setup.py`. È stato fatto per avere un canale su anaconda cloud senza basarsi su `conda-forge`. Ciò significa che i tokenizer che dipendono dalla libreria SentencePiece non saranno disponibili con un'installazione standard di `transformers`.
 
-Ciò include le versioni **lente** di: 
-- `XLNetTokenizer` 
-- `AlbertTokenizer` 
-- `CamembertTokenizer` 
-- `MBartTokenizer` 
-- `PegasusTokenizer` 
-- `T5Tokenizer` 
-- `ReformerTokenizer` 
-- `XLMRobertaTokenizer` 
+Ciò include le versioni **lente** di:
+- `XLNetTokenizer`
+- `AlbertTokenizer`
+- `CamembertTokenizer`
+- `MBartTokenizer`
+- `PegasusTokenizer`
+- `T5Tokenizer`
+- `ReformerTokenizer`
+- `XLMRobertaTokenizer`
 
-##### Come ottenere lo stesso comportamento della v3.x nella v4.x 
+##### Come ottenere lo stesso comportamento della v3.x nella v4.x
 
-Per ottenere lo stesso comportamento della versione `v3.x`, devi installare anche `sentencepiece`: 
+Per ottenere lo stesso comportamento della versione `v3.x`, devi installare anche `sentencepiece`:
 
-Nella versione `v3.x`: 
-```bash 
-pip install transformers 
-``` 
-per ottenere lo stesso nella versione `v4.x`: 
-```bash 
-pip install transformers[sentencepiece] 
-``` 
-o 
-```bash 
-pip install transformers stentencepiece 
-``` 
-#### 3. L'architettura delle repo è stato aggiornata in modo che ogni modello abbia la propria cartella 
+Nella versione `v3.x`:
+```bash
+pip install transformers
+```
+per ottenere lo stesso nella versione `v4.x`:
+```bash
+pip install transformers[sentencepiece]
+```
+o
+```bash
+pip install transformers stentencepiece
+```
+#### 3. L'architettura delle repo è stato aggiornata in modo che ogni modello abbia la propria cartella
 
-Con l’aggiunta di nuovi modelli, il numero di file nella cartella `src/transformers` continua a crescere e diventa più difficile navigare e capire. Abbiamo fatto la scelta di inserire ogni modello e i file che lo accompagnano nelle proprie sottocartelle. 
+Con l’aggiunta di nuovi modelli, il numero di file nella cartella `src/transformers` continua a crescere e diventa più difficile navigare e capire. Abbiamo fatto la scelta di inserire ogni modello e i file che lo accompagnano nelle proprie sottocartelle.
 
-Si tratta di una modifica sostanziale in quanto l'importazione di layer intermedi utilizzando direttamente il modulo di un modello deve essere eseguita tramite un percorso diverso. 
+Si tratta di una modifica sostanziale in quanto l'importazione di layer intermedi utilizzando direttamente il modulo di un modello deve essere eseguita tramite un percorso diverso.
 
-##### Come ottenere lo stesso comportamento della v3.x nella v4.x 
+##### Come ottenere lo stesso comportamento della v3.x nella v4.x
 
-Per ottenere lo stesso comportamento della versione `v3.x`, devi aggiornare il percorso utilizzato per accedere ai layer. 
+Per ottenere lo stesso comportamento della versione `v3.x`, devi aggiornare il percorso utilizzato per accedere ai layer.
 
-Nella versione `v3.x`: 
-```bash 
-from transformers.modeling_bert import BertLayer 
-``` 
-per ottenere lo stesso nella versione `v4.x`: 
-```bash 
-from transformers.models.bert.modeling_bert import BertLayer 
-``` 
+Nella versione `v3.x`:
+```bash
+from transformers.modeling_bert import BertLayer
+```
+per ottenere lo stesso nella versione `v4.x`:
+```bash
+from transformers.models.bert.modeling_bert import BertLayer
+```
 
-#### 4. Impostare l'argomento `return_dict` su `True` per impostazione predefinita 
+#### 4. Impostare l'argomento `return_dict` su `True` per impostazione predefinita
 
-L'[argomento `return_dict`](main_classes/output) abilita la restituzione di oggetti python dict-like contenenti gli output del modello, invece delle tuple standard. Questo oggetto è self-documented poiché le chiavi possono essere utilizzate per recuperare valori, comportandosi anche come una tupla e gli utenti possono recuperare oggetti per indexing o slicing. 
+L'[argomento `return_dict`](main_classes/output) abilita la restituzione di oggetti python dict-like contenenti gli output del modello, invece delle tuple standard. Questo oggetto è self-documented poiché le chiavi possono essere utilizzate per recuperare valori, comportandosi anche come una tupla e gli utenti possono recuperare oggetti per indexing o slicing.
 
-Questa è una modifica sostanziale poiché la tupla non può essere decompressa: `value0, value1 = outputs` non funzionerà. 
+Questa è una modifica sostanziale poiché la tupla non può essere decompressa: `value0, value1 = outputs` non funzionerà.
 
-##### Come ottenere lo stesso comportamento della v3.x nella v4.x 
+##### Come ottenere lo stesso comportamento della v3.x nella v4.x
 
-Per ottenere lo stesso comportamento della versione `v3.x`, specifica l'argomento `return_dict` come `False`, sia nella configurazione del modello che nel passaggio successivo. 
+Per ottenere lo stesso comportamento della versione `v3.x`, specifica l'argomento `return_dict` come `False`, sia nella configurazione del modello che nel passaggio successivo.
 
-Nella versione `v3.x`: 
-```bash 
-model = BertModel.from_pretrained("bert-base-cased") 
-outputs = model(**inputs) 
-``` 
-per ottenere lo stesso nella versione `v4.x`: 
-```bash 
-model = BertModel.from_pretrained("bert-base-cased") 
-outputs = model(**inputs, return_dict=False) 
-``` 
-o 
-```bash 
-model = BertModel.from_pretrained("bert-base-cased", return_dict=False) 
-outputs = model(**inputs) 
-``` 
+Nella versione `v3.x`:
+```bash
+model = BertModel.from_pretrained("bert-base-cased")
+outputs = model(**inputs)
+```
+per ottenere lo stesso nella versione `v4.x`:
+```bash
+model = BertModel.from_pretrained("bert-base-cased")
+outputs = model(**inputs, return_dict=False)
+```
+o
+```bash
+model = BertModel.from_pretrained("bert-base-cased", return_dict=False)
+outputs = model(**inputs)
+```
 
-#### 5. Rimozione di alcuni attributi deprecati 
+#### 5. Rimozione di alcuni attributi deprecati
 
-Gli attributi sono stati rimossi se deprecati da almeno un mese. L'elenco completo degli attributi obsoleti è disponibile in [#8604](https://github.com/huggingface/transformers/pull/8604). 
+Gli attributi sono stati rimossi se deprecati da almeno un mese. L'elenco completo degli attributi obsoleti è disponibile in [#8604](https://github.com/huggingface/transformers/pull/8604).
 
-Ecco un elenco di questi attributi/metodi/argomenti e quali dovrebbero essere le loro sostituzioni: 
+Ecco un elenco di questi attributi/metodi/argomenti e quali dovrebbero essere le loro sostituzioni:
 
-In diversi modelli, le etichette diventano coerenti con gli altri modelli: 
-- `masked_lm_labels` diventa `labels` in `AlbertForMaskedLM` e `AlbertForPreTraining`. 
-- `masked_lm_labels` diventa `labels` in `BertForMaskedLM` e `BertForPreTraining`. 
-- `masked_lm_labels` diventa `labels` in `DistilBertForMaskedLM`. 
-- `masked_lm_labels` diventa `labels` in `ElectraForMaskedLM`. 
-- `masked_lm_labels` diventa `labels` in `LongformerForMaskedLM`. 
-- `masked_lm_labels` diventa `labels` in `MobileBertForMaskedLM`. 
-- `masked_lm_labels` diventa `labels` in `RobertaForMaskedLM`. 
-- `lm_labels` diventa `labels` in `BartForConditionalGeneration`. 
-- `lm_labels` diventa `labels` in `GPT2DoubleHeadsModel`. 
-- `lm_labels` diventa `labels` in `OpenAIGPTDoubleHeadsModel`. 
-- `lm_labels` diventa `labels` in `T5ForConditionalGeneration`. 
+In diversi modelli, le etichette diventano coerenti con gli altri modelli:
+- `masked_lm_labels` diventa `labels` in `AlbertForMaskedLM` e `AlbertForPreTraining`.
+- `masked_lm_labels` diventa `labels` in `BertForMaskedLM` e `BertForPreTraining`.
+- `masked_lm_labels` diventa `labels` in `DistilBertForMaskedLM`.
+- `masked_lm_labels` diventa `labels` in `ElectraForMaskedLM`.
+- `masked_lm_labels` diventa `labels` in `LongformerForMaskedLM`.
+- `masked_lm_labels` diventa `labels` in `MobileBertForMaskedLM`.
+- `masked_lm_labels` diventa `labels` in `RobertaForMaskedLM`.
+- `lm_labels` diventa `labels` in `BartForConditionalGeneration`.
+- `lm_labels` diventa `labels` in `GPT2DoubleHeadsModel`.
+- `lm_labels` diventa `labels` in `OpenAIGPTDoubleHeadsModel`.
+- `lm_labels` diventa `labels` in `T5ForConditionalGeneration`.
 
-In diversi modelli, il meccanismo di memorizzazione nella cache diventa coerente con gli altri: 
-- `decoder_cached_states` diventa `past_key_values` in tutti i modelli BART-like, FSMT e T5. 
-- `decoder_past_key_values` diventa `past_key_values` in tutti i modelli BART-like, FSMT e T5. 
-- `past` diventa `past_key_values` in tutti i modelli CTRL. 
-- `past` diventa `past_key_values` in tutti i modelli GPT-2. 
+In diversi modelli, il meccanismo di memorizzazione nella cache diventa coerente con gli altri:
+- `decoder_cached_states` diventa `past_key_values` in tutti i modelli BART-like, FSMT e T5.
+- `decoder_past_key_values` diventa `past_key_values` in tutti i modelli BART-like, FSMT e T5.
+- `past` diventa `past_key_values` in tutti i modelli CTRL.
+- `past` diventa `past_key_values` in tutti i modelli GPT-2.
 
-Per quanto riguarda le classi tokenizer: 
-- L'attributo tokenizer `max_len` diventa `model_max_length`. 
-- L'attributo tokenizer `return_lengths` diventa `return_length`. 
-- L'argomento di codifica del tokenizer `is_pretokenized` diventa `is_split_into_words`. 
+Per quanto riguarda le classi tokenizer:
+- L'attributo tokenizer `max_len` diventa `model_max_length`.
+- L'attributo tokenizer `return_lengths` diventa `return_length`.
+- L'argomento di codifica del tokenizer `is_pretokenized` diventa `is_split_into_words`.
 
-Per quanto riguarda la classe `Trainer`: 
-- L'argomento `tb_writer` di `Trainer` è stato rimosso in favore della funzione richiamabile `TensorBoardCallback(tb_writer=...)`. 
-- L'argomento `prediction_loss_only` di `Trainer` è stato rimosso in favore dell'argomento di classe `args.prediction_loss_only`. 
-- L'attributo `data_collator` di `Trainer` sarà richiamabile. 
-- Il metodo `_log` di `Trainer` è deprecato a favore di `log`. 
-- Il metodo `_training_step` di `Trainer` è deprecato a favore di `training_step`. 
-- Il metodo `_prediction_loop` di `Trainer` è deprecato a favore di `prediction_loop`. 
-- Il metodo `is_local_master` di `Trainer` è deprecato a favore di `is_local_process_zero`. 
-- Il metodo `is_world_master` di `Trainer` è deprecato a favore di `is_world_process_zero`. 
+Per quanto riguarda la classe `Trainer`:
+- L'argomento `tb_writer` di `Trainer` è stato rimosso in favore della funzione richiamabile `TensorBoardCallback(tb_writer=...)`.
+- L'argomento `prediction_loss_only` di `Trainer` è stato rimosso in favore dell'argomento di classe `args.prediction_loss_only`.
+- L'attributo `data_collator` di `Trainer` sarà richiamabile.
+- Il metodo `_log` di `Trainer` è deprecato a favore di `log`.
+- Il metodo `_training_step` di `Trainer` è deprecato a favore di `training_step`.
+- Il metodo `_prediction_loop` di `Trainer` è deprecato a favore di `prediction_loop`.
+- Il metodo `is_local_master` di `Trainer` è deprecato a favore di `is_local_process_zero`.
+- Il metodo `is_world_master` di `Trainer` è deprecato a favore di `is_world_process_zero`.
 
-Per quanto riguarda la classe `TFTrainer`: 
-- L'argomento `prediction_loss_only` di `TFTrainer` è stato rimosso a favore dell'argomento di classe `args.prediction_loss_only`. 
-- Il metodo `_log` di `Trainer` è deprecato a favore di `log`. 
-- Il metodo `_prediction_loop` di `TFTrainer` è deprecato a favore di `prediction_loop`. 
-- Il metodo `_setup_wandb` di `TFTrainer` è deprecato a favore di `setup_wandb`. 
-- Il metodo `_run_model` di `TFTrainer` è deprecato a favore di `run_model`. 
+Per quanto riguarda la classe `TrainingArguments`:
+- L'argomento `evaluate_during_training` di `TrainingArguments` è deprecato a favore di `evaluation_strategy`.
 
-Per quanto riguarda la classe `TrainingArguments`: 
-- L'argomento `evaluate_during_training` di `TrainingArguments` è deprecato a favore di `evaluation_strategy`. 
+Per quanto riguarda il modello Transfo-XL:
+- L'attributo di configurazione `tie_weight` di Transfo-XL diventa `tie_words_embeddings`.
+- Il metodo di modellazione `reset_length` di Transfo-XL diventa `reset_memory_length`.
 
-Per quanto riguarda il modello Transfo-XL: 
-- L'attributo di configurazione `tie_weight` di Transfo-XL diventa `tie_words_embeddings`. 
-- Il metodo di modellazione `reset_length` di Transfo-XL diventa `reset_memory_length`. 
-
-Per quanto riguarda le pipeline: 
-- L'argomento `topk` di `FillMaskPipeline` diventa `top_k`. 
+Per quanto riguarda le pipeline:
+- L'argomento `topk` di `FillMaskPipeline` diventa `top_k`.
 
 
 
-## Passaggio da pytorch-transformers a 🤗 Transformers 
+## Passaggio da pytorch-transformers a 🤗 Transformers
 
-Ecco un breve riepilogo di ciò a cui prestare attenzione durante il passaggio da `pytorch-transformers` a 🤗 Transformers. 
+Ecco un breve riepilogo di ciò a cui prestare attenzione durante il passaggio da `pytorch-transformers` a 🤗 Transformers.
 
-### L’ordine posizionale di alcune parole chiave di input dei modelli (`attention_mask`, `token_type_ids`...) è cambiato 
+### L’ordine posizionale di alcune parole chiave di input dei modelli (`attention_mask`, `token_type_ids`...) è cambiato
 
-Per usare Torchscript (vedi #1010, #1204 e #1195) l'ordine specifico delle **parole chiave di input** di alcuni modelli (`attention_mask`, `token_type_ids`...) è stato modificato. 
+Per usare Torchscript (vedi #1010, #1204 e #1195) l'ordine specifico delle **parole chiave di input** di alcuni modelli (`attention_mask`, `token_type_ids`...) è stato modificato.
 
-Se inizializzavi i modelli usando parole chiave per gli argomenti, ad esempio `model(inputs_ids, attention_mask=attention_mask, token_type_ids=token_type_ids)`, questo non dovrebbe causare alcun cambiamento. 
+Se inizializzavi i modelli usando parole chiave per gli argomenti, ad esempio `model(inputs_ids, attention_mask=attention_mask, token_type_ids=token_type_ids)`, questo non dovrebbe causare alcun cambiamento.
 
-Se inizializzavi i modelli con input posizionali per gli argomenti, ad esempio `model(inputs_ids, attention_mask, token_type_ids)`, potrebbe essere necessario ricontrollare l'ordine esatto degli argomenti di input. 
+Se inizializzavi i modelli con input posizionali per gli argomenti, ad esempio `model(inputs_ids, attention_mask, token_type_ids)`, potrebbe essere necessario ricontrollare l'ordine esatto degli argomenti di input.
 
-## Migrazione da pytorch-pretrained-bert 
+## Migrazione da pytorch-pretrained-bert
 
-Ecco un breve riepilogo di ciò a cui prestare attenzione durante la migrazione da `pytorch-pretrained-bert` a 🤗 Transformers 
+Ecco un breve riepilogo di ciò a cui prestare attenzione durante la migrazione da `pytorch-pretrained-bert` a 🤗 Transformers
 
-### I modelli restituiscono sempre `tuple` 
+### I modelli restituiscono sempre `tuple`
 
-La principale modifica di rilievo durante la migrazione da `pytorch-pretrained-bert` a 🤗 Transformers è che il metodo dei modelli di previsione dà sempre una `tupla` con vari elementi a seconda del modello e dei parametri di configurazione. 
+La principale modifica di rilievo durante la migrazione da `pytorch-pretrained-bert` a 🤗 Transformers è che il metodo dei modelli di previsione dà sempre una `tupla` con vari elementi a seconda del modello e dei parametri di configurazione.
 
-Il contenuto esatto delle tuple per ciascun modello è mostrato in dettaglio nelle docstring dei modelli e nella [documentazione](https://huggingface.co/transformers/). 
+Il contenuto esatto delle tuple per ciascun modello è mostrato in dettaglio nelle docstring dei modelli e nella [documentazione](https://huggingface.co/transformers/).
 
-In quasi tutti i casi, andrà bene prendendo il primo elemento dell'output come quello che avresti precedentemente utilizzato in `pytorch-pretrained-bert`. 
+In quasi tutti i casi, andrà bene prendendo il primo elemento dell'output come quello che avresti precedentemente utilizzato in `pytorch-pretrained-bert`.
 
 Ecco un esempio di conversione da `pytorch-pretrained-bert`
- a 🤗 Transformers per un modello di classificazione `BertForSequenceClassification`: 
+ a 🤗 Transformers per un modello di classificazione `BertForSequenceClassification`:
 
-```python 
-# Carichiamo il nostro modello 
-model = BertForSequenceClassification.from_pretrained("bert-base-uncased") 
+```python
+# Carichiamo il nostro modello
+model = BertForSequenceClassification.from_pretrained("bert-base-uncased")
 
-# Se usavi questa riga in pytorch-pretrained-bert : 
-loss = model(input_ids, labels=labels) 
+# Se usavi questa riga in pytorch-pretrained-bert :
+loss = model(input_ids, labels=labels)
 
-# Ora usa questa riga in 🤗 Transformers per estrarre la perdita dalla tupla di output: 
-outputs = model(input_ids, labels=labels) 
-loss = outputs[0] 
+# Ora usa questa riga in 🤗 Transformers per estrarre la perdita dalla tupla di output:
+outputs = model(input_ids, labels=labels)
+loss = outputs[0]
 
-# In 🤗 Transformers puoi anche avere accesso ai logit: 
-loss, logits = outputs[:2] 
+# In 🤗 Transformers puoi anche avere accesso ai logit:
+loss, logits = outputs[:2]
 
-# Ed anche agli attention weight se configuri il modello per restituirli (e anche altri output, vedi le docstring e la documentazione) 
-model = BertForSequenceClassification.from_pretrained(" bert-base-uncased", output_attentions=True) 
-outputs = model(input_ids, labels=labels) 
-loss, logits, attentions = outputs 
-``` 
+# Ed anche agli attention weight se configuri il modello per restituirli (e anche altri output, vedi le docstring e la documentazione)
+model = BertForSequenceClassification.from_pretrained(" bert-base-uncased", output_attentions=True)
+outputs = model(input_ids, labels=labels)
+loss, logits, attentions = outputs
+```
 
-### Serializzazione 
+### Serializzazione
 
-Modifica sostanziale nel metodo `from_pretrained()`: 
+Modifica sostanziale nel metodo `from_pretrained()`:
 
-1. I modelli sono ora impostati in modalità di valutazione in maniera predefinita quando usi il metodo `from_pretrained()`. Per addestrarli non dimenticare di riportarli in modalità di addestramento (`model.train()`) per attivare i moduli di dropout. 
+1. I modelli sono ora impostati in modalità di valutazione in maniera predefinita quando usi il metodo `from_pretrained()`. Per addestrarli non dimenticare di riportarli in modalità di addestramento (`model.train()`) per attivare i moduli di dropout.
 
-2. Gli argomenti aggiuntivi `*inputs` e `**kwargs` forniti al metodo `from_pretrained()` venivano passati direttamente al metodo `__init__()` della classe sottostante del modello. Ora sono usati per aggiornare prima l'attributo di configurazione del modello, che può non funzionare con le classi del modello derivate costruite basandosi sui precedenti esempi di `BertForSequenceClassification`. Più precisamente, gli argomenti posizionali `*inputs` forniti a `from_pretrained()` vengono inoltrati direttamente al metodo `__init__()`  del modello mentre gli argomenti keyword `**kwargs` (i) che corrispondono agli attributi della classe di configurazione, vengono utilizzati per aggiornare tali attributi (ii) che non corrispondono ad alcun attributo della classe di configurazione, vengono inoltrati al metodo `__init__()`. 
+2. Gli argomenti aggiuntivi `*inputs` e `**kwargs` forniti al metodo `from_pretrained()` venivano passati direttamente al metodo `__init__()` della classe sottostante del modello. Ora sono usati per aggiornare prima l'attributo di configurazione del modello, che può non funzionare con le classi del modello derivate costruite basandosi sui precedenti esempi di `BertForSequenceClassification`. Più precisamente, gli argomenti posizionali `*inputs` forniti a `from_pretrained()` vengono inoltrati direttamente al metodo `__init__()`  del modello mentre gli argomenti keyword `**kwargs` (i) che corrispondono agli attributi della classe di configurazione, vengono utilizzati per aggiornare tali attributi (ii) che non corrispondono ad alcun attributo della classe di configurazione, vengono inoltrati al metodo `__init__()`.
 
-Inoltre, sebbene non si tratti di una modifica sostanziale, i metodi di serializzazione sono stati standardizzati e probabilmente dovresti passare al nuovo metodo `save_pretrained(save_directory)` se prima usavi qualsiasi altro metodo di serializzazione. 
+Inoltre, sebbene non si tratti di una modifica sostanziale, i metodi di serializzazione sono stati standardizzati e probabilmente dovresti passare al nuovo metodo `save_pretrained(save_directory)` se prima usavi qualsiasi altro metodo di serializzazione.
 
-Ecco un esempio: 
+Ecco un esempio:
 
-```python 
-### Carichiamo un modello e un tokenizer 
-model = BertForSequenceClassification.from_pretrained("bert-base-uncased") 
-tokenizer = BertTokenizer.from_pretrained("bert-base-uncased") 
+```python
+### Carichiamo un modello e un tokenizer
+model = BertForSequenceClassification.from_pretrained("bert-base-uncased")
+tokenizer = BertTokenizer.from_pretrained("bert-base-uncased")
 
-### Facciamo fare alcune cose al nostro modello e tokenizer 
-# Es: aggiungiamo nuovi token al vocabolario e agli embending del nostro modello 
-tokenizer.add_tokens(["[SPECIAL_TOKEN_1]", "[SPECIAL_TOKEN_2]"]) 
-model.resize_token_embeddings(len(tokenizer)) 
+### Facciamo fare alcune cose al nostro modello e tokenizer
+# Es: aggiungiamo nuovi token al vocabolario e agli embending del nostro modello
+tokenizer.add_tokens(["[SPECIAL_TOKEN_1]", "[SPECIAL_TOKEN_2]"])
+model.resize_token_embeddings(len(tokenizer))
 # Alleniamo il nostro modello
-train(model) 
+train(model)
 
-### Ora salviamo il nostro modello e il tokenizer in una cartella 
-model.save_pretrained("./my_saved_model_directory/") 
-tokenizer.save_pretrained("./my_saved_model_directory/") 
+### Ora salviamo il nostro modello e il tokenizer in una cartella
+model.save_pretrained("./my_saved_model_directory/")
+tokenizer.save_pretrained("./my_saved_model_directory/")
 
-### Ricarichiamo il modello e il tokenizer 
-model = BertForSequenceClassification.from_pretrained("./my_saved_model_directory/") 
-tokenizer = BertTokenizer.from_pretrained("./my_saved_model_directory/") 
-``` 
+### Ricarichiamo il modello e il tokenizer
+model = BertForSequenceClassification.from_pretrained("./my_saved_model_directory/")
+tokenizer = BertTokenizer.from_pretrained("./my_saved_model_directory/")
+```
 
-### Ottimizzatori: BertAdam e OpenAIAdam ora sono AdamW, lo scheduling è quello standard PyTorch 
+### Ottimizzatori: BertAdam e OpenAIAdam ora sono AdamW, lo scheduling è quello standard PyTorch
 
-I due ottimizzatori precedenti inclusi, `BertAdam` e `OpenAIAdam`, sono stati sostituiti da un singolo `AdamW` che presenta alcune differenze: 
+I due ottimizzatori precedenti inclusi, `BertAdam` e `OpenAIAdam`, sono stati sostituiti da un singolo `AdamW` che presenta alcune differenze:
 
-- implementa solo la correzione del weights decay, 
-- lo scheduling ora è esterno (vedi sotto), 
-- anche il gradient clipping ora è esterno (vedi sotto). 
+- implementa solo la correzione del weights decay,
+- lo scheduling ora è esterno (vedi sotto),
+- anche il gradient clipping ora è esterno (vedi sotto).
 
 Il nuovo ottimizzatore `AdamW` corrisponde alle API di `Adam` di PyTorch e ti consente di utilizzare metodi PyTorch o apex per lo scheduling e il clipping.
 
-Lo scheduling è ora standard [PyTorch learning rate schedulers](https://pytorch.org/docs/stable/optim.html#how-to-adjust-learning-rate) e non fanno più parte dell'ottimizzatore. 
+Lo scheduling è ora standard [PyTorch learning rate schedulers](https://pytorch.org/docs/stable/optim.html#how-to-adjust-learning-rate) e non fanno più parte dell'ottimizzatore.
 
-Ecco un esempio di linear warmup e decay con `BertAdam` e con `AdamW`: 
+Ecco un esempio di linear warmup e decay con `BertAdam` e con `AdamW`:
 
-```python 
-# Parametri: 
-lr = 1e-3 
-max_grad_norm = 1.0 
-num_training_steps = 1000 
-num_warmup_steps = 100 
-warmup_proportion = float( num_warmup_steps) / float(num_training_steps) # 0.1 
+```python
+# Parametri:
+lr = 1e-3
+max_grad_norm = 1.0
+num_training_steps = 1000
+num_warmup_steps = 100
+warmup_proportion = float( num_warmup_steps) / float(num_training_steps) # 0.1
 
-### In precedenza l'ottimizzatore BertAdam veniva istanziato in questo modo: 
-optimizer = BertAdam( 
-   model.parameters(), 
-   lr=lr, 
-   schedule="warmup_linear", 
-   warmup=warmup_proportion, 
-   num_training_steps=num_training_steps, 
-) 
-### e usato in questo modo: 
-for batch in train_data: 
-   loss = model(batch) 
-   loss.backward() 
-   optimizer.step() 
+### In precedenza l'ottimizzatore BertAdam veniva istanziato in questo modo:
+optimizer = BertAdam(
+   model.parameters(),
+   lr=lr,
+   schedule="warmup_linear",
+   warmup=warmup_proportion,
+   num_training_steps=num_training_steps,
+)
+### e usato in questo modo:
+for batch in train_data:
+   loss = model(batch)
+   loss.backward()
+   optimizer.step()
 
-### In 🤗 Transformers, ottimizzatore e schedule sono divisi e usati in questo modo: 
-optimizer = AdamW( 
-   model.parameters(), lr=lr, correct_bias=False 
-) # Per riprodurre il comportamento specifico di BertAdam impostare correct_bias=False 
-scheduler = get_linear_schedule_with_warmup( 
-   optimizer, num_warmup_steps=num_warmup_steps, num_training_steps=num_training_steps 
-) # PyTorch scheduler 
-### e va usato così: 
-for batch in train_data: 
-   loss = model(batch) 
-   loss.backward() 
-   torch.nn.utils.clip_grad_norm_( 
-   model.parameters(), max_grad_norm 
-   ) # Gradient clipping non è più in AdamW (quindi puoi usare amp senza problemi) 
-   optimizer.step() 
+### In 🤗 Transformers, ottimizzatore e schedule sono divisi e usati in questo modo:
+optimizer = AdamW(
+   model.parameters(), lr=lr, correct_bias=False
+) # Per riprodurre il comportamento specifico di BertAdam impostare correct_bias=False
+scheduler = get_linear_schedule_with_warmup(
+   optimizer, num_warmup_steps=num_warmup_steps, num_training_steps=num_training_steps
+) # PyTorch scheduler
+### e va usato così:
+for batch in train_data:
+   loss = model(batch)
+   loss.backward()
+   torch.nn.utils.clip_grad_norm_(
+   model.parameters(), max_grad_norm
+   ) # Gradient clipping non è più in AdamW (quindi puoi usare amp senza problemi)
+   optimizer.step()
    scheduler.step()
 ```

docs/source/ja/installation.md

@@ -135,10 +135,10 @@ Python環境は次回の実行時に🤗 Transformersの`main`バージョンを
 
 ## condaでのインストール
 
-`huggingface`のcondaチャンネルからインストールします:
+`conda-forge`のcondaチャンネルからインストールします:
 
 ```bash
-conda install -c huggingface transformers
+conda install conda-forge::transformers
 ```
 
 ## キャッシュの設定
@@ -241,4 +241,4 @@ python examples/pytorch/translation/run_translation.py --model_name_or_path t5-s
 
 Hubに保存されているファイルをダウンロードする方法の詳細については、[How to download files from the Hub](https://huggingface.co/docs/hub/how-to-downstream)セクションを参照してください。
 
-</Tip>
+</Tip>

docs/source/ja/internal/generation_utils.md

@@ -45,7 +45,7 @@ inputs = tokenizer("Hello, my dog is cute and ", return_tensors="pt")
 generation_output = model.generate(**inputs, return_dict_in_generate=True, output_scores=True)
 ```
 
-`generation_output` オブジェクトは、できる限り [`~generation.GreedySearchDecoderOnlyOutput`] です。
+`generation_output` オブジェクトは、できる限り [`~generation.GenerateDecoderOnlyOutput`] です。
 以下のそのクラスのドキュメントを参照してください。これは、次の属性があることを意味します。
 
 - `sequences`: 生成されたトークンのシーケンス
@@ -76,25 +76,13 @@ generation_output[:2]
 
 ### PyTorch
 
-[[autodoc]] generation.GreedySearchEncoderDecoderOutput
+[[autodoc]] generation.GenerateDecoderOnlyOutput
 
-[[autodoc]] generation.GreedySearchDecoderOnlyOutput
+[[autodoc]] generation.GenerateEncoderDecoderOutput
 
-[[autodoc]] generation.SampleEncoderDecoderOutput
+[[autodoc]] generation.GenerateBeamDecoderOnlyOutput
 
-[[autodoc]] generation.SampleDecoderOnlyOutput
-
-[[autodoc]] generation.BeamSearchEncoderDecoderOutput
-
-[[autodoc]] generation.BeamSearchDecoderOnlyOutput
-
-[[autodoc]] generation.BeamSampleEncoderDecoderOutput
-
-[[autodoc]] generation.BeamSampleDecoderOnlyOutput
-
-[[autodoc]] generation.ContrastiveSearchEncoderDecoderOutput
-
-[[autodoc]] generation.ContrastiveSearchDecoderOnlyOutput
+[[autodoc]] generation.GenerateBeamEncoderDecoderOutput
 
 ### TensorFlow
 

docs/source/ja/main_classes/deepspeed.md

@@ -1910,7 +1910,7 @@ SW: Model with 2783M total params, 65M largest layer params.
 
 3. 次の出力:
 
-   ```bash 
+   ```bash
     python -c 'import torch; print(f"torch: {torch.__version__}")'
     python -c 'import transformers; print(f"transformers: {transformers.__version__}")'
     python -c 'import deepspeed; print(f"deepspeed: {deepspeed.__version__}")'
@@ -1994,7 +1994,6 @@ SW: Model with 2783M total params, 65M largest layer params.
 
 ### Notes
 
-- DeepSpeed は PyTorch [`Trainer`] では動作しますが、TF [`TFTrainer`] では動作しません。
 - DeepSpeed には pip でインストール可能な PyPI パッケージがありますが、ハードウェアに最も適合するように、また有効にする必要がある場合は、[ソース](https://github.com/microsoft/deepspeed#installation) からインストールすることを強くお勧めします。
   1 ビット Adam などの特定の機能は、pypi ディストリビューションでは利用できません。
 - 🤗 Transformers で DeepSpeed を使用するために [`Trainer`] を使用する必要はありません - 任意のモデルを使用できます

docs/source/ko/add_new_pipeline.md

@@ -15,7 +15,7 @@ rendered properly in your Markdown viewer.
 
 # 어떻게 사용자 정의 파이프라인을 생성하나요? [[how-to-create-a-custom-pipeline]]
 
-이 가이드에서는 사용자 정의 파이프라인을 어떻게 생성하고 [허브](hf.co/models)에 공유하거나 🤗 Transformers 라이브러리에 추가하는 방법을 살펴보겠습니다.
+이 가이드에서는 사용자 정의 파이프라인을 어떻게 생성하고 [허브](https://hf.co/models)에 공유하거나 🤗 Transformers 라이브러리에 추가하는 방법을 살펴보겠습니다.
 
 먼저 파이프라인이 수용할 수 있는 원시 입력을 결정해야 합니다.
 문자열, 원시 바이트, 딕셔너리 또는 가장 원하는 입력일 가능성이 높은 것이면 무엇이든 가능합니다.

docs/source/ko/installation.md

@@ -49,7 +49,7 @@ Windows의 경우:
 .env/Scripts/activate
 ```
 
-이제 🤗 Transformers를 설치할 준비가 되었습니다. 다음 명령을 입력해주세요. 
+이제 🤗 Transformers를 설치할 준비가 되었습니다. 다음 명령을 입력해주세요.
 
 ```bash
 pip install transformers
@@ -135,10 +135,10 @@ Python 환경을 다시 실행하면 업데이트된 🤗 Transformers의 `main`
 
 ## conda로 설치하기[[install-with-conda]]
 
-`huggingface` conda 채널에서 설치할 수 있습니다.
+`conda-forge` conda 채널에서 설치할 수 있습니다.
 
 ```bash
-conda install -c huggingface transformers
+conda install conda-forge::transformers
 ```
 
 ## 캐시 구성하기[[cache-setup]]
@@ -242,4 +242,4 @@ Another option for using 🤗 Transformers offline is to download the files ahea
 
 Hub에 저장된 파일을 다운로드하는 방법을 더 자세히 알아보려면 [Hub에서 파일 다운로드하기](https://huggingface.co/docs/hub/how-to-downstream) 섹션을 참고해주세요.
 
-</Tip>
+</Tip>

docs/source/pt/installation.md

@@ -144,10 +144,10 @@ O ambiente de Python que foi criado para a instalação do 🤗 Transformers enc
 
 ## Instalação usando o Conda
 
-É possível instalar o 🤗 Transformers a partir do canal conda `huggingface` com o seguinte comando:
+É possível instalar o 🤗 Transformers a partir do canal conda `conda-forge` com o seguinte comando:
 
 ```bash
-conda install -c huggingface transformers
+conda install conda-forge::transformers
 ```
 
 ## Configuração do Cachê
@@ -166,7 +166,7 @@ No Windows, este diretório pré-definido é dado por `C:\Users\username\.cache\
     O 🤗 Transformers usará as variáveis de ambiente do shell `PYTORCH_TRANSFORMERS_CACHE` ou `PYTORCH_PRETRAINED_BERT_CACHE`
     se estiver vindo de uma versão anterior da biblioteca que tenha configurado essas variáveis de ambiente, a menos que
     você especifique a variável de ambiente do shell `TRANSFORMERS_CACHE`.
-    
+
 </Tip>
 
 

docs/source/zh/_toctree.yml

@@ -60,6 +60,10 @@
   - local: perf_torch_compile
     title: 使用 `torch.compile()` 优化推理
   title: 性能和可扩展性
+- sections:
+  - local: contributing
+    title: 如何为 🤗 Transformers 做贡献？
+  title: 贡献
 - sections:
   - local: task_summary
     title: 🤗Transformers能做什么

docs/source/zh/contributing.md

@@ -0,0 +1,331 @@
+<!---
+Copyright 2020 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+-->
+
+# 为 🤗 Transformers 做贡献
+
+欢迎所有人为 🤗 Transformers 做出贡献，我们重视每个人的贡献。代码贡献并不是帮助社区的唯一途径。回答问题、帮助他人和改进文档也非常有价值。
+
+宣传 🤗 Transformers 也会帮助我们！比如在博客文章里介绍一下这个库是如何帮助你完成了很棒的项目，每次它帮助你时都在 Twitter 上大声宣传，或者给这个代码仓库点⭐️来表示感谢。
+
+无论你选择以哪种方式做出贡献，请注意并尊重我们的[行为准则](https://github.com/huggingface/transformers/blob/main/CODE_OF_CONDUCT.md)。
+
+**本指南的灵感来源于 [scikit-learn贡献指南](https://github.com/scikit-learn/scikit-learn/blob/main/CONTRIBUTING.md) ，它令人印象深刻.**
+
+## 做贡献的方法
+
+有多种方法可以为 🤗 Transformers 做贡献：
+
+* 修复现有代码中尚未解决的问题。
+* 提交与 bug 或所需新功能相关的 issue。
+* 实现新的模型。
+* 为示例或文档做贡献。
+
+如果你不知道从哪里开始，有一个特别的 [Good First Issue](https://github.com/huggingface/transformers/contribute) 列表。它会列出一些适合初学者的开放的 issues，并帮助你开始为开源项目做贡献。只需要在你想要处理的 issue 下发表评论就行。 
+
+如果想要稍微更有挑战性的内容，你也可以查看 [Good Second Issue](https://github.com/huggingface/transformers/labels/Good%20Second%20Issue) 列表。总的来说，如果你觉得自己知道该怎么做，就去做吧，我们会帮助你达到目标的！🚀
+
+> 所有的贡献对社区来说都同样宝贵。🥰
+
+## 修复尚未解决的问题
+
+如果你发现现有代码中存在问题，并且已经想到了解决方法，请随时[开始贡献](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md/#create-a-pull-request) 并创建一个 Pull Request！
+
+## 提交与 bug 相关的 issue 或功能请求
+
+在提交与错误相关的 issue 或功能请求时，请尽量遵循下面的指南。这能让我们更容易迅速回复你，并提供良好的反馈意见。
+
+### 你发现了 bug 吗？
+
+🤗 Transformers 之所以强大可靠，要感谢用户报告了他们遇到的问题。
+
+在提出issue之前，请你**确认该 bug 尚未被报告**（使用 GitHub 的 Issues 下面的搜索栏）。issue 也应该是与库本身的 bug 有关，而不是与你的代码有关。如果不确定 bug 是在你的代码中还是在库中，请先在[论坛](https://discuss.huggingface.co/)中询问。这有助于我们更快地解决与库相关的问题。
+
+一旦你确认该 bug 尚未被报告，请在你的 issue 中包含以下信息，以便我们快速解决：
+
+* 使用的**操作系统类型和版本**，以及 **Python**、**PyTorch** 和 **TensorFlow** 的版本。
+* 一个简短、独立的代码片段，可以让我们在不到30秒内重现这个问题。
+* 如果发生异常，请提供*完整的* traceback。
+* 附上你认为可能有帮助的任何其他附加信息，如屏幕截图。
+
+想要自动获取操作系统和软件版本，请运行以下命令：
+
+```bash
+transformers-cli env
+```
+
+你也可以从代码仓库的根目录下运行相同的命令：
+
+```bash
+python src/transformers/commands/transformers_cli.py env
+```
+
+### 你想要新功能吗？
+
+如果你希望在 🤗 Transformers 中看到新功能，请提出一个 issue 并包含以下内容：
+
+1. 这个新功能的*动机*是什么呢？是因为使用这个库时遇到了问题或者感到了某种不满吗？是因为你的项目需要这个功能吗？或者是你自己开发了某项内容，并且认为它可能会对社区有所帮助？
+
+   不管是什么，我们都很想听！
+
+2. 请尽可能详细地描述你想要的功能。你告诉我们的越多，我们就能更好地帮助你。
+3. 请提供一个*代码片段*，演示该功能的使用方法。
+4. 如果这个功能与某篇论文相关，请包含链接。
+
+如果你描述得足够清晰，那么在你创建 issue 时，我们已经完成了80%的工作。
+
+我们已经添加了[模板](https://github.com/huggingface/transformers/tree/main/templates)，可能有助于你提出 issue。
+
+## 你想要实现一个新模型吗？
+
+我们会持续发布新模型，如果你想要实现一个新模型，请提供以下信息:
+
+* 模型的简要描述和论文链接。
+* 如果实现是开源的，请提供实现的链接。
+* 如果模型权重可用，请提供模型权重的链接。
+
+如果你想亲自贡献模型，请告诉我们。让我们帮你把它添加到 🤗 Transformers！
+
+我们已经添加了[详细的指南和模板](https://github.com/huggingface/transformers/tree/main/templates)来帮助你添加新模型。我们还有一个更技术性的指南，告诉你[如何将模型添加到 🤗 Transformers](https://huggingface.co/docs/transformers/add_new_model)。
+
+## 你想要添加文档吗？
+
+我们始终在寻求改进文档，使其更清晰准确。请告诉我们如何改进文档，比如拼写错误以及任何缺失、不清楚或不准确的内容。我们非常乐意进行修改，如果你有兴趣，我们也可以帮助你做出贡献！
+
+有关如何生成、构建和编写文档的更多详细信息，请查看文档 [README](https://github.com/huggingface/transformers/tree/main/docs)。
+
+## 创建 Pull Request
+
+在开始编写任何代码之前，我们强烈建议你先搜索现有的 PR(Pull Request) 或 issue，以确保没有其他人已经在做同样的事情。如果你不确定，提出 issue 来获取反馈意见是一个好办法。
+
+要为 🤗 Transformers 做贡献，你需要基本的 `git` 使用技能。虽然 `git` 不是一个很容易使用的工具，但它提供了非常全面的手册，在命令行中输入 `git --help` 并享受吧！如果你更喜欢书籍，[Pro Git](https://git-scm.com/book/en/v2)是一本很好的参考书。
+
+要为 🤗 Transformers 做贡献，你需要 **[Python 3.8]((https://github.com/huggingface/transformers/blob/main/setup.py#L426))** 或更高版本。请按照以下步骤开始贡献：
+
+1. 点击[仓库](https://github.com/huggingface/transformers)页面上的 **[Fork](https://github.com/huggingface/transformers/fork)** 按钮，这会在你的 GitHub 账号下拷贝一份代码。
+
+2. 把派生仓库克隆到本地磁盘，并将基础仓库添加为远程仓库：
+
+   ```bash
+   git clone git@github.com:<your Github handle>/transformers.git
+   cd transformers
+   git remote add upstream https://github.com/huggingface/transformers.git
+   ```
+
+3. 创建一个新的分支来保存你的更改：
+
+   ```bash
+   git checkout -b a-descriptive-name-for-my-changes
+   ```
+
+   🚨 **不要**在 `main` 分支工作!
+
+4. 在虚拟环境中运行以下命令来设置开发环境：
+
+   ```bash
+   pip install -e ".[dev]"
+   ```
+
+   如果在虚拟环境中已经安装了 🤗 Transformers，请先使用 `pip uninstall transformers` 卸载它，然后再用 `-e` 参数以可编辑模式重新安装。
+   
+   根据你的操作系统，以及 Transformers 的可选依赖项数量的增加，可能会在执行此命令时出现失败。如果出现这种情况，请确保已经安装了你想使用的深度学习框架（PyTorch, TensorFlow 和 Flax），然后执行以下操作：
+
+   ```bash
+   pip install -e ".[quality]"
+   ```
+
+   大多数情况下，这些应该够用了。
+
+5. 在你的分支上开发相关功能。
+
+   在编写代码时，请确保测试套件通过。用下面的方式运行受你的更改影响的测试：
+
+   ```bash
+   pytest tests/<TEST_TO_RUN>.py
+   ```
+
+   想了解更多关于测试的信息，请阅读[测试](https://huggingface.co/docs/transformers/testing)指南。
+
+   🤗 Transformers 使用 `black` 和 `ruff` 来保持代码风格的一致性。进行更改后，使用以下命令自动执行格式更正和代码验证：
+
+   ```bash
+   make fixup
+   ```
+
+   它已经被优化为仅适用于你创建的 PR 所修改过的文件。
+
+   如果想要逐个运行检查，可以使用以下命令：
+
+   ```bash
+   make style
+   ```
+
+   🤗 Transformers 还使用了 `ruff` 和一些自定义脚本来检查编码错误。虽然质量管理是通过 CI 进行的，但你也可以使用以下命令来运行相同的检查：
+
+   ```bash
+   make quality
+   ```
+
+   最后，我们有许多脚本来确保在添加新模型时不会忘记更新某些文件。你可以使用以下命令运行这些脚本：
+
+   ```bash
+   make repo-consistency
+   ```
+
+   想要了解有关这些检查及如何解决相关问题的更多信息，请阅读 [检查 Pull Request](https://huggingface.co/docs/transformers/pr_checks) 指南。
+
+   如果你修改了 `docs/source` 目录下的文档，请确保文档仍然能够被构建。这个检查也会在你创建 PR 时在 CI 中运行。如果要进行本地检查，请确保安装了文档构建工具：
+   
+   ```bash
+   pip install ".[docs]"
+   ```
+
+   在仓库的根目录下运行以下命令：
+
+   ```bash
+   doc-builder build transformers docs/source/en --build_dir ~/tmp/test-build
+   ```
+
+   这将会在 `~/tmp/test-build` 文件夹中构建文档，你可以使用自己喜欢的编辑器查看生成的 Markdown 文件。当你创建 PR 时，也可以在GitHub上预览文档。
+
+   当你对修改满意后，使用 `git add` 把修改的文件添加到暂存区，然后使用 `git commit` 在本地记录你的更改:
+
+   ```bash
+   git add modified_file.py
+   git commit
+   ```
+
+   请记得写一个[好的提交信息](https://chris.beams.io/posts/git-commit/)来清晰地传达你所做的更改！
+
+   为了保持你的代码副本与原始仓库的最新状态一致，在你创建 PR *之前*或者在管理员要求的情况下，把你的分支在 `upstream/branch` 上进行 rebase：
+
+   ```bash
+   git fetch upstream
+   git rebase upstream/main
+   ```
+
+   把你的更改推送到你的分支：
+
+   ```bash
+   git push -u origin a-descriptive-name-for-my-changes
+   ```
+
+   如果你已经创建了一个 PR，你需要使用 `--force` 参数进行强制推送。如果 PR 还没有被创建，你可以正常推送你的更改。
+
+6. 现在你可以转到 GitHub 上你的账号下的派生仓库，点击 **Pull Request** 来创建一个 PR。 请确保勾选我们 [checklist](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md/#pull-request-checklist) 下的所有项目。准备好这些后，可以将你的更改发送给项目管理员进行审查。
+
+7. 如果管理员要求你进行更改，别气馁，我们的核心贡献者也会经历相同的事情！请在你的本地分支上进行工作，并将更改推送到派生仓库，以便于每个人都可以在 PR 中看到你的更改。这样它们会自动出现在 PR 中。
+
+### Pull request 的检查清单
+
+☐ Pull request 的标题应该总结你的贡献内容。<br>
+☐ 如果你的 Pull request 解决了一个issue，请在 Pull request 描述中提及该 issue 的编号，以确保它们被关联起来（这样查看 issue 的人就知道你正在处理它）。<br>
+☐ 如果是正在进行中的工作，请在标题前加上 [WIP]。这有助于避免重复工作和区分哪些 PR 可以合并。<br>
+☐ 确保可以通过现有的测试。<br>
+☐ 如果添加了新功能，请同时添加对应的测试。<br>
+   - 如果添加一个新模型，请使用 `ModelTester.all_model_classes = (MyModel, MyModelWithLMHead,...)` 来触发通用测试。
+   - 如果你正在添加新的 `@slow` 测试，请确保通过以下检查：`RUN_SLOW=1 python -m pytest tests/models/my_new_model/test_my_new_model.py`
+   - 如果你正在添加一个新的分词器，请编写测试并确保通过以下检查：`RUN_SLOW=1 python -m pytest tests/models/{your_model_name}/test_tokenization_{your_model_name}.py`
+   - CircleCI 不会运行时间较长的测试，但 GitHub Actions 每晚会运行所有测试！<br>
+
+☐ 所有公共 method 必须具有信息文档（比如 [`modeling_bert.py`](https://github.com/huggingface/transformers/blob/main/src/transformers/models/bert/modeling_bert.py)）。<br>
+☐ 由于代码仓库的体积正在迅速增长，请避免添加图像、视频和其他非文本文件，它们会增加仓库的负担。请使用 [`hf-internal-testing`](https://huggingface.co/hf-internal-testing) 等 Hub 仓库来托管这些文件，并通过 URL 引用它们。我们建议将与文档相关的图片放置在以下仓库中：[huggingface/documentation-images](https://huggingface.co/datasets/huggingface/documentation-images)。你可以在这个数据集仓库上创建一个 PR，并请求 Hugging Face 成员进行合并。
+
+要了解更多有关在 Pull request 上运行的检查的信息，请查看我们的 [检查 Pull Request](https://huggingface.co/docs/transformers/pr_checks) 指南。
+
+### 测试
+
+包含了广泛的测试套件来测试库的行为和一些示例。库测试可以在 [tests](https://github.com/huggingface/transformers/tree/main/tests) 文件夹中找到，示例测试可以在 [examples](https://github.com/huggingface/transformers/tree/main/examples) 文件夹中找到。
+
+我们喜欢使用 `pytest` 和 `pytest-xdist`，因为它运行更快。在仓库的根目录，指定一个*子文件夹的路径或测试文件*来运行测试。
+
+```bash
+python -m pytest -n auto --dist=loadfile -s -v ./tests/models/my_new_model
+```
+
+同样地，在 `examples` 目录，指定一个*子文件夹的路径或测试文件* 来运行测试。例如，以下命令会测试 PyTorch `examples` 目录中的文本分类子文件夹：
+
+```bash
+pip install -r examples/xxx/requirements.txt  # 仅在第一次需要
+python -m pytest -n auto --dist=loadfile -s -v ./examples/pytorch/text-classification
+```
+
+实际上这就是我们的 `make test` 和 `make test-examples` 命令的实现方式（不包括 `pip install`）！
+
+你也可以指定一个较小的测试集来仅测试特定功能。
+
+默认情况下，会跳过时间较长的测试，但你可以将 `RUN_SLOW` 环境变量设置为 `yes` 来运行它们。这将下载以 GB 为单位的模型文件，所以确保你有足够的磁盘空间、良好的网络连接和足够的耐心！
+
+<Tip warning={true}>
+
+记得指定一个*子文件夹的路径或测试文件*来运行测试。否则你将会运行 `tests` 或 `examples` 文件夹中的所有测试，它会花费很长时间！
+
+</Tip>
+
+```bash
+RUN_SLOW=yes python -m pytest -n auto --dist=loadfile -s -v ./tests/models/my_new_model
+RUN_SLOW=yes python -m pytest -n auto --dist=loadfile -s -v ./examples/pytorch/text-classification
+```
+
+和时间较长的测试一样，还有其他环境变量在测试过程中，在默认情况下是未启用的：
+- `RUN_CUSTOM_TOKENIZERS`: 启用自定义分词器的测试。
+- `RUN_PT_FLAX_CROSS_TESTS`: 启用 PyTorch + Flax 整合的测试。
+- `RUN_PT_TF_CROSS_TESTS`: 启用 TensorFlow + PyTorch 整合的测试。
+
+更多环境变量和额外信息可以在 [testing_utils.py](src/transformers/testing_utils.py) 中找到。
+
+🤗 Transformers 只是使用 `pytest` 作为测试运行程序，但测试套件本身没用任何与 `pytest` 相关的功能。
+
+这意味着完全支持 `unittest` 。以下是如何使用 `unittest` 运行测试的方法：
+
+```bash
+python -m unittest discover -s tests -t . -v
+python -m unittest discover -s examples -t examples -v
+```
+
+### 风格指南
+
+🤗 Transformers 的文档遵循 [Google Python Style Guide](https://google.github.io/styleguide/pyguide.html)。请查看我们的 [文档编写指南](https://github.com/huggingface/transformers/tree/main/docs#writing-documentation---specification) 来获取更多信息。
+
+### 在 Windows 上开发
+
+在 Windows 上（除非你正在使用 [Windows Subsystem for Linux](https://learn.microsoft.com/en-us/windows/wsl/) 或 WSL），你需要配置 git 将 Windows 的 `CRLF` 行结束符转换为 Linux 的 `LF` 行结束符：
+
+```bash
+git config core.autocrlf input
+```
+
+在 Windows 上有一种方法可以运行 `make` 命令，那就是使用 MSYS2：
+
+1. [下载 MSYS2](https://www.msys2.org/)，假设已经安装在 `C:\msys64`。
+2. 从命令行打开 `C:\msys64\msys2.exe` （可以在 **开始** 菜单中找到）。
+3. 在 shell 中运行： `pacman -Syu` ，并使用 `pacman -S make` 安装 `make`。
+4. 把 `C:\msys64\usr\bin` 添加到你的 PATH 环境变量中。
+
+现在你可以在任何终端（Powershell、cmd.exe 等）中使用 `make` 命令了！ 🎉
+
+### 将派生仓库与上游主仓库（Hugging Face 仓库）同步
+
+更新派生仓库的主分支时，请按照以下步骤操作。这是为了避免向每个上游 PR 添加参考注释，同时避免向参与这些 PR 的开发人员发送不必要的通知。
+
+1. 可以的话，请避免使用派生仓库上的分支和 PR 来与上游进行同步，而是直接合并到派生仓库的主分支。
+2. 如果确实需要一个 PR，在检查你的分支后，请按照以下步骤操作：
+
+```bash
+git checkout -b your-branch-for-syncing
+git pull --squash --no-commit upstream main
+git commit -m '<your message without GitHub references>'
+git push --set-upstream origin your-branch-for-syncing
+```

docs/source/zh/installation.md

@@ -70,7 +70,7 @@ pip install 'transformers[tf-cpu]'
 <Tip warning={true}>
 
 M1 / ARM用户
-    
+
 在安装 TensorFlow 2.0 前，你需要安装以下库：
 ```
 brew install cmake
@@ -147,10 +147,10 @@ git pull
 
 ## 使用 conda 安装
 
-从 conda 的 `huggingface` 频道安装：
+从 conda 的 `conda-forge` 频道安装：
 
 ```bash
-conda install -c huggingface transformers
+conda install conda-forge::transformers
 ```
 
 ## 缓存设置

docs/source/zh/internal/generation_utils.md

@@ -43,7 +43,7 @@ inputs = tokenizer("Hello, my dog is cute and ", return_tensors="pt")
 generation_output = model.generate(**inputs, return_dict_in_generate=True, output_scores=True)
 ```
 
-`generation_output` 的对象是 [`~generation.GreedySearchDecoderOnlyOutput`] 的一个实例，从该类的文档中我们可以看到，这意味着它具有以下属性：
+`generation_output` 的对象是 [`~generation.GenerateDecoderOnlyOutput`] 的一个实例，从该类的文档中我们可以看到，这意味着它具有以下属性：
 
 - `sequences`: 生成的tokens序列
 - `scores`（可选）: 每个生成步骤的语言建模头的预测分数
@@ -70,25 +70,13 @@ generation_output[:2]
 
 ### PyTorch
 
-[[autodoc]] generation.GreedySearchEncoderDecoderOutput
+[[autodoc]] generation.GenerateDecoderOnlyOutput
 
-[[autodoc]] generation.GreedySearchDecoderOnlyOutput
+[[autodoc]] generation.GenerateEncoderDecoderOutput
 
-[[autodoc]] generation.SampleEncoderDecoderOutput
+[[autodoc]] generation.GenerateBeamDecoderOnlyOutput
 
-[[autodoc]] generation.SampleDecoderOnlyOutput
-
-[[autodoc]] generation.BeamSearchEncoderDecoderOutput
-
-[[autodoc]] generation.BeamSearchDecoderOnlyOutput
-
-[[autodoc]] generation.BeamSampleEncoderDecoderOutput
-
-[[autodoc]] generation.BeamSampleDecoderOnlyOutput
-
-[[autodoc]] generation.ContrastiveSearchEncoderDecoderOutput
-
-[[autodoc]] generation.ContrastiveSearchDecoderOnlyOutput
+[[autodoc]] generation.GenerateBeamEncoderDecoderOutput
 
 ### TensorFlow
 

docs/source/zh/main_classes/deepspeed.md

@@ -249,7 +249,7 @@ recommend ZeRO-3 config as starting one. -->
 注意：
 
 - 如果您需要在特定的 GPU 上运行，而不是 GPU 0，则无法使用 `CUDA_VISIBLE_DEVICES` 来限制可用 GPU 的可见范围。相反，您必须使用以下语法：
-  
+
   ```bash
   deepspeed --include localhost:1 examples/pytorch/translation/run_translation.py ...
   ```
@@ -1845,7 +1845,6 @@ SW: Model with 2783M total params, 65M largest layer params.
 
 ### 注意事项
 
-- DeepSpeed 与 PyTorch [`Trainer`] 一起工作，但不与 TF [`TFTrainer`] 一起工作。
 - 尽管 DeepSpeed 有一个可安装的 PyPI 包，但强烈建议从源代码安装它，以最好地匹配您的硬件，如果您需要启用某些功能，如 1-bit Adam，这些功能在 pypi 发行版中不可用。
 - 您不必使用🤗  Transformers的 [`Trainer`] 来使用 DeepSpeed   - 您可以使用任何模型与自己的训练器，您还需要根据 [DeepSpeed 集成说明](https://www.deepspeed.ai/getting-started/#writing-deepspeed-models) 调整后者。
 

examples/flax/conftest.py

@@ -21,7 +21,7 @@ from os.path import abspath, dirname, join
 
 
 # allow having multiple repository checkouts and not needing to remember to rerun
-# 'pip install -e .[dev]' when switching between checkouts and running tests.
+# `pip install -e '.[dev]'` when switching between checkouts and running tests.
 git_repo_path = abspath(join(dirname(dirname(dirname(__file__))), "src"))
 sys.path.insert(1, git_repo_path)
 

examples/flax/image-captioning/run_image_captioning_flax.py

@@ -853,7 +853,7 @@ def main():
                 yield batch
 
     # Metric
-    metric = evaluate.load("rouge")
+    metric = evaluate.load("rouge", cache_dir=model_args.cache_dir)
 
     def postprocess_text(preds, labels):
         preds = [pred.strip() for pred in preds]

examples/flax/question-answering/run_qa.py

@@ -62,7 +62,7 @@ from transformers.utils import check_min_version, send_example_telemetry
 logger = logging.getLogger(__name__)
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.37.0.dev0")
+check_min_version("4.37.0")
 
 Array = Any
 Dataset = datasets.arrow_dataset.Dataset
@@ -807,7 +807,9 @@ def main():
         references = [{"id": ex["id"], "answers": ex[answer_column_name]} for ex in examples]
         return EvalPrediction(predictions=formatted_predictions, label_ids=references)
 
-    metric = evaluate.load("squad_v2" if data_args.version_2_with_negative else "squad")
+    metric = evaluate.load(
+        "squad_v2" if data_args.version_2_with_negative else "squad", cache_dir=model_args.cache_dir
+    )
 
     def compute_metrics(p: EvalPrediction):
         return metric.compute(predictions=p.predictions, references=p.label_ids)

examples/flax/speech-recognition/run_flax_speech_recognition_seq2seq.py

@@ -60,7 +60,7 @@ from transformers.utils.versions import require_version
 
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risk.
-check_min_version("4.37.0.dev0")
+check_min_version("4.37.0")
 
 require_version("datasets>=2.14.0", "To fix: pip install -r examples/flax/speech-recogintion/requirements.txt")
 
@@ -577,7 +577,7 @@ def main():
         return
 
     # 8. Load Metric
-    metric = evaluate.load("wer")
+    metric = evaluate.load("wer", cache_dir=model_args.cache_dir)
 
     def compute_metrics(preds, labels):
         # replace padded labels by the padding token

examples/flax/summarization/run_summarization_flax.py

@@ -710,7 +710,7 @@ def main():
         )
 
     # Metric
-    metric = evaluate.load("rouge")
+    metric = evaluate.load("rouge", cache_dir=model_args.cache_dir)
 
     def postprocess_text(preds, labels):
         preds = [pred.strip() for pred in preds]

examples/flax/text-classification/run_flax_glue.py

@@ -55,7 +55,7 @@ from transformers.utils import check_min_version, send_example_telemetry
 
 logger = logging.getLogger(__name__)
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
-check_min_version("4.37.0.dev0")
+check_min_version("4.37.0")
 
 Array = Any
 Dataset = datasets.arrow_dataset.Dataset
@@ -599,9 +599,9 @@ def main():
     p_eval_step = jax.pmap(eval_step, axis_name="batch")
 
     if data_args.task_name is not None:
-        metric = evaluate.load("glue", data_args.task_name)
+        metric = evaluate.load("glue", data_args.task_name, cache_dir=model_args.cache_dir)
     else:
-        metric = evaluate.load("accuracy")
+        metric = evaluate.load("accuracy", cache_dir=model_args.cache_dir)
 
     logger.info(f"===== Starting training ({num_epochs} epochs) =====")
     train_time = 0