Add Qwen2MoE (#29377)

* add support for qwen2 MoE models

* update docs

* add support for qwen2 MoE models

* update docs

* update model name & test

* update readme

* update class names & readme & model_doc of Qwen2MoE.

* update architecture name

* fix qwen2_moe tests

* use Qwen2Tokenizer instead of Qwen2MoeTokenizer

* update modeling_qwen2_moe.py

* fix model architecture

* fix qwen2_moe tests

* use Qwen2Tokenizer instead of Qwen2MoeTokenizer

* update modeling_qwen2_moe.py

* fix model architecture

* fix style

* fix test when there are sparse and non sparse layers

* fixup

* Update README.md

Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* fixup

* fixup

* add archive back

* add support for qwen2 MoE models

* update docs

* update model name & test

* update readme

* update class names & readme & model_doc of Qwen2MoE.

* update architecture name

* fix qwen2_moe tests

* use Qwen2Tokenizer instead of Qwen2MoeTokenizer

* update modeling_qwen2_moe.py

* fix model architecture

* fixup

* fix qwen2_moe tests

* use Qwen2Tokenizer instead of Qwen2MoeTokenizer

* fix style

* fix test when there are sparse and non sparse layers

* fixup

* add archive back

* fix integration test

* fixup

---------

Co-authored-by: bozheng-hit <dsoul0621@gmail.com>
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

README.md

@@ -473,6 +473,7 @@ Current number of checkpoints: ![](https://img.shields.io/endpoint?url=https://h
 1. **[PVTv2](https://huggingface.co/docs/transformers/model_doc/pvt_v2)** (from Shanghai AI Laboratory, Nanjing University, The University of Hong Kong etc.) released with the paper [PVT v2: Improved Baselines with Pyramid Vision Transformer](https://arxiv.org/abs/2106.13797) 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. **[Qwen2MoE](https://huggingface.co/docs/transformers/main/model_doc/qwen2_moe)** (from the Qwen team, Alibaba Group) released with [blog post](https://qwenlm.github.io/blog/qwen1.5/) by Bo Zheng, Dayiheng Liu, Rui Men, Junyang Lin, Zhou San, Bowen Yu, An Yang, Mingfeng Xue, Fei Huang, Binyuan Hui, Mei Li, Tianyu Liu, Xingzhang Ren, Xuancheng Ren, Kexin Yang, Chang Zhou, Jingren Zhou.
 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.

README_de.md

@@ -469,6 +469,7 @@ Aktuelle Anzahl der Checkpoints: ![](https://img.shields.io/endpoint?url=https:/
 1. **[PVTv2](https://huggingface.co/docs/transformers/model_doc/pvt_v2)** (from Shanghai AI Laboratory, Nanjing University, The University of Hong Kong etc.) released with the paper [PVT v2: Improved Baselines with Pyramid Vision Transformer](https://arxiv.org/abs/2106.13797) 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. **[Qwen2MoE](https://huggingface.co/docs/transformers/main/model_doc/qwen2_moe)** (from the Qwen team, Alibaba Group) released with the paper [blog post](https://qwenlm.github.io/blog/qwen1.5/) by Bo Zheng, Dayiheng Liu, Rui Men, Junyang Lin, Zhou San, Bowen Yu, An Yang, Mingfeng Xue, Fei Huang, Binyuan Hui, Mei Li, Tianyu Liu, Xingzhang Ren, Xuancheng Ren, Kexin Yang, Chang Zhou, Jingren Zhou.
 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.

README_es.md

@@ -446,6 +446,7 @@ Número actual de puntos de control: ![](https://img.shields.io/endpoint?url=htt
 1. **[PVTv2](https://huggingface.co/docs/transformers/model_doc/pvt_v2)** (from Shanghai AI Laboratory, Nanjing University, The University of Hong Kong etc.) released with the paper [PVT v2: Improved Baselines with Pyramid Vision Transformer](https://arxiv.org/abs/2106.13797) 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. **[Qwen2MoE](https://huggingface.co/docs/transformers/main/model_doc/qwen2_moe)** (from the Qwen team, Alibaba Group) released with the paper [blog post](https://qwenlm.github.io/blog/qwen1.5/) by Bo Zheng, Dayiheng Liu, Rui Men, Junyang Lin, Zhou San, Bowen Yu, An Yang, Mingfeng Xue, Fei Huang, Binyuan Hui, Mei Li, Tianyu Liu, Xingzhang Ren, Xuancheng Ren, Kexin Yang, Chang Zhou, Jingren Zhou.
 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.

README_fr.md

@@ -467,6 +467,7 @@ Nombre actuel de points de contrôle : ![](https://img.shields.io/endpoint?url=h
 1. **[PVTv2](https://huggingface.co/docs/transformers/model_doc/pvt_v2)** (de Shanghai AI Laboratory, Nanjing University, The University of Hong Kong etc.) publié dans l'article [PVT v2: Improved Baselines with Pyramid Vision Transformer](https://arxiv.org/abs/2106.13797) parWenhai 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)** (de NVIDIA) a été publié dans l'article [Integer Quantization for Deep Learning Inference: Principles and Empirical Evaluation](https://arxiv.org/abs/2004.09602) par Hao Wu, Patrick Judd, Xiaojie Zhang, Mikhail Isaev et Paulius Micikevicius.
 1. **[Qwen2](https://huggingface.co/docs/transformers/model_doc/qwen2)** (de l'équipe Qwen, Alibaba Group) a été publié avec le rapport technique [Qwen Technical Report](https://arxiv.org/abs/2309.16609) par 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 et Tianhang Zhu.
+1. **[Qwen2MoE](https://huggingface.co/docs/transformers/main/model_doc/qwen2_moe)** (de l'équipe Qwen, Alibaba Group) a été publié avec le rapport technique [blog post](https://qwenlm.github.io/blog/qwen1.5/) par Bo Zheng, Dayiheng Liu, Rui Men, Junyang Lin, Zhou San, Bowen Yu, An Yang, Mingfeng Xue, Fei Huang, Binyuan Hui, Mei Li, Tianyu Liu, Xingzhang Ren, Xuancheng Ren, Kexin Yang, Chang Zhou, Jingren Zhou.
 1. **[RAG](https://huggingface.co/docs/transformers/model_doc/rag)** (de Facebook) a été publié dans l'article [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://arxiv.org/abs/2005.11401) par 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)** (de Google Research) a été publié dans l'article [REALM: Retrieval-Augmented Language Model Pre-Training](https://arxiv.org/abs/2002.08909) par Kelvin Guu, Kenton Lee, Zora Tung, Panupong Pasupat et Ming-Wei Chang.
 1. **[Reformer](https://huggingface.co/docs/transformers/model_doc/reformer)** (de Google Research) a été publié dans l'article [Reformer: The Efficient Transformer](https://arxiv.org/abs/2001.04451) par Nikita Kitaev, Łukasz Kaiser, Anselm Levskaya.

README_hd.md

@@ -420,6 +420,7 @@ conda install conda-forge::transformers
 1. **[PVTv2](https://huggingface.co/docs/transformers/model_doc/pvt_v2)** (Shanghai AI Laboratory, 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 [PVT v2: Improved Baselines with Pyramid Vision Transformer](https://arxiv.org/abs/2106.13797) के साथ जारी किया गया 
 1. **[QDQBert](https://huggingface.co/docs/transformers/model_doc/qdqbert)** (NVIDIA से) साथ वाला पेपर [Integer Quantization for Deep Learning Inference: Principles and Empirical Evaluation](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. **[Qwen2MoE](https://huggingface.co/docs/transformers/main/model_doc/qwen2_moe)** (the Qwen team, Alibaba Group से) Bo Zheng, Dayiheng Liu, Rui Men, Junyang Lin, Zhou San, Bowen Yu, An Yang, Mingfeng Xue, Fei Huang, Binyuan Hui, Mei Li, Tianyu Liu, Xingzhang Ren, Xuancheng Ren, Kexin Yang, Chang Zhou, Jingren Zhou. द्वाराअनुसंधान पत्र [blog post](https://qwenlm.github.io/blog/qwen1.5/) के साथ जारी किया गया
 1. **[RAG](https://huggingface.co/docs/transformers/model_doc/rag)** (फेसबुक से) साथ में कागज [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](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: Retrieval-Augmented Language Model Pre-Training](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.

README_ja.md

@@ -480,6 +480,7 @@ Flax、PyTorch、TensorFlowをcondaでインストールする方法は、それ
 1. **[PVTv2](https://huggingface.co/docs/transformers/model_doc/pvt_v2)** (Shanghai AI Laboratory, 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 [PVT v2: Improved Baselines with Pyramid Vision Transformer](https://arxiv.org/abs/2106.13797) 
 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. **[Qwen2MoE](https://huggingface.co/docs/transformers/main/model_doc/qwen2_moe)** (the Qwen team, Alibaba Group から) Bo Zheng, Dayiheng Liu, Rui Men, Junyang Lin, Zhou San, Bowen Yu, An Yang, Mingfeng Xue, Fei Huang, Binyuan Hui, Mei Li, Tianyu Liu, Xingzhang Ren, Xuancheng Ren, Kexin Yang, Chang Zhou, Jingren Zhou. から公開された研究論文 [blog post](https://qwenlm.github.io/blog/qwen1.5/)
 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)

README_ko.md

@@ -395,6 +395,7 @@ Flax, PyTorch, TensorFlow 설치 페이지에서 이들을 conda로 설치하는
 1. **[PVTv2](https://huggingface.co/docs/transformers/model_doc/pvt_v2)** (Shanghai AI Laboratory, 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 [PVT v2: Improved Baselines with Pyramid Vision Transformer](https://arxiv.org/abs/2106.13797)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. **[Qwen2MoE](https://huggingface.co/docs/transformers/main/model_doc/qwen2_moe)** (the Qwen team, Alibaba Group 에서 제공)은 Bo Zheng, Dayiheng Liu, Rui Men, Junyang Lin, Zhou San, Bowen Yu, An Yang, Mingfeng Xue, Fei Huang, Binyuan Hui, Mei Li, Tianyu Liu, Xingzhang Ren, Xuancheng Ren, Kexin Yang, Chang Zhou, Jingren Zhou.의 [blog post](https://qwenlm.github.io/blog/qwen1.5/)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.

README_pt-br.md

@@ -478,6 +478,7 @@ Número atual de pontos de verificação: ![](https://img.shields.io/endpoint?ur
 1. **[PVTv2](https://huggingface.co/docs/transformers/model_doc/pvt_v2)** (from Shanghai AI Laboratory, Nanjing University, The University of Hong Kong etc.) released with the paper [PVT v2: Improved Baselines with Pyramid Vision Transformer](https://arxiv.org/abs/2106.13797) 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. **[Qwen2MoE](https://huggingface.co/docs/transformers/main/model_doc/qwen2_moe)** (from the Qwen team, Alibaba Group) released with the paper [blog post](https://qwenlm.github.io/blog/qwen1.5/) by Bo Zheng, Dayiheng Liu, Rui Men, Junyang Lin, Zhou San, Bowen Yu, An Yang, Mingfeng Xue, Fei Huang, Binyuan Hui, Mei Li, Tianyu Liu, Xingzhang Ren, Xuancheng Ren, Kexin Yang, Chang Zhou, Jingren Zhou.
 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.

README_ru.md

@@ -468,6 +468,7 @@ conda install conda-forge::transformers
 1. **[PVTv2](https://huggingface.co/docs/transformers/model_doc/pvt_v2)** (from Shanghai AI Laboratory, Nanjing University, The University of Hong Kong etc.) released with the paper [PVT v2: Improved Baselines with Pyramid Vision Transformer](https://arxiv.org/abs/2106.13797) 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. **[Qwen2MoE](https://huggingface.co/docs/transformers/main/model_doc/qwen2_moe)** (from the Qwen team, Alibaba Group) released with the paper [blog post](https://qwenlm.github.io/blog/qwen1.5/) by Bo Zheng, Dayiheng Liu, Rui Men, Junyang Lin, Zhou San, Bowen Yu, An Yang, Mingfeng Xue, Fei Huang, Binyuan Hui, Mei Li, Tianyu Liu, Xingzhang Ren, Xuancheng Ren, Kexin Yang, Chang Zhou, Jingren Zhou.
 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.

README_te.md

@@ -470,6 +470,7 @@ Flax, PyTorch లేదా TensorFlow యొక్క ఇన్‌స్టా
 1. **[PVTv2](https://huggingface.co/docs/transformers/model_doc/pvt_v2)** (from Shanghai AI Laboratory, Nanjing University, The University of Hong Kong etc.) released with the paper [PVT v2: Improved Baselines with Pyramid Vision Transformer](https://arxiv.org/abs/2106.13797) 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. **[Qwen2MoE](https://huggingface.co/docs/transformers/main/model_doc/qwen2_moe)** (from the Qwen team, Alibaba Group) released with the paper [blog post](https://qwenlm.github.io/blog/qwen1.5/) by Bo Zheng, Dayiheng Liu, Rui Men, Junyang Lin, Zhou San, Bowen Yu, An Yang, Mingfeng Xue, Fei Huang, Binyuan Hui, Mei Li, Tianyu Liu, Xingzhang Ren, Xuancheng Ren, Kexin Yang, Chang Zhou, Jingren Zhou.
 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.

README_vi.md

@@ -469,6 +469,7 @@ Số lượng điểm kiểm tra hiện tại: ![](https://img.shields.io/endpoi
 1. **[PVTv2](https://huggingface.co/docs/transformers/model_doc/pvt_v2)** (từ Shanghai AI Laboratory, Nanjing University, The University of Hong Kong etc.) được phát hành với bài báo [PVT v2: Improved Baselines with Pyramid Vision Transformer](https://arxiv.org/abs/2106.13797) 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)** (từ NVIDIA) được phát hành với bài báo [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)** (từ the Qwen team, Alibaba Group) được phát hành với bài báo [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. **[Qwen2MoE](https://huggingface.co/docs/transformers/main/model_doc/qwen2_moe)** (từ the Qwen team, Alibaba Group) được phát hành với bài báo [blog post](https://qwenlm.github.io/blog/qwen1.5/) by Bo Zheng, Dayiheng Liu, Rui Men, Junyang Lin, Zhou San, Bowen Yu, An Yang, Mingfeng Xue, Fei Huang, Binyuan Hui, Mei Li, Tianyu Liu, Xingzhang Ren, Xuancheng Ren, Kexin Yang, Chang Zhou, Jingren Zhou.
 1. **[RAG](https://huggingface.co/docs/transformers/model_doc/rag)** (từ Facebook) được phát hành với bài báo [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)** (từ Google Research) được phát hành với bài báo [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)** (từ Google Research) được phát hành với bài báo [Reformer: The Efficient Transformer](https://arxiv.org/abs/2001.04451) by Nikita Kitaev, Łukasz Kaiser, Anselm Levskaya.

README_zh-hans.md

@@ -419,6 +419,7 @@ conda install conda-forge::transformers
 1. **[PVTv2](https://huggingface.co/docs/transformers/model_doc/pvt_v2)** (来自 Shanghai AI Laboratory, Nanjing University, The University of Hong Kong etc.) 伴随论文 [PVT v2: Improved Baselines with Pyramid Vision Transformer](https://arxiv.org/abs/2106.13797) 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. **[Qwen2MoE](https://huggingface.co/docs/transformers/main/model_doc/qwen2_moe)** (来自 the Qwen team, Alibaba Group) 伴随论文 [blog post](https://qwenlm.github.io/blog/qwen1.5/) by Bo Zheng, Dayiheng Liu, Rui Men, Junyang Lin, Zhou San, Bowen Yu, An Yang, Mingfeng Xue, Fei Huang, Binyuan Hui, Mei Li, Tianyu Liu, Xingzhang Ren, Xuancheng Ren, Kexin Yang, Chang Zhou, Jingren Zhou 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。

README_zh-hant.md

@@ -431,6 +431,7 @@ conda install conda-forge::transformers
 1. **[PVTv2](https://huggingface.co/docs/transformers/model_doc/pvt_v2)** (from Shanghai AI Laboratory, Nanjing University, The University of Hong Kong etc.) released with the paper [PVT v2: Improved Baselines with Pyramid Vision Transformer](https://arxiv.org/abs/2106.13797) 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. **[Qwen2MoE](https://huggingface.co/docs/transformers/main/model_doc/qwen2_moe)** (from the Qwen team, Alibaba Group) released with the paper [blog post](https://qwenlm.github.io/blog/qwen1.5/) by Bo Zheng, Dayiheng Liu, Rui Men, Junyang Lin, Zhou San, Bowen Yu, An Yang, Mingfeng Xue, Fei Huang, Binyuan Hui, Mei Li, Tianyu Liu, Xingzhang Ren, Xuancheng Ren, Kexin Yang, Chang Zhou, Jingren Zhou.
 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.

docs/source/en/_toctree.yml

@@ -462,6 +462,8 @@
         title: QDQBert
       - local: model_doc/qwen2
         title: Qwen2
+      - local: model_doc/qwen2_moe
+        title: Qwen2MoE
       - local: model_doc/rag
         title: RAG
       - local: model_doc/realm

docs/source/en/index.md

@@ -240,6 +240,7 @@ Flax), PyTorch, and/or TensorFlow.
 |                        [PVTv2](model_doc/pvt_v2)                         |       ✅        |         ❌         |      ❌      |
 |                       [QDQBert](model_doc/qdqbert)                       |       ✅        |         ❌         |      ❌      |
 |                         [Qwen2](model_doc/qwen2)                         |       ✅        |         ❌         |      ❌      |
+|                     [Qwen2MoE](model_doc/qwen2_moe)                      |       ✅        |         ❌         |      ❌      |
 |                           [RAG](model_doc/rag)                           |       ✅        |         ✅         |      ❌      |
 |                         [REALM](model_doc/realm)                         |       ✅        |         ❌         |      ❌      |
 |                      [Reformer](model_doc/reformer)                      |       ✅        |         ❌         |      ❌      |

docs/source/en/model_doc/qwen2_moe.md

@@ -0,0 +1,77 @@
+<!--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.
+
+-->
+
+# Qwen2MoE
+
+## Overview
+
+Qwen2MoE 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
+
+Qwen2MoE 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. Qwen2MoE has the following architectural choices:
+
+- Qwen2MoE 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.
+- Qwen2MoE employs Mixture of Experts (MoE) architecture, where the models are upcycled from dense language models. For instance, `Qwen1.5-MoE-A2.7B` is upcycled from `Qwen-1.8B`. It has 14.3B parameters in total and 2.7B activated parameters during runtime, while it achieves comparable performance with `Qwen1.5-7B`, with only 20% of the training resources.
+
+For more details refer to the [release blog post](https://qwenlm.github.io/blog/qwen1.5/).
+
+## Usage tips
+
+`Qwen1.5-MoE-A2.7B` and `Qwen1.5-MoE-A2.7B-Chat` can be found on the [Huggingface Hub](https://huggingface.co/Qwen)
+
+In the following, we demonstrate how to use `Qwen1.5-MoE-A2.7B-Chat` 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("Qwen/Qwen1.5-MoE-A2.7B-Chat", device_map="auto")
+>>> tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen1.5-MoE-A2.7B-Chat")
+
+>>> 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]
+```
+
+## Qwen2MoeConfig
+
+[[autodoc]] Qwen2MoeConfig
+
+## Qwen2MoeModel
+
+[[autodoc]] Qwen2MoeModel
+    - forward
+
+## Qwen2MoeForCausalLM
+
+[[autodoc]] Qwen2MoeForCausalLM
+    - forward
+
+## Qwen2MoeForSequenceClassification
+
+[[autodoc]] Qwen2MoeForSequenceClassification
+    - forward

docs/source/en/perf_infer_gpu_one.md

@@ -59,6 +59,7 @@ FlashAttention-2 is currently supported for the following architectures:
 * [StableLm](https://huggingface.co/docs/transformers/model_doc/stablelm#transformers.StableLmModel)
 * [Starcoder2](https://huggingface.co/docs/transformers/model_doc/starcoder2#transformers.Starcoder2Model)
 * [Qwen2](https://huggingface.co/docs/transformers/model_doc/qwen2#transformers.Qwen2Model)
+* [Qwen2MoE](https://huggingface.co/docs/transformers/model_doc/qwen2_moe#transformers.Qwen2MoeModel)
 * [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.
@@ -187,6 +188,7 @@ For now, Transformers supports SDPA inference and training for the following arc
 * [StableLm](https://huggingface.co/docs/transformers/model_doc/stablelm#transformers.StableLmModel)
 * [Starcoder2](https://huggingface.co/docs/transformers/model_doc/starcoder2#transformers.Starcoder2Model)
 * [Qwen2](https://huggingface.co/docs/transformers/model_doc/qwen2#transformers.Qwen2Model)
+* [Qwen2MoE](https://huggingface.co/docs/transformers/model_doc/qwen2_moe#transformers.Qwen2MoeModel)
 
 <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), [Cohere](../model_doc/cohere), [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), [Gemma](../model_doc/gemma), [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), [Mamba](../model_doc/mamba), [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), [MusicGen Melody](../model_doc/musicgen_melody), [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), [StableLm](../model_doc/stablelm), [Starcoder2](../model_doc/starcoder2), [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), [Cohere](../model_doc/cohere), [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), [Gemma](../model_doc/gemma), [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), [Mamba](../model_doc/mamba), [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), [MusicGen Melody](../model_doc/musicgen_melody), [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), [Qwen2MoE](../model_doc/qwen2_moe), [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), [StableLm](../model_doc/stablelm), [Starcoder2](../model_doc/starcoder2), [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/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), [Gemma](../model_doc/gemma), [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), [StableLm](../model_doc/stablelm), [Starcoder2](../model_doc/starcoder2), [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), [Gemma](../model_doc/gemma), [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), [Qwen2MoE](../model_doc/qwen2_moe), [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), [StableLm](../model_doc/stablelm), [Starcoder2](../model_doc/starcoder2), [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)
 
 
 

src/transformers/__init__.py

@@ -733,6 +733,10 @@ _import_structure = {
         "Qwen2Config",
         "Qwen2Tokenizer",
     ],
+    "models.qwen2_moe": [
+        "QWEN2MOE_PRETRAINED_CONFIG_ARCHIVE_MAP",
+        "Qwen2MoeConfig",
+    ],
     "models.rag": ["RagConfig", "RagRetriever", "RagTokenizer"],
     "models.realm": [
         "REALM_PRETRAINED_CONFIG_ARCHIVE_MAP",
@@ -3081,6 +3085,14 @@ else:
             "Qwen2PreTrainedModel",
         ]
     )
+    _import_structure["models.qwen2_moe"].extend(
+        [
+            "Qwen2MoeForCausalLM",
+            "Qwen2MoeForSequenceClassification",
+            "Qwen2MoeModel",
+            "Qwen2MoePreTrainedModel",
+        ]
+    )
     _import_structure["models.rag"].extend(
         [
             "RagModel",
@@ -5605,6 +5617,7 @@ if TYPE_CHECKING:
     from .models.pvt_v2 import PvtV2Config
     from .models.qdqbert import QDQBERT_PRETRAINED_CONFIG_ARCHIVE_MAP, QDQBertConfig
     from .models.qwen2 import QWEN2_PRETRAINED_CONFIG_ARCHIVE_MAP, Qwen2Config, Qwen2Tokenizer
+    from .models.qwen2_moe import QWEN2MOE_PRETRAINED_CONFIG_ARCHIVE_MAP, Qwen2MoeConfig
     from .models.rag import RagConfig, RagRetriever, RagTokenizer
     from .models.realm import (
         REALM_PRETRAINED_CONFIG_ARCHIVE_MAP,
@@ -7649,6 +7662,12 @@ if TYPE_CHECKING:
             Qwen2Model,
             Qwen2PreTrainedModel,
         )
+        from .models.qwen2_moe import (
+            Qwen2MoeForCausalLM,
+            Qwen2MoeForSequenceClassification,
+            Qwen2MoeModel,
+            Qwen2MoePreTrainedModel,
+        )
         from .models.rag import (
             RagModel,
             RagPreTrainedModel,

src/transformers/models/__init__.py

@@ -184,6 +184,7 @@ from . import (
     pvt_v2,
     qdqbert,
     qwen2,
+    qwen2_moe,
     rag,
     realm,
     reformer,

src/transformers/models/auto/configuration_auto.py

@@ -195,6 +195,7 @@ CONFIG_MAPPING_NAMES = OrderedDict(
         ("pvt_v2", "PvtV2Config"),
         ("qdqbert", "QDQBertConfig"),
         ("qwen2", "Qwen2Config"),
+        ("qwen2_moe", "Qwen2MoeConfig"),
         ("rag", "RagConfig"),
         ("realm", "RealmConfig"),
         ("reformer", "ReformerConfig"),
@@ -467,6 +468,7 @@ MODEL_NAMES_MAPPING = OrderedDict(
         ("pvt_v2", "PVTv2"),
         ("qdqbert", "QDQBert"),
         ("qwen2", "Qwen2"),
+        ("qwen2_moe", "Qwen2MoE"),
         ("rag", "RAG"),
         ("realm", "REALM"),
         ("reformer", "Reformer"),

src/transformers/models/auto/modeling_auto.py

@@ -182,6 +182,7 @@ MODEL_MAPPING_NAMES = OrderedDict(
         ("pvt_v2", "PvtV2Model"),
         ("qdqbert", "QDQBertModel"),
         ("qwen2", "Qwen2Model"),
+        ("qwen2_moe", "Qwen2MoeModel"),
         ("reformer", "ReformerModel"),
         ("regnet", "RegNetModel"),
         ("rembert", "RemBertModel"),
@@ -467,6 +468,7 @@ MODEL_FOR_CAUSAL_LM_MAPPING_NAMES = OrderedDict(
         ("prophetnet", "ProphetNetForCausalLM"),
         ("qdqbert", "QDQBertLMHeadModel"),
         ("qwen2", "Qwen2ForCausalLM"),
+        ("qwen2_moe", "Qwen2MoeForCausalLM"),
         ("reformer", "ReformerModelWithLMHead"),
         ("rembert", "RemBertForCausalLM"),
         ("roberta", "RobertaForCausalLM"),
@@ -871,6 +873,7 @@ MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING_NAMES = OrderedDict(
         ("plbart", "PLBartForSequenceClassification"),
         ("qdqbert", "QDQBertForSequenceClassification"),
         ("qwen2", "Qwen2ForSequenceClassification"),
+        ("qwen2_moe", "Qwen2MoeForSequenceClassification"),
         ("reformer", "ReformerForSequenceClassification"),
         ("rembert", "RemBertForSequenceClassification"),
         ("roberta", "RobertaForSequenceClassification"),

src/transformers/models/auto/tokenization_auto.py

@@ -354,6 +354,13 @@ else:
                     "Qwen2TokenizerFast" if is_tokenizers_available() else None,
                 ),
             ),
+            (
+                "qwen2_moe",
+                (
+                    "Qwen2Tokenizer",
+                    "Qwen2TokenizerFast" if is_tokenizers_available() else None,
+                ),
+            ),
             ("rag", ("RagTokenizer", None)),
             ("realm", ("RealmTokenizer", "RealmTokenizerFast" if is_tokenizers_available() else None)),
             (

src/transformers/models/qwen2_moe/__init__.py

@@ -0,0 +1,62 @@
+# Copyright 2024 The Qwen Team and The HuggingFace Inc. team. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+from typing import TYPE_CHECKING
+
+from ...utils import (
+    OptionalDependencyNotAvailable,
+    _LazyModule,
+    is_torch_available,
+)
+
+
+_import_structure = {
+    "configuration_qwen2_moe": ["QWEN2MOE_PRETRAINED_CONFIG_ARCHIVE_MAP", "Qwen2MoeConfig"],
+}
+
+
+try:
+    if not is_torch_available():
+        raise OptionalDependencyNotAvailable()
+except OptionalDependencyNotAvailable:
+    pass
+else:
+    _import_structure["modeling_qwen2_moe"] = [
+        "Qwen2MoeForCausalLM",
+        "Qwen2MoeModel",
+        "Qwen2MoePreTrainedModel",
+        "Qwen2MoeForSequenceClassification",
+    ]
+
+
+if TYPE_CHECKING:
+    from .configuration_qwen2_moe import QWEN2MOE_PRETRAINED_CONFIG_ARCHIVE_MAP, Qwen2MoeConfig
+
+    try:
+        if not is_torch_available():
+            raise OptionalDependencyNotAvailable()
+    except OptionalDependencyNotAvailable:
+        pass
+    else:
+        from .modeling_qwen2_moe import (
+            Qwen2MoeForCausalLM,
+            Qwen2MoeForSequenceClassification,
+            Qwen2MoeModel,
+            Qwen2MoePreTrainedModel,
+        )
+
+
+else:
+    import sys
+
+    sys.modules[__name__] = _LazyModule(__name__, globals()["__file__"], _import_structure, module_spec=__spec__)

src/transformers/models/qwen2_moe/configuration_qwen2_moe.py

@@ -0,0 +1,175 @@
+# coding=utf-8
+# Copyright 2024 The Qwen team, Alibaba Group and the HuggingFace Inc. team. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+""" Qwen2MoE model configuration"""
+
+from ...configuration_utils import PretrainedConfig
+from ...utils import logging
+
+
+logger = logging.get_logger(__name__)
+
+QWEN2MOE_PRETRAINED_CONFIG_ARCHIVE_MAP = {
+    "Qwen/Qwen1.5-MoE-A2.7B": "https://huggingface.co/Qwen/Qwen1.5-MoE-A2.7B/resolve/main/config.json",
+}
+
+
+class Qwen2MoeConfig(PretrainedConfig):
+    r"""
+    This is the configuration class to store the configuration of a [`Qwen2MoeModel`]. It is used to instantiate a
+    Qwen2MoE model according to the specified arguments, defining the model architecture. Instantiating a configuration
+    with the defaults will yield a similar configuration to that of
+    Qwen1.5-MoE-A2.7B" [Qwen/Qwen1.5-MoE-A2.7B"](https://huggingface.co/Qwen/Qwen1.5-MoE-A2.7B").
+
+    Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
+    documentation from [`PretrainedConfig`] for more information.
+
+
+    Args:
+        vocab_size (`int`, *optional*, defaults to 151936):
+            Vocabulary size of the Qwen2MoE model. Defines the number of different tokens that can be represented by the
+            `inputs_ids` passed when calling [`Qwen2MoeModel`]
+        hidden_size (`int`, *optional*, defaults to 2048):
+            Dimension of the hidden representations.
+        intermediate_size (`int`, *optional*, defaults to 5632):
+            Dimension of the MLP representations.
+        num_hidden_layers (`int`, *optional*, defaults to 24):
+            Number of hidden layers in the Transformer encoder.
+        num_attention_heads (`int`, *optional*, defaults to 16):
+            Number of attention heads for each attention layer in the Transformer encoder.
+        num_key_value_heads (`int`, *optional*, defaults to 16):
+            This is the number of key_value heads that should be used to implement Grouped Query Attention. If
+            `num_key_value_heads=num_attention_heads`, the model will use Multi Head Attention (MHA), if
+            `num_key_value_heads=1 the model will use Multi Query Attention (MQA) otherwise GQA is used. When
+            converting a multi-head checkpoint to a GQA checkpoint, each group key and value head should be constructed
+            by meanpooling all the original heads within that group. For more details checkout [this
+            paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to `32`.
+        hidden_act (`str` or `function`, *optional*, defaults to `"silu"`):
+            The non-linear activation function (function or string) in the decoder.
+        max_position_embeddings (`int`, *optional*, defaults to 32768):
+            The maximum sequence length that this model might ever be used with.
+        initializer_range (`float`, *optional*, defaults to 0.02):
+            The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
+        rms_norm_eps (`float`, *optional*, defaults to 1e-06):
+            The epsilon used by the rms normalization layers.
+        use_cache (`bool`, *optional*, defaults to `True`):
+            Whether or not the model should return the last key/values attentions (not used by all models). Only
+            relevant if `config.is_decoder=True`.
+        tie_word_embeddings (`bool`, *optional*, defaults to `False`):
+            Whether the model's input and output word embeddings should be tied.
+        rope_theta (`float`, *optional*, defaults to 10000.0):
+            The base period of the RoPE embeddings.
+        use_sliding_window (`bool`, *optional*, defaults to `False`):
+            Whether to use sliding window attention.
+        sliding_window (`int`, *optional*, defaults to 4096):
+            Sliding window attention (SWA) window size. If not specified, will default to `4096`.
+        max_window_layers (`int`, *optional*, defaults to 28):
+            The number of layers that use SWA (Sliding Window Attention). The bottom layers use SWA while the top use full attention.
+        attention_dropout (`float`, *optional*, defaults to 0.0):
+            The dropout ratio for the attention probabilities.
+        decoder_sparse_step (`int`, *optional*, defaults to 1):
+            The frequency of the MoE layer.
+        moe_intermediate_size (`int`, *optional*, defaults to 1408):
+            Intermediate size of the routed expert.
+        shared_expert_intermediate_size (`int`, *optional*, defaults to 5632):
+            Intermediate size of the shared expert.
+        num_experts_per_tok (`int`, *optional*, defaults to 4):
+            Number of selected experts.
+        num_experts (`int`, *optional*, defaults to 60):
+            Number of routed experts.
+        norm_topk_prob (`bool`, *optional*, defaults to `False`):
+            Whether to normalize the topk probabilities.
+        output_router_logits (`bool`, *optional*, defaults to `False`):
+            Whether or not the router logits should be returned by the model. Enabeling this will also
+            allow the model to output the auxiliary loss, including load balancing loss and router z-loss.
+        router_aux_loss_coef (`float`, *optional*, defaults to 0.001):
+            The aux loss factor for the total loss.
+
+    ```python
+    >>> from transformers import Qwen2MoeModel, Qwen2MoeConfig
+
+    >>> # Initializing a Qwen2MoE style configuration
+    >>> configuration = Qwen2MoeConfig()
+
+    >>> # Initializing a model from the Qwen1.5-MoE-A2.7B" style configuration
+    >>> model = Qwen2MoeModel(configuration)
+
+    >>> # Accessing the model configuration
+    >>> configuration = model.config
+    ```"""
+
+    model_type = "qwen2_moe"
+    keys_to_ignore_at_inference = ["past_key_values"]
+
+    def __init__(
+        self,
+        vocab_size=151936,
+        hidden_size=2048,
+        intermediate_size=5632,
+        num_hidden_layers=24,
+        num_attention_heads=16,
+        num_key_value_heads=16,
+        hidden_act="silu",
+        max_position_embeddings=32768,
+        initializer_range=0.02,
+        rms_norm_eps=1e-6,
+        use_cache=True,
+        tie_word_embeddings=False,
+        rope_theta=10000.0,
+        use_sliding_window=False,
+        sliding_window=4096,
+        max_window_layers=28,
+        attention_dropout=0.0,
+        decoder_sparse_step=1,
+        moe_intermediate_size=1408,
+        shared_expert_intermediate_size=5632,
+        num_experts_per_tok=4,
+        num_experts=60,
+        norm_topk_prob=False,
+        output_router_logits=False,
+        router_aux_loss_coef=0.001,
+        **kwargs,
+    ):
+        self.vocab_size = vocab_size
+        self.max_position_embeddings = max_position_embeddings
+        self.hidden_size = hidden_size
+        self.intermediate_size = intermediate_size
+        self.num_hidden_layers = num_hidden_layers
+        self.num_attention_heads = num_attention_heads
+        self.use_sliding_window = use_sliding_window
+        self.sliding_window = sliding_window
+        self.max_window_layers = max_window_layers
+
+        self.num_key_value_heads = num_key_value_heads
+        self.hidden_act = hidden_act
+        self.initializer_range = initializer_range
+        self.rms_norm_eps = rms_norm_eps
+        self.use_cache = use_cache
+        self.rope_theta = rope_theta
+        self.attention_dropout = attention_dropout
+
+        # MoE arguments
+        self.decoder_sparse_step = decoder_sparse_step
+        self.moe_intermediate_size = moe_intermediate_size
+        self.shared_expert_intermediate_size = shared_expert_intermediate_size
+        self.num_experts_per_tok = num_experts_per_tok
+        self.num_experts = num_experts
+        self.norm_topk_prob = norm_topk_prob
+        self.output_router_logits = output_router_logits
+        self.router_aux_loss_coef = router_aux_loss_coef
+
+        super().__init__(
+            tie_word_embeddings=tie_word_embeddings,
+            **kwargs,
+        )

src/transformers/utils/dummy_pt_objects.py

@@ -6939,6 +6939,34 @@ class Qwen2PreTrainedModel(metaclass=DummyObject):
         requires_backends(self, ["torch"])
 
 
+class Qwen2MoeForCausalLM(metaclass=DummyObject):
+    _backends = ["torch"]
+
+    def __init__(self, *args, **kwargs):
+        requires_backends(self, ["torch"])
+
+
+class Qwen2MoeForSequenceClassification(metaclass=DummyObject):
+    _backends = ["torch"]
+
+    def __init__(self, *args, **kwargs):
+        requires_backends(self, ["torch"])
+
+
+class Qwen2MoeModel(metaclass=DummyObject):
+    _backends = ["torch"]
+
+    def __init__(self, *args, **kwargs):
+        requires_backends(self, ["torch"])
+
+
+class Qwen2MoePreTrainedModel(metaclass=DummyObject):
+    _backends = ["torch"]
+
+    def __init__(self, *args, **kwargs):
+        requires_backends(self, ["torch"])
+
+
 class RagModel(metaclass=DummyObject):
     _backends = ["torch"]
 

tests/models/qwen2_moe/test_modeling_qwen2_moe.py

@@ -0,0 +1,672 @@
+# coding=utf-8
+# Copyright 2024 The Qwen team, Alibaba Group and The HuggingFace Inc. team. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+""" Testing suite for the PyTorch Qwen2MoE model. """
+
+
+import gc
+import tempfile
+import unittest
+
+import pytest
+
+from transformers import AutoTokenizer, Qwen2MoeConfig, is_torch_available, set_seed
+from transformers.testing_utils import (
+    backend_empty_cache,
+    require_bitsandbytes,
+    require_flash_attn,
+    require_torch,
+    require_torch_gpu,
+    require_torch_sdpa,
+    slow,
+    torch_device,
+)
+
+from ...generation.test_utils import GenerationTesterMixin
+from ...test_configuration_common import ConfigTester
+from ...test_modeling_common import ModelTesterMixin, ids_tensor
+from ...test_pipeline_mixin import PipelineTesterMixin
+
+
+if is_torch_available():
+    import torch
+
+    from transformers import (
+        Qwen2MoeForCausalLM,
+        Qwen2MoeForSequenceClassification,
+        Qwen2MoeModel,
+    )
+
+
+class Qwen2MoeModelTester:
+    def __init__(
+        self,
+        parent,
+        batch_size=13,
+        seq_length=7,
+        is_training=True,
+        use_input_mask=True,
+        use_token_type_ids=True,
+        use_labels=True,
+        vocab_size=99,
+        hidden_size=32,
+        num_hidden_layers=5,
+        max_window_layers=3,
+        use_sliding_window=True,
+        sliding_window=2,
+        num_attention_heads=4,
+        num_key_value_heads=2,
+        intermediate_size=37,
+        hidden_act="gelu",
+        hidden_dropout_prob=0.1,
+        attention_probs_dropout_prob=0.1,
+        max_position_embeddings=512,
+        expert_interval=1,
+        moe_intermediate_size=12,
+        shared_expert_intermediate_size=36,
+        shared_expert_gate=True,
+        num_experts_per_tok=2,
+        num_experts=8,
+        norm_topk_prob=False,
+        output_router_logits=False,
+        router_aux_loss_coef=0.001,
+        type_vocab_size=16,
+        type_sequence_label_size=2,
+        initializer_range=0.02,
+        num_labels=3,
+        num_choices=4,
+        pad_token_id=0,
+        bos_token_id=1,
+        scope=None,
+    ):
+        self.parent = parent
+        self.batch_size = batch_size
+        self.seq_length = seq_length
+        self.is_training = is_training
+        self.use_input_mask = use_input_mask
+        self.use_token_type_ids = use_token_type_ids
+        self.use_labels = use_labels
+        self.vocab_size = vocab_size
+        self.hidden_size = hidden_size
+        self.num_hidden_layers = num_hidden_layers
+        self.max_window_layers = max_window_layers
+        self.use_sliding_window = use_sliding_window
+        self.sliding_window = sliding_window
+        self.num_attention_heads = num_attention_heads
+        self.num_key_value_heads = num_key_value_heads
+        self.intermediate_size = intermediate_size
+        self.hidden_act = hidden_act
+        self.hidden_dropout_prob = hidden_dropout_prob
+        self.attention_probs_dropout_prob = attention_probs_dropout_prob
+        self.max_position_embeddings = max_position_embeddings
+        self.type_vocab_size = type_vocab_size
+        self.type_sequence_label_size = type_sequence_label_size
+        self.initializer_range = initializer_range
+        self.num_labels = num_labels
+        self.num_choices = num_choices
+        self.pad_token_id = pad_token_id
+        self.bos_token_id = bos_token_id
+        self.scope = scope
+        self.expert_interval = expert_interval
+        self.moe_intermediate_size = moe_intermediate_size
+        self.shared_expert_intermediate_size = shared_expert_intermediate_size
+        self.shared_expert_gate = shared_expert_gate
+        self.num_experts_per_tok = num_experts_per_tok
+        self.num_experts = num_experts
+        self.norm_topk_prob = norm_topk_prob
+        self.output_router_logits = output_router_logits
+        self.router_aux_loss_coef = router_aux_loss_coef
+
+    # Copied from tests.models.llama.test_modeling_llama.LlamaModelTester.prepare_config_and_inputs
+    def prepare_config_and_inputs(self):
+        input_ids = ids_tensor([self.batch_size, self.seq_length], self.vocab_size)
+
+        input_mask = None
+        if self.use_input_mask:
+            input_mask = torch.tril(torch.ones(self.batch_size, self.seq_length)).to(torch_device)
+
+        token_type_ids = None
+        if self.use_token_type_ids:
+            token_type_ids = ids_tensor([self.batch_size, self.seq_length], self.type_vocab_size)
+
+        sequence_labels = None
+        token_labels = None
+        choice_labels = None
+        if self.use_labels:
+            sequence_labels = ids_tensor([self.batch_size], self.type_sequence_label_size)
+            token_labels = ids_tensor([self.batch_size, self.seq_length], self.num_labels)
+            choice_labels = ids_tensor([self.batch_size], self.num_choices)
+
+        config = self.get_config()
+
+        return config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
+
+    def get_config(self):
+        return Qwen2MoeConfig(
+            vocab_size=self.vocab_size,
+            hidden_size=self.hidden_size,
+            num_hidden_layers=self.num_hidden_layers,
+            max_window_layers=self.max_window_layers,
+            use_sliding_window=self.use_sliding_window,
+            sliding_window=self.sliding_window,
+            num_attention_heads=self.num_attention_heads,
+            num_key_value_heads=self.num_key_value_heads,
+            intermediate_size=self.intermediate_size,
+            hidden_act=self.hidden_act,
+            hidden_dropout_prob=self.hidden_dropout_prob,
+            attention_probs_dropout_prob=self.attention_probs_dropout_prob,
+            max_position_embeddings=self.max_position_embeddings,
+            expert_interval=self.expert_interval,
+            moe_intermediate_size=self.moe_intermediate_size,
+            shared_expert_intermediate_size=self.shared_expert_intermediate_size,
+            shared_expert_gate=self.shared_expert_gate,
+            num_experts_per_tok=self.num_experts_per_tok,
+            num_experts=self.num_experts,
+            norm_topk_prob=self.norm_topk_prob,
+            output_router_logits=self.output_router_logits,
+            router_aux_loss_coef=self.router_aux_loss_coef,
+            type_vocab_size=self.type_vocab_size,
+            is_decoder=False,
+            initializer_range=self.initializer_range,
+            pad_token_id=self.pad_token_id,
+            bos_token_id=self.bos_token_id,
+        )
+
+    # Copied from tests.models.llama.test_modeling_llama.LlamaModelTester.create_and_check_model with Llama->Qwen2Moe
+    def create_and_check_model(
+        self, config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
+    ):
+        model = Qwen2MoeModel(config=config)
+        model.to(torch_device)
+        model.eval()
+        result = model(input_ids, attention_mask=input_mask)
+        result = model(input_ids)
+        self.parent.assertEqual(result.last_hidden_state.shape, (self.batch_size, self.seq_length, self.hidden_size))
+
+    # Copied from tests.models.llama.test_modeling_llama.LlamaModelTester.create_and_check_model_as_decoder with Llama->Qwen2Moe
+    def create_and_check_model_as_decoder(
+        self,
+        config,
+        input_ids,
+        token_type_ids,
+        input_mask,
+        sequence_labels,
+        token_labels,
+        choice_labels,
+        encoder_hidden_states,
+        encoder_attention_mask,
+    ):
+        config.add_cross_attention = True
+        model = Qwen2MoeModel(config)
+        model.to(torch_device)
+        model.eval()
+        result = model(
+            input_ids,
+            attention_mask=input_mask,
+            encoder_hidden_states=encoder_hidden_states,
+            encoder_attention_mask=encoder_attention_mask,
+        )
+        result = model(
+            input_ids,
+            attention_mask=input_mask,
+            encoder_hidden_states=encoder_hidden_states,
+        )
+        result = model(input_ids, attention_mask=input_mask)
+        self.parent.assertEqual(result.last_hidden_state.shape, (self.batch_size, self.seq_length, self.hidden_size))
+
+    # Copied from tests.models.llama.test_modeling_llama.LlamaModelTester.create_and_check_for_causal_lm with Llama->Qwen2Moe
+    def create_and_check_for_causal_lm(
+        self,
+        config,
+        input_ids,
+        token_type_ids,
+        input_mask,
+        sequence_labels,
+        token_labels,
+        choice_labels,
+        encoder_hidden_states,
+        encoder_attention_mask,
+    ):
+        model = Qwen2MoeForCausalLM(config=config)
+        model.to(torch_device)
+        model.eval()
+        result = model(input_ids, attention_mask=input_mask, labels=token_labels)
+        self.parent.assertEqual(result.logits.shape, (self.batch_size, self.seq_length, self.vocab_size))
+
+    # Copied from tests.models.llama.test_modeling_llama.LlamaModelTester.create_and_check_decoder_model_past_large_inputs with Llama->Qwen2Moe
+    def create_and_check_decoder_model_past_large_inputs(
+        self,
+        config,
+        input_ids,
+        token_type_ids,
+        input_mask,
+        sequence_labels,
+        token_labels,
+        choice_labels,
+        encoder_hidden_states,
+        encoder_attention_mask,
+    ):
+        config.is_decoder = True
+        config.add_cross_attention = True
+        model = Qwen2MoeForCausalLM(config=config)
+        model.to(torch_device)
+        model.eval()
+
+        # first forward pass
+        outputs = model(
+            input_ids,
+            attention_mask=input_mask,
+            encoder_hidden_states=encoder_hidden_states,
+            encoder_attention_mask=encoder_attention_mask,
+            use_cache=True,
+        )
+        past_key_values = outputs.past_key_values
+
+        # create hypothetical multiple next token and extent to next_input_ids
+        next_tokens = ids_tensor((self.batch_size, 3), config.vocab_size)
+        next_mask = ids_tensor((self.batch_size, 3), vocab_size=2)
+
+        # append to next input_ids and
+        next_input_ids = torch.cat([input_ids, next_tokens], dim=-1)
+        next_attention_mask = torch.cat([input_mask, next_mask], dim=-1)
+
+        output_from_no_past = model(
+            next_input_ids,
+            attention_mask=next_attention_mask,
+            encoder_hidden_states=encoder_hidden_states,
+            encoder_attention_mask=encoder_attention_mask,
+            output_hidden_states=True,
+        )["hidden_states"][0]
+        output_from_past = model(
+            next_tokens,
+            attention_mask=next_attention_mask,
+            encoder_hidden_states=encoder_hidden_states,
+            encoder_attention_mask=encoder_attention_mask,
+            past_key_values=past_key_values,
+            output_hidden_states=True,
+        )["hidden_states"][0]
+
+        # select random slice
+        random_slice_idx = ids_tensor((1,), output_from_past.shape[-1]).item()
+        output_from_no_past_slice = output_from_no_past[:, -3:, random_slice_idx].detach()
+        output_from_past_slice = output_from_past[:, :, random_slice_idx].detach()
+
+        self.parent.assertTrue(output_from_past_slice.shape[1] == next_tokens.shape[1])
+
+        # test that outputs are equal for slice
+        self.parent.assertTrue(torch.allclose(output_from_past_slice, output_from_no_past_slice, atol=1e-3))
+
+    # Copied from tests.models.llama.test_modeling_llama.LlamaModelTester.prepare_config_and_inputs_for_common
+    def prepare_config_and_inputs_for_common(self):
+        config_and_inputs = self.prepare_config_and_inputs()
+        (
+            config,
+            input_ids,
+            token_type_ids,
+            input_mask,
+            sequence_labels,
+            token_labels,
+            choice_labels,
+        ) = config_and_inputs
+        inputs_dict = {"input_ids": input_ids, "attention_mask": input_mask}
+        return config, inputs_dict
+
+
+@require_torch
+# Copied from tests.models.mistral.test_modeling_mistral.MistralModelTest with Mistral->Qwen2Moe
+class Qwen2MoeModelTest(ModelTesterMixin, GenerationTesterMixin, PipelineTesterMixin, unittest.TestCase):
+    all_model_classes = (
+        (Qwen2MoeModel, Qwen2MoeForCausalLM, Qwen2MoeForSequenceClassification) if is_torch_available() else ()
+    )
+    all_generative_model_classes = (Qwen2MoeForCausalLM,) if is_torch_available() else ()
+    pipeline_model_mapping = (
+        {
+            "feature-extraction": Qwen2MoeModel,
+            "text-classification": Qwen2MoeForSequenceClassification,
+            "text-generation": Qwen2MoeForCausalLM,
+            "zero-shot": Qwen2MoeForSequenceClassification,
+        }
+        if is_torch_available()
+        else {}
+    )
+    test_headmasking = False
+    test_pruning = False
+
+    # TODO (ydshieh): Check this. See https://app.circleci.com/pipelines/github/huggingface/transformers/79245/workflows/9490ef58-79c2-410d-8f51-e3495156cf9c/jobs/1012146
+    def is_pipeline_test_to_skip(
+        self, pipeline_test_casse_name, config_class, model_architecture, tokenizer_name, processor_name
+    ):
+        return True
+
+    def setUp(self):
+        self.model_tester = Qwen2MoeModelTester(self)
+        self.config_tester = ConfigTester(self, config_class=Qwen2MoeConfig, hidden_size=37)
+
+    def test_config(self):
+        self.config_tester.run_common_tests()
+
+    def test_model(self):
+        config_and_inputs = self.model_tester.prepare_config_and_inputs()
+        self.model_tester.create_and_check_model(*config_and_inputs)
+
+    def test_model_various_embeddings(self):
+        config_and_inputs = self.model_tester.prepare_config_and_inputs()
+        for type in ["absolute", "relative_key", "relative_key_query"]:
+            config_and_inputs[0].position_embedding_type = type
+            self.model_tester.create_and_check_model(*config_and_inputs)
+
+    def test_Qwen2Moe_sequence_classification_model(self):
+        config, input_dict = self.model_tester.prepare_config_and_inputs_for_common()
+        print(config)
+        config.num_labels = 3
+        input_ids = input_dict["input_ids"]
+        attention_mask = input_ids.ne(1).to(torch_device)
+        sequence_labels = ids_tensor([self.model_tester.batch_size], self.model_tester.type_sequence_label_size)
+        model = Qwen2MoeForSequenceClassification(config)
+        model.to(torch_device)
+        model.eval()
+        result = model(input_ids, attention_mask=attention_mask, labels=sequence_labels)
+        self.assertEqual(result.logits.shape, (self.model_tester.batch_size, self.model_tester.num_labels))
+
+    def test_Qwen2Moe_sequence_classification_model_for_single_label(self):
+        config, input_dict = self.model_tester.prepare_config_and_inputs_for_common()
+        config.num_labels = 3
+        config.problem_type = "single_label_classification"
+        input_ids = input_dict["input_ids"]
+        attention_mask = input_ids.ne(1).to(torch_device)
+        sequence_labels = ids_tensor([self.model_tester.batch_size], self.model_tester.type_sequence_label_size)
+        model = Qwen2MoeForSequenceClassification(config)
+        model.to(torch_device)
+        model.eval()
+        result = model(input_ids, attention_mask=attention_mask, labels=sequence_labels)
+        self.assertEqual(result.logits.shape, (self.model_tester.batch_size, self.model_tester.num_labels))
+
+    def test_Qwen2Moe_sequence_classification_model_for_multi_label(self):
+        config, input_dict = self.model_tester.prepare_config_and_inputs_for_common()
+        config.num_labels = 3
+        config.problem_type = "multi_label_classification"
+        input_ids = input_dict["input_ids"]
+        attention_mask = input_ids.ne(1).to(torch_device)
+        sequence_labels = ids_tensor(
+            [self.model_tester.batch_size, config.num_labels], self.model_tester.type_sequence_label_size
+        ).to(torch.float)
+        model = Qwen2MoeForSequenceClassification(config)
+        model.to(torch_device)
+        model.eval()
+        result = model(input_ids, attention_mask=attention_mask, labels=sequence_labels)
+        self.assertEqual(result.logits.shape, (self.model_tester.batch_size, self.model_tester.num_labels))
+
+    @unittest.skip("Qwen2Moe buffers include complex numbers, which breaks this test")
+    def test_save_load_fast_init_from_base(self):
+        pass
+
+    @unittest.skip("Qwen2Moe uses GQA on all models so the KV cache is a non standard format")
+    def test_past_key_values_format(self):
+        pass
+
+    @require_flash_attn
+    @require_torch_gpu
+    @pytest.mark.flash_attn_test
+    @slow
+    def test_flash_attn_2_generate_padding_right(self):
+        import torch
+
+        for model_class in self.all_generative_model_classes:
+            config, _ = self.model_tester.prepare_config_and_inputs_for_common()
+            model = model_class(config)
+
+            with tempfile.TemporaryDirectory() as tmpdirname:
+                model.save_pretrained(tmpdirname)
+                model = model_class.from_pretrained(tmpdirname, torch_dtype=torch.float16, low_cpu_mem_usage=True).to(
+                    torch_device
+                )
+
+                dummy_input = torch.LongTensor([[0, 2, 3, 4], [0, 2, 3, 4]]).to(torch_device)
+                dummy_attention_mask = torch.LongTensor([[1, 1, 1, 1], [1, 1, 1, 0]]).to(torch_device)
+
+                model.generate(dummy_input, attention_mask=dummy_attention_mask, max_new_tokens=1, do_sample=False)
+
+                model = model_class.from_pretrained(
+                    tmpdirname,
+                    torch_dtype=torch.float16,
+                    attn_implementation="flash_attention_2",
+                    low_cpu_mem_usage=True,
+                ).to(torch_device)
+
+                with self.assertRaises(ValueError):
+                    _ = model.generate(
+                        dummy_input, attention_mask=dummy_attention_mask, max_new_tokens=1, do_sample=False
+                    )
+
+    @require_flash_attn
+    @require_torch_gpu
+    @pytest.mark.flash_attn_test
+    @slow
+    def test_flash_attn_2_generate_use_cache(self):
+        import torch
+
+        max_new_tokens = 30
+
+        for model_class in self.all_generative_model_classes:
+            config, inputs_dict = self.model_tester.prepare_config_and_inputs_for_common()
+
+            dummy_input = inputs_dict[model_class.main_input_name]
+            if dummy_input.dtype in [torch.float32, torch.bfloat16]:
+                dummy_input = dummy_input.to(torch.float16)
+
+            # make sure that all models have enough positions for generation
+            if hasattr(config, "max_position_embeddings"):
+                config.max_position_embeddings = max_new_tokens + dummy_input.shape[1] + 1
+
+            model = model_class(config)
+
+            with tempfile.TemporaryDirectory() as tmpdirname:
+                model.save_pretrained(tmpdirname)
+
+                dummy_attention_mask = inputs_dict.get("attention_mask", torch.ones_like(dummy_input))
+                # NOTE: Qwen2Moe apparently does not support right padding + use_cache with FA2.
+                dummy_attention_mask[:, -1] = 1
+
+                model = model_class.from_pretrained(
+                    tmpdirname,
+                    torch_dtype=torch.float16,
+                    attn_implementation="flash_attention_2",
+                    low_cpu_mem_usage=True,
+                ).to(torch_device)
+
+                # Just test that a large cache works as expected
+                _ = model.generate(
+                    dummy_input,
+                    attention_mask=dummy_attention_mask,
+                    max_new_tokens=max_new_tokens,
+                    do_sample=False,
+                    use_cache=True,
+                )
+
+    @require_flash_attn
+    @require_torch_gpu
+    @pytest.mark.flash_attn_test
+    @slow
+    def test_flash_attn_2_inference_padding_right(self):
+        self.skipTest("Qwen2Moe flash attention does not support right padding")
+
+    # Ignore copy
+    def test_load_balancing_loss(self):
+        r"""
+        Let's make sure we can actually compute the loss and do a backward on it.
+        """
+        config, input_dict = self.model_tester.prepare_config_and_inputs_for_common()
+        config.num_labels = 3
+        config.num_experts = 8
+        config.expert_interval = 2
+        config.output_router_logits = True
+        input_ids = input_dict["input_ids"]
+        attention_mask = input_ids.ne(1).to(torch_device)
+        model = Qwen2MoeForCausalLM(config)
+        model.to(torch_device)
+        model.eval()
+        result = model(input_ids, attention_mask=attention_mask)
+        self.assertEqual(result.router_logits[0].shape, (91, config.num_experts))
+        torch.testing.assert_close(result.aux_loss.cpu(), torch.tensor(2, dtype=torch.float32), rtol=1e-2, atol=1e-2)
+
+        # First, we make sure that adding padding tokens doesn't change the loss
+        # loss(input_ids, attention_mask=None) == loss(input_ids + padding, attention_mask=attention_mask_with_padding)
+        pad_length = 1000
+        # Add padding tokens (assume that pad_token_id=1) to input_ids
+        padding_block = torch.ones(input_ids.shape[0], pad_length, dtype=torch.int32).to(torch_device)
+        padded_input_ids = torch.cat((padding_block, input_ids), dim=1)  # this is to simulate padding to the left
+        padded_attention_mask = padded_input_ids.ne(1).to(torch_device)
+
+        padded_result = model(padded_input_ids, attention_mask=padded_attention_mask)
+        torch.testing.assert_close(result.aux_loss.cpu(), padded_result.aux_loss.cpu(), rtol=1e-4, atol=1e-4)
+
+        # We make sure that the loss of includding padding tokens != the loss without padding tokens
+        # if attention_mask=None --> we don't exclude padding tokens
+        include_padding_result = model(padded_input_ids, attention_mask=None)
+
+        # This is to mimic torch.testing.assert_not_close
+        self.assertNotAlmostEqual(include_padding_result.aux_loss.item(), result.aux_loss.item())
+
+
+@require_torch
+class Qwen2MoeIntegrationTest(unittest.TestCase):
+    @slow
+    def test_model_a2_7b_logits(self):
+        input_ids = [1, 306, 4658, 278, 6593, 310, 2834, 338]
+        model = Qwen2MoeForCausalLM.from_pretrained("Qwen/Qwen1.5-MoE-A2.7B", device_map="auto")
+        input_ids = torch.tensor([input_ids]).to(model.model.embed_tokens.weight.device)
+        with torch.no_grad():
+            out = model(input_ids).logits.cpu()
+        # Expected mean on dim = -1
+        EXPECTED_MEAN = torch.tensor([[-4.2125, -3.6416, -4.9136, -4.3005, -4.9938, -3.4393, -3.5195, -4.1621]])
+        torch.testing.assert_close(out.mean(-1), EXPECTED_MEAN, atol=1e-2, rtol=1e-2)
+        # slicing logits[0, 0, 0:30]
+        EXPECTED_SLICE = torch.tensor([2.3013, -0.6595, -0.1389, -1.4095, -1.7381, -1.7609, -2.0449, -2.4289, -3.0271, -2.1351, -0.6568, -4.6012, -1.9102, -0.7475, -3.1377, 4.6904, 7.1936, 7.0991, 6.4414, 6.1720, 6.2617, 5.8751, 5.6997, 5.6011, 5.5828, -3.9505, -0.5384, -0.3392, 1.2445, 2.0714])  # fmt: skip
+        print(out[0, 0, :30])
+        torch.testing.assert_close(out[0, 0, :30], EXPECTED_SLICE, atol=1e-4, rtol=1e-4)
+
+        del model
+        backend_empty_cache(torch_device)
+        gc.collect()
+
+    @slow
+    def test_model_a2_7b_generation(self):
+        EXPECTED_TEXT_COMPLETION = """To be or not to be, that is the question. This is the question that has been asked by many people over the"""
+        prompt = "To be or not to"
+        tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen1.5-MoE-A2.7B", use_fast=False)
+        model = Qwen2MoeForCausalLM.from_pretrained("Qwen/Qwen1.5-MoE-A2.7B", device_map="auto")
+        input_ids = tokenizer.encode(prompt, return_tensors="pt").to(model.model.embed_tokens.weight.device)
+
+        # greedy generation outputs
+        generated_ids = model.generate(input_ids, max_new_tokens=20, temperature=0)
+        text = tokenizer.decode(generated_ids[0], skip_special_tokens=True)
+        self.assertEqual(EXPECTED_TEXT_COMPLETION, text)
+
+        del model
+        backend_empty_cache(torch_device)
+        gc.collect()
+
+    @require_bitsandbytes
+    @slow
+    @require_flash_attn
+    def test_model_a2_7b_long_prompt(self):
+        EXPECTED_OUTPUT_TOKEN_IDS = [306, 338]
+        # An input with 4097 tokens that is above the size of the sliding window
+        input_ids = [1] + [306, 338] * 2048
+        model = Qwen2MoeForCausalLM.from_pretrained(
+            "Qwen/Qwen1.5-MoE-A2.7B",
+            device_map="auto",
+            load_in_4bit=True,
+            attn_implementation="flash_attention_2",
+        )
+        input_ids = torch.tensor([input_ids]).to(model.model.embed_tokens.weight.device)
+        generated_ids = model.generate(input_ids, max_new_tokens=4, temperature=0)
+        self.assertEqual(EXPECTED_OUTPUT_TOKEN_IDS, generated_ids[0][-2:].tolist())
+
+        # Assisted generation
+        assistant_model = model
+        assistant_model.generation_config.num_assistant_tokens = 2
+        assistant_model.generation_config.num_assistant_tokens_schedule = "constant"
+        generated_ids = model.generate(input_ids, max_new_tokens=4, temperature=0)
+        self.assertEqual(EXPECTED_OUTPUT_TOKEN_IDS, generated_ids[0][-2:].tolist())
+
+        del assistant_model
+        del model
+        backend_empty_cache(torch_device)
+        gc.collect()
+
+    @slow
+    @require_torch_sdpa
+    def test_model_a2_7b_long_prompt_sdpa(self):
+        EXPECTED_OUTPUT_TOKEN_IDS = [306, 338]
+        # An input with 4097 tokens that is above the size of the sliding window
+        input_ids = [1] + [306, 338] * 2048
+        model = Qwen2MoeForCausalLM.from_pretrained(
+            "Qwen/Qwen1.5-MoE-A2.7B",
+            device_map="auto",
+            attn_implementation="sdpa",
+        )
+        input_ids = torch.tensor([input_ids]).to(model.model.embed_tokens.weight.device)
+        generated_ids = model.generate(input_ids, max_new_tokens=4, temperature=0)
+        self.assertEqual(EXPECTED_OUTPUT_TOKEN_IDS, generated_ids[0][-2:].tolist())
+
+        # Assisted generation
+        assistant_model = model
+        assistant_model.generation_config.num_assistant_tokens = 2
+        assistant_model.generation_config.num_assistant_tokens_schedule = "constant"
+        generated_ids = assistant_model.generate(input_ids, max_new_tokens=4, temperature=0)
+        self.assertEqual(EXPECTED_OUTPUT_TOKEN_IDS, generated_ids[0][-2:].tolist())
+
+        del assistant_model
+
+        backend_empty_cache(torch_device)
+        gc.collect()
+
+        EXPECTED_TEXT_COMPLETION = """To be or not to be, that is the question. This is the question that has been asked by many people over the"""
+        prompt = "To be or not to"
+        tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen1.5-MoE-A2.7B", use_fast=False)
+
+        input_ids = tokenizer.encode(prompt, return_tensors="pt").to(model.model.embed_tokens.weight.device)
+
+        # greedy generation outputs
+        generated_ids = model.generate(input_ids, max_new_tokens=20, temperature=0)
+        text = tokenizer.decode(generated_ids[0], skip_special_tokens=True)
+        self.assertEqual(EXPECTED_TEXT_COMPLETION, text)
+
+    @slow
+    def test_speculative_generation(self):
+        EXPECTED_TEXT_COMPLETION = (
+            "To be or not to be, that is the question.\nThe answer is to be, of course. But what does it"
+        )
+        prompt = "To be or not to"
+        tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen1.5-MoE-A2.7B", use_fast=False)
+        model = Qwen2MoeForCausalLM.from_pretrained(
+            "Qwen/Qwen1.5-MoE-A2.7B", device_map="auto", torch_dtype=torch.float16
+        )
+        assistant_model = Qwen2MoeForCausalLM.from_pretrained(
+            "Qwen/Qwen1.5-MoE-A2.7B", device_map="auto", torch_dtype=torch.float16
+        )
+        input_ids = tokenizer.encode(prompt, return_tensors="pt").to(model.model.embed_tokens.weight.device)
+
+        # greedy generation outputs
+        set_seed(0)
+        generated_ids = model.generate(
+            input_ids, max_new_tokens=20, do_sample=True, temperature=0.3, assistant_model=assistant_model
+        )
+        text = tokenizer.decode(generated_ids[0], skip_special_tokens=True)
+        self.assertEqual(EXPECTED_TEXT_COMPLETION, text)
+
+        del model
+        backend_empty_cache(torch_device)
+        gc.collect()

utils/not_doctested.txt

@@ -201,6 +201,7 @@ docs/source/en/model_doc/prophetnet.md
 docs/source/en/model_doc/pvt.md
 docs/source/en/model_doc/qdqbert.md
 docs/source/en/model_doc/qwen2.md
+docs/source/en/model_doc/qwen2_moe.md
 docs/source/en/model_doc/rag.md
 docs/source/en/model_doc/realm.md
 docs/source/en/model_doc/reformer.md
@@ -759,6 +760,8 @@ src/transformers/models/qwen2/configuration_qwen2.py
 src/transformers/models/qwen2/modeling_qwen2.py
 src/transformers/models/qwen2/tokenization_qwen2.py
 src/transformers/models/qwen2/tokenization_qwen2_fast.py
+src/transformers/models/qwen2_moe/configuration_qwen2_moe.py
+src/transformers/models/qwen2_moe/modeling_qwen2_moe.py
 src/transformers/models/rag/configuration_rag.py
 src/transformers/models/rag/modeling_rag.py
 src/transformers/models/rag/modeling_tf_rag.py