Add UDOP (#22940)

* First draft * More improvements * More improvements * More fixes * Fix copies * More improvements * More fixes * More improvements * Convert checkpoint * More improvements, set up tests * Fix more tests * Add UdopModel * More improvements * Fix equivalence test * More fixes * Redesign model * Extend conversion script * Use real inputs for conversion script * Add image processor * Improve conversion script * Add UdopTokenizer * Add fast tokenizer * Add converter * Update README's * Add processor * Add fully fledged tokenizer * Add fast tokenizer * Use processor in conversion script * Add tokenizer tests * Fix one more test * Fix more tests * Fix tokenizer tests * Enable fast tokenizer tests * Fix more tests * Fix additional_special_tokens of fast tokenizer * Fix tokenizer tests * Fix more tests * Fix equivalence test * Rename image to pixel_values * Rename seg_data to bbox * More renamings * Remove vis_special_token * More improvements * Add docs * Fix copied from * Update slow tokenizer * Update fast tokenizer design * Make text input optional * Add first draft of processor tests * Fix more processor tests * Fix decoder_start_token_id * Fix test_initialization * Add integration test * More improvements * Improve processor, add test * Add more copied from * Add more copied from * Add more copied from * Add more copied from * Remove print statement * Update README and auto mapping * Delete files * Delete another file * Remove code * Fix test * Fix docs * Remove asserts * Add doc tests * Include UDOP in exotic model tests * Add expected tesseract decodings * Add sentencepiece * Use same design as T5 * Add UdopEncoderModel * Add UdopEncoderModel to tests * More fixes * Fix fast tokenizer * Fix one more test * Remove parallelisable attribute * Fix copies * Remove legacy file * Copy from T5Tokenizer * Fix rebase * More fixes, copy from T5 * More fixes * Fix init * Use ArthurZ/udop for tests * Make all model tests pass * Remove UdopForConditionalGeneration from auto mapping * Fix more tests * fixups * more fixups * fix the tokenizers * remove un-necessary changes * nits * nits * replace truncate_sequences_boxes with truncate_sequences for fix-copies * nit current path * add a test for input ids * ids that we should get taken from c9f7a32f57440d90ff79890270d376a1cc0acb68 * nits converting * nits * apply ruff * nits * nits * style * fix slow order of addition * fix udop fast range as well * fixup * nits * Add docstrings * Fix gradient checkpointing * Update code examples * Skip tests * Update integration test * Address comment * Make fixup * Remove extra ids from tokenizer * Skip test * Apply suggestions from code review Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com> * Update year * Address comment * Address more comments * Address comments * Add copied from * Update CI * Rename script * Update model id * Add AddedToken, skip tests * Update CI * Fix doc tests * Do not use Tesseract for the doc tests * Remove kwargs * Add original inputs * Update casting * Fix doc test * Update question * Update question * Use LayoutLMv3ImageProcessor * Update organization * Improve docs * Update forward signature * Make images optional * Remove deprecated device argument * Add comment, add add_prefix_space * More improvements * Remove kwargs --------- Co-authored-by: ArthurZucker <arthur.zucker@gmail.com> Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

Add UDOP (#22940)
* First draft * More improvements * More improvements * More fixes * Fix copies * More improvements * More fixes * More improvements * Convert checkpoint * More improvements, set up tests * Fix more tests * Add UdopModel * More improvements * Fix equivalence test * More fixes * Redesign model * Extend conversion script * Use real inputs for conversion script * Add image processor * Improve conversion script * Add UdopTokenizer * Add fast tokenizer * Add converter * Update README's * Add processor * Add fully fledged tokenizer * Add fast tokenizer * Use processor in conversion script * Add tokenizer tests * Fix one more test * Fix more tests * Fix tokenizer tests * Enable fast tokenizer tests * Fix more tests * Fix additional_special_tokens of fast tokenizer * Fix tokenizer tests * Fix more tests * Fix equivalence test * Rename image to pixel_values * Rename seg_data to bbox * More renamings * Remove vis_special_token * More improvements * Add docs * Fix copied from * Update slow tokenizer * Update fast tokenizer design * Make text input optional * Add first draft of processor tests * Fix more processor tests * Fix decoder_start_token_id * Fix test_initialization * Add integration test * More improvements * Improve processor, add test * Add more copied from * Add more copied from * Add more copied from * Add more copied from * Remove print statement * Update README and auto mapping * Delete files * Delete another file * Remove code * Fix test * Fix docs * Remove asserts * Add doc tests * Include UDOP in exotic model tests * Add expected tesseract decodings * Add sentencepiece * Use same design as T5 * Add UdopEncoderModel * Add UdopEncoderModel to tests * More fixes * Fix fast tokenizer * Fix one more test * Remove parallelisable attribute * Fix copies * Remove legacy file * Copy from T5Tokenizer * Fix rebase * More fixes, copy from T5 * More fixes * Fix init * Use ArthurZ/udop for tests * Make all model tests pass * Remove UdopForConditionalGeneration from auto mapping * Fix more tests * fixups * more fixups * fix the tokenizers * remove un-necessary changes * nits * nits * replace truncate_sequences_boxes with truncate_sequences for fix-copies * nit current path * add a test for input ids * ids that we should get taken from c9f7a32f57440d90ff79890270d376a1cc0acb68 * nits converting * nits * apply ruff * nits * nits * style * fix slow order of addition * fix udop fast range as well * fixup * nits * Add docstrings * Fix gradient checkpointing * Update code examples * Skip tests * Update integration test * Address comment * Make fixup * Remove extra ids from tokenizer * Skip test * Apply suggestions from code review Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com> * Update year * Address comment * Address more comments * Address comments * Add copied from * Update CI * Rename script * Update model id * Add AddedToken, skip tests * Update CI * Fix doc tests * Do not use Tesseract for the doc tests * Remove kwargs * Add original inputs * Update casting * Fix doc test * Update question * Update question * Use LayoutLMv3ImageProcessor * Update organization * Improve docs * Update forward signature * Make images optional * Remove deprecated device argument * Add comment, add add_prefix_space * More improvements * Remove kwargs --------- Co-authored-by: ArthurZucker <arthur.zucker@gmail.com> Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
836921fd · NielsRogge · GitHub · ed74d978 · 836921fd · 836921fd
Unverified Commit 836921fd authored Mar 04, 2024 by NielsRogge Committed by GitHub Mar 04, 2024
20 changed files
--- a/.circleci/create_circleci_config.py
+++ b/.circleci/create_circleci_config.py
@@ -475,6 +475,7 @@ exotic_models_job = CircleCIJob(
        "pip install -U --upgrade-strategy eager 'git+https://github.com/facebookresearch/detectron2.git'",
        "sudo apt install tesseract-ocr",
        "pip install -U --upgrade-strategy eager pytesseract",
+        "pip install --upgrade-strategy eager sentencepiece",
        "pip install -U --upgrade-strategy eager natten==0.15.1+torch210cpu -f https://shi-labs.com/natten/wheels",
        "pip install -U --upgrade-strategy eager python-Levenshtein",
        "pip install -U --upgrade-strategy eager opencv-python",
@@ -485,6 +486,7 @@ exotic_models_job = CircleCIJob(
        "tests/models/*layoutlmv*",
        "tests/models/*nat",
        "tests/models/deta",
+        "tests/models/udop",
        "tests/models/nougat",
    ],
    pytest_num_workers=1,

--- a/README.md
+++ b/README.md
@@ -511,6 +511,7 @@ Current number of checkpoints: ![](https://img.shields.io/endpoint?url=https://h
 1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (from Microsoft), released together with the paper [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) by Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei.
 1. **[TVLT](https://huggingface.co/docs/transformers/model_doc/tvlt)** (from UNC Chapel Hill) released with the paper [TVLT: Textless Vision-Language Transformer](https://arxiv.org/abs/2209.14156) by Zineng Tang, Jaemin Cho, Yixin Nie, Mohit Bansal.
 1. **[TVP](https://huggingface.co/docs/transformers/model_doc/tvp)** (from Intel) released with the paper [Text-Visual Prompting for Efficient 2D Temporal Video Grounding](https://arxiv.org/abs/2303.04995) by Yimeng Zhang, Xin Chen, Jinghan Jia, Sijia Liu, Ke Ding.
+1. **[UDOP](https://huggingface.co/docs/transformers/main/model_doc/udop)** (from Microsoft Research) released with the paper [Unifying Vision, Text, and Layout for Universal Document Processing](https://arxiv.org/abs/2212.02623) by Zineng Tang, Ziyi Yang, Guoxin Wang, Yuwei Fang, Yang Liu, Chenguang Zhu, Michael Zeng, Cha Zhang, Mohit Bansal.
 1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (from Google Research) released with the paper [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) by Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
 1. **[UMT5](https://huggingface.co/docs/transformers/model_doc/umt5)** (from Google Research) released with the paper [UniMax: Fairer and More Effective Language Sampling for Large-Scale Multilingual Pretraining](https://openreview.net/forum?id=kXwdL1cWOAi) by Hyung Won Chung, Xavier Garcia, Adam Roberts, Yi Tay, Orhan Firat, Sharan Narang, Noah Constant.
 1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (from Microsoft Research) released with the paper [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597) by Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang.

--- a/README_es.md
+++ b/README_es.md
@@ -484,6 +484,7 @@ Número actual de puntos de control: ![](https://img.shields.io/endpoint?url=htt
 1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (from Microsoft), released together with the paper [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) by Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei.
 1. **[TVLT](https://huggingface.co/docs/transformers/model_doc/tvlt)** (from UNC Chapel Hill) released with the paper [TVLT: Textless Vision-Language Transformer](https://arxiv.org/abs/2209.14156) by Zineng Tang, Jaemin Cho, Yixin Nie, Mohit Bansal.
 1. **[TVP](https://huggingface.co/docs/transformers/model_doc/tvp)** (from Intel) released with the paper [Text-Visual Prompting for Efficient 2D Temporal Video Grounding](https://arxiv.org/abs/2303.04995) by Yimeng Zhang, Xin Chen, Jinghan Jia, Sijia Liu, Ke Ding.
+1. **[UDOP](https://huggingface.co/docs/transformers/main/model_doc/udop)** (from Microsoft Research) released with the paper [Unifying Vision, Text, and Layout for Universal Document Processing](https://arxiv.org/abs/2212.02623) by Zineng Tang, Ziyi Yang, Guoxin Wang, Yuwei Fang, Yang Liu, Chenguang Zhu, Michael Zeng, Cha Zhang, Mohit Bansal.
 1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (from Google Research) released with the paper [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) by Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
 1. **[UMT5](https://huggingface.co/docs/transformers/model_doc/umt5)** (from Google Research) released with the paper [UniMax: Fairer and More Effective Language Sampling for Large-Scale Multilingual Pretraining](https://openreview.net/forum?id=kXwdL1cWOAi) by Hyung Won Chung, Xavier Garcia, Adam Roberts, Yi Tay, Orhan Firat, Sharan Narang, Noah Constant.
 1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (from Microsoft Research) released with the paper [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597) by Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang.

--- a/README_fr.md
+++ b/README_fr.md
@@ -505,6 +505,7 @@ Nombre actuel de points de contrôle : ![](https://img.shields.io/endpoint?url=h
 1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (de Microsoft), publié dans l'article [TrOCR : Reconnaissance optique de caractères basée sur un transformateur avec des modèles pré-entraînés](https://arxiv.org/abs/2109.10282) par Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei.
 1. **[TVLT](https://huggingface.co/docs/transformers/model_doc/tvlt)** (de l'UNC Chapel Hill) a été publié dans l'article [TVLT : Transformer Vision-Language sans texte](https://arxiv.org/abs/2209.14156) par Zineng Tang, Jaemin Cho, Yixin Nie, Mohit Bansal.
 1. **[TVP](https://huggingface.co/docs/transformers/model_doc/tvp)** (d'Intel) a été publié dans l'article [Text-Visual Prompting for Efficient 2D Temporal Video Grounding](https://arxiv.org/abs/2303.04995) par Yimeng Zhang, Xin Chen, Jinghan Jia, Sijia Liu, Ke Ding.
+1. **[UDOP](https://huggingface.co/docs/transformers/main/model_doc/udop)** (de Microsoft Research) publié dans l'article [Unifying Vision, Text, and Layout for Universal Document Processing](https://arxiv.org/abs/2212.02623) parZineng Tang, Ziyi Yang, Guoxin Wang, Yuwei Fang, Yang Liu, Chenguang Zhu, Michael Zeng, Cha Zhang, Mohit Bansal.
 1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (de Google Research) a été publié dans l'article [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) par Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler.
 1. **[UMT5](https://huggingface.co/docs/transformers/model_doc/umt5)** (de Google Research) a été publié dans l'article [UniMax : Échantillonnage linguistique plus équitable et plus efficace pour l'entraînement préalable multilingue à grande échelle](https://openreview.net/forum?id=kXwdL1cWOAi) par Hyung Won Chung, Xavier Garcia, Adam Roberts, Yi Tay, Orhan Firat, Sharan Narang, Noah Constant.
 1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (de Microsoft Research) a été publié dans l'article [UniSpeech : Apprentissage unifié de la représentation de la parole avec des données étiquetées et non étiquetées](https://arxiv.org/abs/2101.07597) par Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang.

--- a/README_hd.md
+++ b/README_hd.md
@@ -458,6 +458,7 @@ conda install conda-forge::transformers
 1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (from Microsoft) released with the paper [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) by Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei.
 1. **[TVLT](https://huggingface.co/docs/transformers/model_doc/tvlt)** (from UNC Chapel Hill) released with the paper [TVLT: Textless Vision-Language Transformer](https://arxiv.org/abs/2209.14156) by Zineng Tang, Jaemin Cho, Yixin Nie, Mohit Bansal.
 1. **[TVP](https://huggingface.co/docs/transformers/model_doc/tvp)** (from Intel) released with the paper [Text-Visual Prompting for Efficient 2D Temporal Video Grounding](https://arxiv.org/abs/2303.04995) by Yimeng Zhang, Xin Chen, Jinghan Jia, Sijia Liu, Ke Ding.
+1. **[UDOP](https://huggingface.co/docs/transformers/main/model_doc/udop)** (Microsoft Research से) Zineng Tang, Ziyi Yang, Guoxin Wang, Yuwei Fang, Yang Liu, Chenguang Zhu, Michael Zeng, Cha Zhang, Mohit Bansal. द्वाराअनुसंधान पत्र [Unifying Vision, Text, and Layout for Universal Document Processing](https://arxiv.org/abs/2212.02623) के साथ जारी किया गया
 1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (from Google Research) released with the paper [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) by Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
 1. **[UMT5](https://huggingface.co/docs/transformers/model_doc/umt5)** (Google Research से) Hyung Won Chung, Xavier Garcia, Adam Roberts, Yi Tay, Orhan Firat, Sharan Narang, Noah Constant. द्वाराअनुसंधान पत्र [UniMax: Fairer and More Effective Language Sampling for Large-Scale Multilingual Pretraining](https://openreview.net/forum?id=kXwdL1cWOAi) के साथ जारी किया गया
 1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (माइक्रोसॉफ्ट रिसर्च से) साथ में दिया गया पेपर [UniSpeech: यूनिफाइड स्पीच रिप्रेजेंटेशन लर्निंग विद लेबलेड एंड अनलेबल्ड डेटा](https://arxiv.org/abs/2101.07597) चेंगई वांग, यू वू, याओ कियान, केनिची कुमातानी, शुजी लियू, फुरु वेई, माइकल ज़ेंग, ज़ुएदोंग हुआंग द्वारा।

--- a/README_ja.md
+++ b/README_ja.md
@@ -518,6 +518,7 @@ Flax、PyTorch、TensorFlowをcondaでインストールする方法は、それ
 1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (Microsoft から), Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei から公開された研究論文: [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282)
 1. **[TVLT](https://huggingface.co/docs/transformers/model_doc/tvlt)** (from UNC Chapel Hill から), Zineng Tang, Jaemin Cho, Yixin Nie, Mohit Bansal から公開された研究論文: [TVLT: Textless Vision-Language Transformer](https://arxiv.org/abs/2209.14156)
 1. **[TVP](https://huggingface.co/docs/transformers/model_doc/tvp)** (Intel から), Yimeng Zhang, Xin Chen, Jinghan Jia, Sijia Liu, Ke Ding から公開された研究論文: [Text-Visual Prompting for Efficient 2D Temporal Video Grounding](https://arxiv.org/abs/2303.04995)
+1. **[UDOP](https://huggingface.co/docs/transformers/main/model_doc/udop)** (Microsoft Research から) Zineng Tang, Ziyi Yang, Guoxin Wang, Yuwei Fang, Yang Liu, Chenguang Zhu, Michael Zeng, Cha Zhang, Mohit Bansal. から公開された研究論文 [Unifying Vision, Text, and Layout for Universal Document Processing](https://arxiv.org/abs/2212.02623)
 1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (Google Research から) Yi Tay, Mostafa Dehghani, Vinh Q から公開された研究論文: [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
 1. **[UMT5](https://huggingface.co/docs/transformers/model_doc/umt5)** (Google Research から) Hyung Won Chung, Xavier Garcia, Adam Roberts, Yi Tay, Orhan Firat, Sharan Narang, Noah Constant. から公開された研究論文 [UniMax: Fairer and More Effective Language Sampling for Large-Scale Multilingual Pretraining](https://openreview.net/forum?id=kXwdL1cWOAi)
 1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (Microsoft Research から) Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang から公開された研究論文: [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597)

--- a/README_ko.md
+++ b/README_ko.md
@@ -433,6 +433,7 @@ Flax, PyTorch, TensorFlow 설치 페이지에서 이들을 conda로 설치하는
 1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (Microsoft 에서) Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei 의 [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) 논문과 함께 발표했습니다.
 1. **[TVLT](https://huggingface.co/docs/transformers/model_doc/tvlt)** (from UNC Chapel Hill 에서) Zineng Tang, Jaemin Cho, Yixin Nie, Mohit Bansal 의 [TVLT: Textless Vision-Language Transformer](https://arxiv.org/abs/2209.14156) 논문과 함께 발표했습니다.
 1. **[TVP](https://huggingface.co/docs/transformers/model_doc/tvp)** (Intel 에서) Yimeng Zhang, Xin Chen, Jinghan Jia, Sijia Liu, Ke Ding 의 [Text-Visual Prompting for Efficient 2D Temporal Video Grounding](https://arxiv.org/abs/2303.04995) 논문과 함께 발표했습니다.
+1. **[UDOP](https://huggingface.co/docs/transformers/main/model_doc/udop)** (Microsoft Research 에서 제공)은 Zineng Tang, Ziyi Yang, Guoxin Wang, Yuwei Fang, Yang Liu, Chenguang Zhu, Michael Zeng, Cha Zhang, Mohit Bansal.의 [Unifying Vision, Text, and Layout for Universal Document Processing](https://arxiv.org/abs/2212.02623)논문과 함께 발표했습니다.
 1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (Google Research 에서) Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzle 의 [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) 논문과 함께 발표했습니다.
 1. **[UMT5](https://huggingface.co/docs/transformers/model_doc/umt5)** (Google Research 에서 제공)은 Hyung Won Chung, Xavier Garcia, Adam Roberts, Yi Tay, Orhan Firat, Sharan Narang, Noah Constant.의 [UniMax: Fairer and More Effective Language Sampling for Large-Scale Multilingual Pretraining](https://openreview.net/forum?id=kXwdL1cWOAi)논문과 함께 발표했습니다.
 1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (Microsoft Research 에서) Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang 의 [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597) 논문과 함께 발표했습니다.

--- a/README_zh-hans.md
+++ b/README_zh-hans.md
@@ -457,6 +457,7 @@ conda install conda-forge::transformers
 1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (来自 Microsoft) 伴随论文 [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) 由 Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei 发布。
 1. **[TVLT](https://huggingface.co/docs/transformers/model_doc/tvlt)** (来自 UNC Chapel Hill) 伴随论文 [TVLT: Textless Vision-Language Transformer](https://arxiv.org/abs/2209.14156) 由 Zineng Tang, Jaemin Cho, Yixin Nie, Mohit Bansal 发布。
 1. **[TVP](https://huggingface.co/docs/transformers/model_doc/tvp)** (来自 Intel) 伴随论文 [Text-Visual Prompting for Efficient 2D Temporal Video Grounding](https://arxiv.org/abs/2303.04995) 由 Yimeng Zhang, Xin Chen, Jinghan Jia, Sijia Liu, Ke Ding 发布.
+1. **[UDOP](https://huggingface.co/docs/transformers/main/model_doc/udop)** (来自 Microsoft Research) 伴随论文 [Unifying Vision, Text, and Layout for Universal Document Processing](https://arxiv.org/abs/2212.02623) 由 Zineng Tang, Ziyi Yang, Guoxin Wang, Yuwei Fang, Yang Liu, Chenguang Zhu, Michael Zeng, Cha Zhang, Mohit Bansal 发布。
 1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (from Google Research) released with the paper [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) by Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
 1. **[UMT5](https://huggingface.co/docs/transformers/model_doc/umt5)** (来自 Google Research) 伴随论文 [UniMax: Fairer and More Effective Language Sampling for Large-Scale Multilingual Pretraining](https://openreview.net/forum?id=kXwdL1cWOAi) 由 Hyung Won Chung, Xavier Garcia, Adam Roberts, Yi Tay, Orhan Firat, Sharan Narang, Noah Constant 发布。
 1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (来自 Microsoft Research) 伴随论文 [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597) 由 Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang 发布。

--- a/README_zh-hant.md
+++ b/README_zh-hant.md
@@ -469,6 +469,7 @@ conda install conda-forge::transformers
 1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (from Microsoft) released with the paper [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) by Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei.
 1. **[TVLT](https://huggingface.co/docs/transformers/model_doc/tvlt)** (from UNC Chapel Hill) released with the paper [TVLT: Textless Vision-Language Transformer](https://arxiv.org/abs/2209.14156) by Zineng Tang, Jaemin Cho, Yixin Nie, Mohit Bansal.
 1. **[TVP](https://huggingface.co/docs/transformers/model_doc/tvp)** (from Intel) released with the paper [Text-Visual Prompting for Efficient 2D Temporal Video Grounding](https://arxiv.org/abs/2303.04995) by Yimeng Zhang, Xin Chen, Jinghan Jia, Sijia Liu, Ke Ding.
+1. **[UDOP](https://huggingface.co/docs/transformers/main/model_doc/udop)** (from Microsoft Research) released with the paper [Unifying Vision, Text, and Layout for Universal Document Processing](https://arxiv.org/abs/2212.02623) by Zineng Tang, Ziyi Yang, Guoxin Wang, Yuwei Fang, Yang Liu, Chenguang Zhu, Michael Zeng, Cha Zhang, Mohit Bansal.
 1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (from Google Research) released with the paper [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) by Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
 1. **[UMT5](https://huggingface.co/docs/transformers/model_doc/umt5)** (from Google Research) released with the paper [UniMax: Fairer and More Effective Language Sampling for Large-Scale Multilingual Pretraining](https://openreview.net/forum?id=kXwdL1cWOAi) by Hyung Won Chung, Xavier Garcia, Adam Roberts, Yi Tay, Orhan Firat, Sharan Narang, Noah Constant.
 1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (from Microsoft Research) released with the paper [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597) by Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang.

--- a/docs/source/en/_toctree.yml
+++ b/docs/source/en/_toctree.yml
@@ -770,6 +770,8 @@
        title: TVLT
      - local: model_doc/tvp
        title: TVP
+      - local: model_doc/udop
+        title: UDOP
      - local: model_doc/vilt
        title: ViLT
      - local: model_doc/vipllava

--- a/docs/source/en/index.md
+++ b/docs/source/en/index.md
@@ -279,6 +279,7 @@ Flax), PyTorch, and/or TensorFlow.
 |                         [TrOCR](model_doc/trocr)                         |       ✅        |         ❌         |      ❌      |
 |                          [TVLT](model_doc/tvlt)                          |       ✅        |         ❌         |      ❌      |
 |                           [TVP](model_doc/tvp)                           |       ✅        |         ❌         |      ❌      |
+|                          [UDOP](model_doc/udop)                          |       ✅        |         ❌         |      ❌      |
 |                           [UL2](model_doc/ul2)                           |       ✅        |         ✅         |      ✅      |
 |                          [UMT5](model_doc/umt5)                          |       ✅        |         ❌         |      ❌      |
 |                     [UniSpeech](model_doc/unispeech)                     |       ✅        |         ❌         |      ❌      |

--- a/docs/source/en/model_doc/udop.md
+++ b/docs/source/en/model_doc/udop.md
+<!--Copyright 2024 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+-->
+
+# UDOP
+
+## Overview
+
+The UDOP model was proposed in [Unifying Vision, Text, and Layout for Universal Document Processing](https://arxiv.org/abs/2212.02623) by Zineng Tang, Ziyi Yang, Guoxin Wang, Yuwei Fang, Yang Liu, Chenguang Zhu, Michael Zeng, Cha Zhang, Mohit Bansal.
+UDOP adopts an encoder-decoder Transformer architecture based on [T5](t5) for document AI tasks like document image classification, document parsing and document visual question answering.
+
+The abstract from the paper is the following:
+
+We propose Universal Document Processing (UDOP), a foundation Document AI model which unifies text, image, and layout modalities together with varied task formats, including document understanding and generation. UDOP leverages the spatial correlation between textual content and document image to model image, text, and layout modalities with one uniform representation. With a novel Vision-Text-Layout Transformer, UDOP unifies pretraining and multi-domain downstream tasks into a prompt-based sequence generation scheme. UDOP is pretrained on both large-scale unlabeled document corpora using innovative self-supervised objectives and diverse labeled data. UDOP also learns to generate document images from text and layout modalities via masked image reconstruction. To the best of our knowledge, this is the first time in the field of document AI that one model simultaneously achieves high-quality neural document editing and content customization. Our method sets the state-of-the-art on 9 Document AI tasks, e.g., document understanding and QA, across diverse data domains like finance reports, academic papers, and websites. UDOP ranks first on the leaderboard of the Document Understanding Benchmark (DUE).*
+
+<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers/model_doc/udop_architecture.jpg"
+alt="drawing" width="600"/>
+
+<small> UDOP architecture. Taken from the <a href="https://arxiv.org/abs/2212.02623">original paper.</a> </small>
+
+## Usage tips
+
+- In addition to *input_ids*, [`UdopForConditionalGeneration`] also expects the input `bbox`, which are
+  the bounding boxes (i.e. 2D-positions) of the input tokens. These can be obtained using an external OCR engine such
+  as Google's [Tesseract](https://github.com/tesseract-ocr/tesseract) (there's a [Python wrapper](https://pypi.org/project/pytesseract/) available). Each bounding box should be in (x0, y0, x1, y1) format, where (x0, y0) corresponds to the position of the upper left corner in the bounding box, and (x1, y1) represents the
+  position of the lower right corner. Note that one first needs to normalize the bounding boxes to be on a 0-1000
+  scale. To normalize, you can use the following function:
+
+```python
+def normalize_bbox(bbox, width, height):
+    return [
+        int(1000 * (bbox[0] / width)),
+        int(1000 * (bbox[1] / height)),
+        int(1000 * (bbox[2] / width)),
+        int(1000 * (bbox[3] / height)),
+    ]
+```
+
+Here, `width` and `height` correspond to the width and height of the original document in which the token
+occurs. Those can be obtained using the Python Image Library (PIL) library for example, as follows:
+
+```python
+from PIL import Image
+
+# Document can be a png, jpg, etc. PDFs must be converted to images.
+image = Image.open(name_of_your_document).convert("RGB")
+
+width, height = image.size
+```
+
+- At inference time, it's recommended to use the `generate` method to autoregressively generate text given a document image.
+- One can use [`UdopProcessor`] to prepare images and text for the model. By default, this class uses the Tesseract engine to extract a list of words
+and boxes (coordinates) from a given document. Its functionality is equivalent to that of [`LayoutLMv3Processor`], hence it supports passing either
+`apply_ocr=False` in case you prefer to use your own OCR engine or `apply_ocr=True` in case you want the default OCR engine to be used.
+
+This model was contributed by [nielsr](https://huggingface.co/nielsr).
+The original code can be found [here](https://github.com/microsoft/UDOP).
+
+
+## UdopConfig
+
+[[autodoc]] UdopConfig
+
+## UdopTokenizer
+
+[[autodoc]] UdopTokenizer
+    - build_inputs_with_special_tokens
+    - get_special_tokens_mask
+    - create_token_type_ids_from_sequences
+    - save_vocabulary
+
+## UdopTokenizerFast
+
+[[autodoc]] UdopTokenizerFast
+
+## UdopProcessor
+
+[[autodoc]] UdopProcessor
+    - __call__
+
+## UdopModel
+
+[[autodoc]] UdopModel
+    - forward
+
+## UdopForConditionalGeneration
+
+[[autodoc]] UdopForConditionalGeneration
+    - forward
+
+## UdopEncoderModel
+
+[[autodoc]] UdopEncoderModel
+    - forward
\ No newline at end of file
--- a/src/transformers/__init__.py
+++ b/src/transformers/__init__.py
@@ -856,6 +856,11 @@ _import_structure = {
        "TvpConfig",
        "TvpProcessor",
    ],
+    "models.udop": [
+        "UDOP_PRETRAINED_CONFIG_ARCHIVE_MAP",
+        "UdopConfig",
+        "UdopProcessor",
+    ],
    "models.umt5": ["UMT5Config"],
    "models.unispeech": [
        "UNISPEECH_PRETRAINED_CONFIG_ARCHIVE_MAP",
@@ -1135,6 +1140,7 @@ else:
    _import_structure["models.speech_to_text"].append("Speech2TextTokenizer")
    _import_structure["models.speecht5"].append("SpeechT5Tokenizer")
    _import_structure["models.t5"].append("T5Tokenizer")
+    _import_structure["models.udop"].append("UdopTokenizer")
    _import_structure["models.xglm"].append("XGLMTokenizer")
    _import_structure["models.xlm_prophetnet"].append("XLMProphetNetTokenizer")
    _import_structure["models.xlm_roberta"].append("XLMRobertaTokenizer")
@@ -1214,6 +1220,7 @@ else:
    _import_structure["models.splinter"].append("SplinterTokenizerFast")
    _import_structure["models.squeezebert"].append("SqueezeBertTokenizerFast")
    _import_structure["models.t5"].append("T5TokenizerFast")
+    _import_structure["models.udop"].append("UdopTokenizerFast")
    _import_structure["models.whisper"].append("WhisperTokenizerFast")
    _import_structure["models.xglm"].append("XGLMTokenizerFast")
    _import_structure["models.xlm_roberta"].append("XLMRobertaTokenizerFast")
@@ -3411,6 +3418,15 @@ else:
            "TvpPreTrainedModel",
        ]
    )
+    _import_structure["models.udop"].extend(
+        [
+            "UDOP_PRETRAINED_MODEL_ARCHIVE_LIST",
+            "UdopEncoderModel",
+            "UdopForConditionalGeneration",
+            "UdopModel",
+            "UdopPreTrainedModel",
+        ],
+    )
    _import_structure["models.umt5"].extend(
        [
            "UMT5EncoderModel",
@@ -5640,6 +5656,7 @@ if TYPE_CHECKING:
        TvpConfig,
        TvpProcessor,
    )
+    from .models.udop import UDOP_PRETRAINED_CONFIG_ARCHIVE_MAP, UdopConfig, UdopProcessor
    from .models.umt5 import UMT5Config
    from .models.unispeech import (
        UNISPEECH_PRETRAINED_CONFIG_ARCHIVE_MAP,
@@ -5915,6 +5932,7 @@ if TYPE_CHECKING:
        from .models.speech_to_text import Speech2TextTokenizer
        from .models.speecht5 import SpeechT5Tokenizer
        from .models.t5 import T5Tokenizer
+        from .models.udop import UdopTokenizer
        from .models.xglm import XGLMTokenizer
        from .models.xlm_prophetnet import XLMProphetNetTokenizer
        from .models.xlm_roberta import XLMRobertaTokenizer
@@ -5987,6 +6005,7 @@ if TYPE_CHECKING:
        from .models.splinter import SplinterTokenizerFast
        from .models.squeezebert import SqueezeBertTokenizerFast
        from .models.t5 import T5TokenizerFast
+        from .models.udop import UdopTokenizerFast
        from .models.whisper import WhisperTokenizerFast
        from .models.xglm import XGLMTokenizerFast
        from .models.xlm_roberta import XLMRobertaTokenizerFast
@@ -7827,6 +7846,13 @@ if TYPE_CHECKING:
            TvpModel,
            TvpPreTrainedModel,
        )
+        from .models.udop import (
+            UDOP_PRETRAINED_MODEL_ARCHIVE_LIST,
+            UdopEncoderModel,
+            UdopForConditionalGeneration,
+            UdopModel,
+            UdopPreTrainedModel,
+        )
        from .models.umt5 import (
            UMT5EncoderModel,
            UMT5ForConditionalGeneration,

--- a/src/transformers/convert_slow_tokenizer.py
+++ b/src/transformers/convert_slow_tokenizer.py
@@ -1039,6 +1039,17 @@ class T5Converter(SpmConverter):
        )


+class UdopConverter(SpmConverter):
+    def post_processor(self):
+        return processors.TemplateProcessing(
+            single=["$A", "</s>"],
+            pair=["$A", "</s>", "$B", "</s>"],
+            special_tokens=[
+                ("</s>", self.original_tokenizer.convert_tokens_to_ids("</s>")),
+            ],
+        )
+
+
 class WhisperConverter(Converter):
    def converted(self) -> Tokenizer:
        vocab = self.original_tokenizer.encoder
@@ -1471,6 +1482,7 @@ SLOW_TO_FAST_CONVERTERS = {
    "SeamlessM4TTokenizer": SeamlessM4TConverter,
    "SqueezeBertTokenizer": BertConverter,
    "T5Tokenizer": T5Converter,
+    "UdopTokenizer": UdopConverter,
    "WhisperTokenizer": WhisperConverter,
    "XLMRobertaTokenizer": XLMRobertaConverter,
    "XLNetTokenizer": XLNetConverter,

--- a/src/transformers/models/__init__.py
+++ b/src/transformers/models/__init__.py
@@ -220,6 +220,7 @@ from . import (
    trocr,
    tvlt,
    tvp,
+    udop,
    umt5,
    unispeech,
    unispeech_sat,

--- a/src/transformers/models/auto/configuration_auto.py
+++ b/src/transformers/models/auto/configuration_auto.py
@@ -231,6 +231,7 @@ CONFIG_MAPPING_NAMES = OrderedDict(
        ("trocr", "TrOCRConfig"),
        ("tvlt", "TvltConfig"),
        ("tvp", "TvpConfig"),
+        ("udop", "UdopConfig"),
        ("umt5", "UMT5Config"),
        ("unispeech", "UniSpeechConfig"),
        ("unispeech-sat", "UniSpeechSatConfig"),
@@ -454,6 +455,7 @@ CONFIG_ARCHIVE_MAP_MAPPING_NAMES = OrderedDict(
        ("transfo-xl", "TRANSFO_XL_PRETRAINED_CONFIG_ARCHIVE_MAP"),
        ("tvlt", "TVLT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
        ("tvp", "TVP_PRETRAINED_CONFIG_ARCHIVE_MAP"),
+        ("udop", "UDOP_PRETRAINED_CONFIG_ARCHIVE_MAP"),
        ("unispeech", "UNISPEECH_PRETRAINED_CONFIG_ARCHIVE_MAP"),
        ("unispeech-sat", "UNISPEECH_SAT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
        ("univnet", "UNIVNET_PRETRAINED_CONFIG_ARCHIVE_MAP"),
@@ -715,6 +717,7 @@ MODEL_NAMES_MAPPING = OrderedDict(
        ("trocr", "TrOCR"),
        ("tvlt", "TVLT"),
        ("tvp", "TVP"),
+        ("udop", "UDOP"),
        ("ul2", "UL2"),
        ("umt5", "UMT5"),
        ("unispeech", "UniSpeech"),

--- a/src/transformers/models/auto/image_processing_auto.py
+++ b/src/transformers/models/auto/image_processing_auto.py
@@ -108,6 +108,7 @@ IMAGE_PROCESSOR_MAPPING_NAMES = OrderedDict(
        ("timesformer", "VideoMAEImageProcessor"),
        ("tvlt", "TvltImageProcessor"),
        ("tvp", "TvpImageProcessor"),
+        ("udop", "LayoutLMv3ImageProcessor"),
        ("upernet", "SegformerImageProcessor"),
        ("van", "ConvNextImageProcessor"),
        ("videomae", "VideoMAEImageProcessor"),

--- a/src/transformers/models/auto/modeling_auto.py
+++ b/src/transformers/models/auto/modeling_auto.py
@@ -219,6 +219,7 @@ MODEL_MAPPING_NAMES = OrderedDict(
        ("transfo-xl", "TransfoXLModel"),
        ("tvlt", "TvltModel"),
        ("tvp", "TvpModel"),
+        ("udop", "UdopModel"),
        ("umt5", "UMT5Model"),
        ("unispeech", "UniSpeechModel"),
        ("unispeech-sat", "UniSpeechSatModel"),

--- a/src/transformers/models/auto/tokenization_auto.py
+++ b/src/transformers/models/auto/tokenization_auto.py
@@ -418,6 +418,13 @@ else:
            ("tapex", ("TapexTokenizer", None)),
            ("transfo-xl", ("TransfoXLTokenizer", None)),
            ("tvp", ("BertTokenizer", "BertTokenizerFast" if is_tokenizers_available() else None)),
+            (
+                "udop",
+                (
+                    "UdopTokenizer" if is_sentencepiece_available() else None,
+                    "UdopTokenizerFast" if is_tokenizers_available() else None,
+                ),
+            ),
            (
                "umt5",
                (

--- a/src/transformers/models/udop/__init__.py
+++ b/src/transformers/models/udop/__init__.py
+# Copyright 2024 The HuggingFace Team. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from typing import TYPE_CHECKING
+
+from ...utils import (
+    OptionalDependencyNotAvailable,
+    _LazyModule,
+    is_sentencepiece_available,
+    is_tokenizers_available,
+    is_torch_available,
+)
+
+
+_import_structure = {
+    "configuration_udop": ["UDOP_PRETRAINED_CONFIG_ARCHIVE_MAP", "UdopConfig"],
+    "processing_udop": ["UdopProcessor"],
+}
+
+try:
+    if not is_sentencepiece_available():
+        raise OptionalDependencyNotAvailable()
+except OptionalDependencyNotAvailable:
+    pass
+else:
+    _import_structure["tokenization_udop"] = ["UdopTokenizer"]
+
+try:
+    if not is_tokenizers_available():
+        raise OptionalDependencyNotAvailable()
+except OptionalDependencyNotAvailable:
+    pass
+else:
+    _import_structure["tokenization_udop_fast"] = ["UdopTokenizerFast"]
+
+try:
+    if not is_torch_available():
+        raise OptionalDependencyNotAvailable()
+except OptionalDependencyNotAvailable:
+    pass
+else:
+    _import_structure["modeling_udop"] = [
+        "UDOP_PRETRAINED_MODEL_ARCHIVE_LIST",
+        "UdopForConditionalGeneration",
+        "UdopPreTrainedModel",
+        "UdopModel",
+        "UdopEncoderModel",
+    ]
+
+if TYPE_CHECKING:
+    from .configuration_udop import UDOP_PRETRAINED_CONFIG_ARCHIVE_MAP, UdopConfig
+    from .processing_udop import UdopProcessor
+
+    try:
+        if not is_sentencepiece_available():
+            raise OptionalDependencyNotAvailable()
+    except OptionalDependencyNotAvailable:
+        pass
+    else:
+        from .tokenization_udop import UdopTokenizer
+
+    try:
+        if not is_tokenizers_available():
+            raise OptionalDependencyNotAvailable()
+    except OptionalDependencyNotAvailable:
+        pass
+    else:
+        from .tokenization_udop_fast import UdopTokenizerFast
+
+    try:
+        if not is_torch_available():
+            raise OptionalDependencyNotAvailable()
+    except OptionalDependencyNotAvailable:
+        pass
+    else:
+        from .modeling_udop import (
+            UDOP_PRETRAINED_MODEL_ARCHIVE_LIST,
+            UdopEncoderModel,
+            UdopForConditionalGeneration,
+            UdopModel,
+            UdopPreTrainedModel,
+        )
+
+else:
+    import sys
+
+    sys.modules[__name__] = _LazyModule(__name__, globals()["__file__"], _import_structure, module_spec=__spec__)