Add BioGPT (#20420)

* biogpt initial commit

* updated init

* fix faster decoding with use_cache

* 1. fix input_ids and input_embeds with correct device
2. added _keys_to_ignore_on_load_missing
3. updated prepare_inputs_for_generation

* add activation_dropout and scale_embedding

* replace fsmt attention with bart attention

* added test

* run make fix-copies

* doc init and fix build

* updated README with proper information

* 1. added tips to docs
2. updated BioGptTokenizer func

* 1. added tokenizer test
2. refactor tokenizer

* make fixup

* add biogpt fairseq to hf converter

* updated layer names more
similar to original checkpoints

* config update doc string and set defaults

* added "#copied" from bart model and
updated doc strings

* enable model_input_names in tokenizer

* 1.  positionalembedding depending on attention_mask
2. added attention mask to prepare for generation

* added test to verify past and generation

* BioGptLMHeadModel -> BioGptForCausalLM

* fix typo

* tokenization and test
Copyright and updated assertion

* updated Copyright and
one func at time in line

* Copyright updates and
minor doc fix

* replace assertion with ValueError

* rm extra space

* added code syntax

* revert cmnt position change

* add tokenizer to auto

* updated doc string

* tokenizer doc string update

* biogpt hub model update to microsoft/biogpt

* make fixup

* rm cmnt to fix flake8 5.0.4 vs 6 error

README.md

@@ -272,6 +272,7 @@ Current number of checkpoints: ![](https://img.shields.io/endpoint?url=https://h
 1. **[BERTweet](https://huggingface.co/docs/transformers/model_doc/bertweet)** (from VinAI Research) released with the paper [BERTweet: A pre-trained language model for English Tweets](https://aclanthology.org/2020.emnlp-demos.2/) by Dat Quoc Nguyen, Thanh Vu and Anh Tuan Nguyen.
 1. **[BigBird-Pegasus](https://huggingface.co/docs/transformers/model_doc/bigbird_pegasus)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
 1. **[BigBird-RoBERTa](https://huggingface.co/docs/transformers/model_doc/big_bird)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
+1. **[BioGpt](https://huggingface.co/docs/transformers/main/model_doc/biogpt)** (from Microsoft Research AI4Science) released with the paper [BioGPT: generative pre-trained transformer for biomedical text generation and mining](https://academic.oup.com/bib/advance-article/doi/10.1093/bib/bbac409/6713511?guestAccessKey=a66d9b5d-4f83-4017-bb52-405815c907b9) by Renqian Luo, Liai Sun, Yingce Xia, Tao Qin, Sheng Zhang, Hoifung Poon and Tie-Yan Liu.
 1. **[Blenderbot](https://huggingface.co/docs/transformers/model_doc/blenderbot)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
 1. **[BlenderbotSmall](https://huggingface.co/docs/transformers/model_doc/blenderbot-small)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
 1. **[BLOOM](https://huggingface.co/docs/transformers/model_doc/bloom)** (from BigScience workshop) released by the [BigScience Workshop](https://bigscience.huggingface.co/).

README_es.md

@@ -272,6 +272,7 @@ Número actual de puntos de control: ![](https://img.shields.io/endpoint?url=htt
 1. **[BERTweet](https://huggingface.co/docs/transformers/model_doc/bertweet)** (from VinAI Research) released with the paper [BERTweet: A pre-trained language model for English Tweets](https://aclanthology.org/2020.emnlp-demos.2/) by Dat Quoc Nguyen, Thanh Vu and Anh Tuan Nguyen.
 1. **[BigBird-Pegasus](https://huggingface.co/docs/transformers/model_doc/bigbird_pegasus)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
 1. **[BigBird-RoBERTa](https://huggingface.co/docs/transformers/model_doc/big_bird)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
+1. **[BioGpt](https://huggingface.co/docs/transformers/main/model_doc/biogpt)** (from Microsoft Research AI4Science) released with the paper [BioGPT: generative pre-trained transformer for biomedical text generation and mining](https://academic.oup.com/bib/advance-article/doi/10.1093/bib/bbac409/6713511?guestAccessKey=a66d9b5d-4f83-4017-bb52-405815c907b9) by Renqian Luo, Liai Sun, Yingce Xia, Tao Qin, Sheng Zhang, Hoifung Poon and Tie-Yan Liu.
 1. **[Blenderbot](https://huggingface.co/docs/transformers/model_doc/blenderbot)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
 1. **[BlenderbotSmall](https://huggingface.co/docs/transformers/model_doc/blenderbot-small)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
 1. **[BLOOM](https://huggingface.co/docs/transformers/model_doc/bloom)** (from BigScience workshop) released by the [BigScience Workshop](https://bigscience.huggingface.co/).

README_ja.md

@@ -307,6 +307,7 @@ Flax、PyTorch、TensorFlowをcondaでインストールする方法は、それ
 1. **[BERTweet](https://huggingface.co/docs/transformers/model_doc/bertweet)** (from VinAI Research) released with the paper [BERTweet: A pre-trained language model for English Tweets](https://aclanthology.org/2020.emnlp-demos.2/) by Dat Quoc Nguyen, Thanh Vu and Anh Tuan Nguyen.
 1. **[BigBird-Pegasus](https://huggingface.co/docs/transformers/model_doc/bigbird_pegasus)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
 1. **[BigBird-RoBERTa](https://huggingface.co/docs/transformers/model_doc/big_bird)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
+1. **[BioGpt](https://huggingface.co/docs/transformers/main/model_doc/biogpt)** (from Microsoft Research AI4Science) released with the paper [BioGPT: generative pre-trained transformer for biomedical text generation and mining](https://academic.oup.com/bib/advance-article/doi/10.1093/bib/bbac409/6713511?guestAccessKey=a66d9b5d-4f83-4017-bb52-405815c907b9) by Renqian Luo, Liai Sun, Yingce Xia, Tao Qin, Sheng Zhang, Hoifung Poon and Tie-Yan Liu.
 1. **[Blenderbot](https://huggingface.co/docs/transformers/model_doc/blenderbot)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
 1. **[BlenderbotSmall](https://huggingface.co/docs/transformers/model_doc/blenderbot-small)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
 1. **[BLOOM](https://huggingface.co/docs/transformers/model_doc/bloom)** (from BigScience workshop) released by the [BigScience Workshop](https://bigscience.huggingface.co/).

README_ko.md

@@ -222,6 +222,7 @@ Flax, PyTorch, TensorFlow 설치 페이지에서 이들을 conda로 설치하는
 1. **[BERTweet](https://huggingface.co/docs/transformers/model_doc/bertweet)** (from VinAI Research) released with the paper [BERTweet: A pre-trained language model for English Tweets](https://aclanthology.org/2020.emnlp-demos.2/) by Dat Quoc Nguyen, Thanh Vu and Anh Tuan Nguyen.
 1. **[BigBird-Pegasus](https://huggingface.co/docs/transformers/model_doc/bigbird_pegasus)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
 1. **[BigBird-RoBERTa](https://huggingface.co/docs/transformers/model_doc/big_bird)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
+1. **[BioGpt](https://huggingface.co/docs/transformers/main/model_doc/biogpt)** (from Microsoft Research AI4Science) released with the paper [BioGPT: generative pre-trained transformer for biomedical text generation and mining](https://academic.oup.com/bib/advance-article/doi/10.1093/bib/bbac409/6713511?guestAccessKey=a66d9b5d-4f83-4017-bb52-405815c907b9) by Renqian Luo, Liai Sun, Yingce Xia, Tao Qin, Sheng Zhang, Hoifung Poon and Tie-Yan Liu.
 1. **[Blenderbot](https://huggingface.co/docs/transformers/model_doc/blenderbot)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
 1. **[BlenderbotSmall](https://huggingface.co/docs/transformers/model_doc/blenderbot-small)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
 1. **[BLOOM](https://huggingface.co/docs/transformers/model_doc/bloom)** (from BigScience workshop) released by the [BigScience Workshop](https://bigscience.huggingface.co/).

README_zh-hans.md

@@ -246,6 +246,7 @@ conda install -c huggingface transformers
 1. **[BERTweet](https://huggingface.co/docs/transformers/model_doc/bertweet)** (来自 VinAI Research) 伴随论文 [BERTweet: A pre-trained language model for English Tweets](https://aclanthology.org/2020.emnlp-demos.2/) 由 Dat Quoc Nguyen, Thanh Vu and Anh Tuan Nguyen 发布。
 1. **[BigBird-Pegasus](https://huggingface.co/docs/transformers/model_doc/bigbird_pegasus)** (来自 Google Research) 伴随论文 [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) 由 Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed 发布。
 1. **[BigBird-RoBERTa](https://huggingface.co/docs/transformers/model_doc/big_bird)** (来自 Google Research) 伴随论文 [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) 由 Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed 发布。
+1. **[BioGpt](https://huggingface.co/docs/transformers/main/model_doc/biogpt)** (来自 Microsoft Research AI4Science) 伴随论文 [BioGPT: generative pre-trained transformer for biomedical text generation and mining](https://academic.oup.com/bib/advance-article/doi/10.1093/bib/bbac409/6713511?guestAccessKey=a66d9b5d-4f83-4017-bb52-405815c907b9) 由 Renqian Luo, Liai Sun, Yingce Xia, Tao Qin, Sheng Zhang, Hoifung Poon and Tie-Yan Liu 发布。
 1. **[Blenderbot](https://huggingface.co/docs/transformers/model_doc/blenderbot)** (来自 Facebook) 伴随论文 [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) 由 Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston 发布。
 1. **[BlenderbotSmall](https://huggingface.co/docs/transformers/model_doc/blenderbot-small)** (来自 Facebook) 伴随论文 [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) 由 Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston 发布。
 1. **[BLOOM](https://huggingface.co/docs/transformers/model_doc/bloom)** (from BigScience workshop) released by the [BigScience Workshop](https://bigscience.huggingface.co/).

README_zh-hant.md

@@ -258,6 +258,7 @@ conda install -c huggingface transformers
 1. **[BERTweet](https://huggingface.co/docs/transformers/model_doc/bertweet)** (from VinAI Research) released with the paper [BERTweet: A pre-trained language model for English Tweets](https://aclanthology.org/2020.emnlp-demos.2/) by Dat Quoc Nguyen, Thanh Vu and Anh Tuan Nguyen.
 1. **[BigBird-Pegasus](https://huggingface.co/docs/transformers/model_doc/bigbird_pegasus)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
 1. **[BigBird-RoBERTa](https://huggingface.co/docs/transformers/model_doc/big_bird)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
+1. **[BioGpt](https://huggingface.co/docs/transformers/main/model_doc/biogpt)** (from Microsoft Research AI4Science) released with the paper [BioGPT: generative pre-trained transformer for biomedical text generation and mining](https://academic.oup.com/bib/advance-article/doi/10.1093/bib/bbac409/6713511?guestAccessKey=a66d9b5d-4f83-4017-bb52-405815c907b9) by Renqian Luo, Liai Sun, Yingce Xia, Tao Qin, Sheng Zhang, Hoifung Poon and Tie-Yan Liu.
 1. **[Blenderbot](https://huggingface.co/docs/transformers/model_doc/blenderbot)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
 1. **[BlenderbotSmall](https://huggingface.co/docs/transformers/model_doc/blenderbot-small)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
 1. **[BLOOM](https://huggingface.co/docs/transformers/model_doc/bloom)** (from BigScience workshop) released by the [BigScience Workshop](https://bigscience.huggingface.co/).

docs/source/en/_toctree.yml

@@ -209,6 +209,8 @@
         title: BigBird
       - local: model_doc/bigbird_pegasus
         title: BigBirdPegasus
+      - local: model_doc/biogpt
+        title: BioGpt
       - local: model_doc/blenderbot
         title: Blenderbot
       - local: model_doc/blenderbot-small

docs/source/en/index.mdx

@@ -60,6 +60,7 @@ The documentation is organized into five sections:
 1. **[BERTweet](model_doc/bertweet)** (from VinAI Research) released with the paper [BERTweet: A pre-trained language model for English Tweets](https://aclanthology.org/2020.emnlp-demos.2/) by Dat Quoc Nguyen, Thanh Vu and Anh Tuan Nguyen.
 1. **[BigBird-Pegasus](model_doc/bigbird_pegasus)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
 1. **[BigBird-RoBERTa](model_doc/big_bird)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
+1. **[BioGpt](model_doc/biogpt)** (from Microsoft Research AI4Science) released with the paper [BioGPT: generative pre-trained transformer for biomedical text generation and mining](https://academic.oup.com/bib/advance-article/doi/10.1093/bib/bbac409/6713511?guestAccessKey=a66d9b5d-4f83-4017-bb52-405815c907b9) by Renqian Luo, Liai Sun, Yingce Xia, Tao Qin, Sheng Zhang, Hoifung Poon and Tie-Yan Liu.
 1. **[Blenderbot](model_doc/blenderbot)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
 1. **[BlenderbotSmall](model_doc/blenderbot-small)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
 1. **[BLOOM](model_doc/bloom)** (from BigScience workshop) released by the [BigScience Workshop](https://bigscience.huggingface.co/).
@@ -229,6 +230,7 @@ Flax), PyTorch, and/or TensorFlow.
 |        Bert Generation        |       ✅       |       ❌       |       ✅        |         ❌         |      ❌      |
 |            BigBird            |       ✅       |       ✅       |       ✅        |         ❌         |      ✅      |
 |        BigBird-Pegasus        |       ❌       |       ❌       |       ✅        |         ❌         |      ❌      |
+|            BioGpt             |       ✅       |       ❌       |       ✅        |         ❌         |      ❌      |
 |          Blenderbot           |       ✅       |       ✅       |       ✅        |         ✅         |      ✅      |
 |        BlenderbotSmall        |       ✅       |       ✅       |       ✅        |         ✅         |      ✅      |
 |             BLOOM             |       ❌       |       ✅       |       ✅        |         ❌         |      ❌      |

docs/source/en/model_doc/biogpt.mdx

+<!--Copyright 2022 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+-->
+
+# BioGPT
+
+## Overview
+
+The BioGPT model was proposed in [BioGPT: generative pre-trained transformer for biomedical text generation and mining
+](https://academic.oup.com/bib/advance-article/doi/10.1093/bib/bbac409/6713511?guestAccessKey=a66d9b5d-4f83-4017-bb52-405815c907b9) by Renqian Luo, Liai Sun, Yingce Xia, Tao Qin, Sheng Zhang, Hoifung Poon and Tie-Yan Liu. BioGPT is a domain-specific generative pre-trained Transformer language model for biomedical text generation and mining. BioGPT follows the Transformer language model backbone, and is pre-trained on 15M PubMed abstracts from scratch.
+
+The abstract from the paper is the following:
+
+*Pre-trained language models have attracted increasing attention in the biomedical domain, inspired by their great success in the general natural language domain. Among the two main branches of pre-trained language models in the general language domain, i.e. BERT (and its variants) and GPT (and its variants), the first one has been extensively studied in the biomedical domain, such as BioBERT and PubMedBERT. While they have achieved great success on a variety of discriminative downstream biomedical tasks, the lack of generation ability constrains their application scope. In this paper, we propose BioGPT, a domain-specific generative Transformer language model pre-trained on large-scale biomedical literature. We evaluate BioGPT on six biomedical natural language processing tasks and demonstrate that our model outperforms previous models on most tasks. Especially, we get 44.98%, 38.42% and 40.76% F1 score on BC5CDR, KD-DTI and DDI end-to-end relation extraction tasks, respectively, and 78.2% accuracy on PubMedQA, creating a new record. Our case study on text generation further demonstrates the advantage of BioGPT on biomedical literature to generate fluent descriptions for biomedical terms.*
+
+Tips:
+
+- BioGPT is a model with absolute position embeddings so it’s usually advised to pad the inputs on the right rather than the left.
+- BioGPT was trained with a causal language modeling (CLM) objective and is therefore powerful at predicting the next token in a sequence. Leveraging this feature allows BioGPT to generate syntactically coherent text as it can be observed in the run_generation.py example script.
+- The model can take the `past_key_values` (for PyTorch) as input, which is the previously computed key/value attention pairs. Using this (past_key_values or past) value prevents the model from re-computing pre-computed values in the context of text generation. For PyTorch, see past_key_values argument of the BioGptForCausalLM.forward() method for more information on its usage.
+
+This model was contributed by [kamalkraj](https://huggingface.co/kamalkraj). The original code can be found [here](https://github.com/microsoft/BioGPT).
+
+## BioGptConfig
+
+[[autodoc]] BioGptConfig
+
+
+## BioGptTokenizer
+
+[[autodoc]] BioGptTokenizer
+    - save_vocabulary
+
+
+## BioGptModel
+
+[[autodoc]] BioGptModel
+    - forward
+
+
+## BioGptForCausalLM
+
+[[autodoc]] BioGptForCausalLM
+    - forward
\ No newline at end of file

src/transformers/__init__.py

@@ -160,6 +160,7 @@ _import_structure = {
         "BIGBIRD_PEGASUS_PRETRAINED_CONFIG_ARCHIVE_MAP",
         "BigBirdPegasusConfig",
     ],
+    "models.biogpt": ["BIOGPT_PRETRAINED_CONFIG_ARCHIVE_MAP", "BioGptConfig", "BioGptTokenizer"],
     "models.blenderbot": ["BLENDERBOT_PRETRAINED_CONFIG_ARCHIVE_MAP", "BlenderbotConfig", "BlenderbotTokenizer"],
     "models.blenderbot_small": [
         "BLENDERBOT_SMALL_PRETRAINED_CONFIG_ARCHIVE_MAP",
@@ -1046,6 +1047,14 @@ else:
             "BigBirdPegasusPreTrainedModel",
         ]
     )
+    _import_structure["models.biogpt"].extend(
+        [
+            "BIOGPT_PRETRAINED_MODEL_ARCHIVE_LIST",
+            "BioGptForCausalLM",
+            "BioGptModel",
+            "BioGptPreTrainedModel",
+        ]
+    )
     _import_structure["models.blenderbot"].extend(
         [
             "BLENDERBOT_PRETRAINED_MODEL_ARCHIVE_LIST",
@@ -3392,6 +3401,7 @@ if TYPE_CHECKING:
     from .models.bertweet import BertweetTokenizer
     from .models.big_bird import BIG_BIRD_PRETRAINED_CONFIG_ARCHIVE_MAP, BigBirdConfig
     from .models.bigbird_pegasus import BIGBIRD_PEGASUS_PRETRAINED_CONFIG_ARCHIVE_MAP, BigBirdPegasusConfig
+    from .models.biogpt import BIOGPT_PRETRAINED_CONFIG_ARCHIVE_MAP, BioGptConfig, BioGptTokenizer
     from .models.blenderbot import BLENDERBOT_PRETRAINED_CONFIG_ARCHIVE_MAP, BlenderbotConfig, BlenderbotTokenizer
     from .models.blenderbot_small import (
         BLENDERBOT_SMALL_PRETRAINED_CONFIG_ARCHIVE_MAP,
@@ -4166,6 +4176,12 @@ if TYPE_CHECKING:
             BigBirdPegasusModel,
             BigBirdPegasusPreTrainedModel,
         )
+        from .models.biogpt import (
+            BIOGPT_PRETRAINED_MODEL_ARCHIVE_LIST,
+            BioGptForCausalLM,
+            BioGptModel,
+            BioGptPreTrainedModel,
+        )
         from .models.blenderbot import (
             BLENDERBOT_PRETRAINED_MODEL_ARCHIVE_LIST,
             BlenderbotForCausalLM,

src/transformers/models/__init__.py

@@ -30,6 +30,7 @@ from . import (
     bertweet,
     big_bird,
     bigbird_pegasus,
+    biogpt,
     blenderbot,
     blenderbot_small,
     bloom,

src/transformers/models/auto/configuration_auto.py

@@ -37,6 +37,7 @@ CONFIG_MAPPING_NAMES = OrderedDict(
         ("bert-generation", "BertGenerationConfig"),
         ("big_bird", "BigBirdConfig"),
         ("bigbird_pegasus", "BigBirdPegasusConfig"),
+        ("biogpt", "BioGptConfig"),
         ("blenderbot", "BlenderbotConfig"),
         ("blenderbot-small", "BlenderbotSmallConfig"),
         ("bloom", "BloomConfig"),
@@ -189,6 +190,7 @@ CONFIG_ARCHIVE_MAP_MAPPING_NAMES = OrderedDict(
         ("bert", "BERT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
         ("big_bird", "BIG_BIRD_PRETRAINED_CONFIG_ARCHIVE_MAP"),
         ("bigbird_pegasus", "BIGBIRD_PEGASUS_PRETRAINED_CONFIG_ARCHIVE_MAP"),
+        ("biogpt", "BIOGPT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
         ("blenderbot", "BLENDERBOT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
         ("blenderbot-small", "BLENDERBOT_SMALL_PRETRAINED_CONFIG_ARCHIVE_MAP"),
         ("bloom", "BLOOM_PRETRAINED_CONFIG_ARCHIVE_MAP"),
@@ -331,6 +333,7 @@ MODEL_NAMES_MAPPING = OrderedDict(
         ("bertweet", "BERTweet"),
         ("big_bird", "BigBird"),
         ("bigbird_pegasus", "BigBird-Pegasus"),
+        ("biogpt", "BioGpt"),
         ("blenderbot", "Blenderbot"),
         ("blenderbot-small", "BlenderbotSmall"),
         ("bloom", "BLOOM"),

src/transformers/models/auto/modeling_auto.py

@@ -36,6 +36,7 @@ MODEL_MAPPING_NAMES = OrderedDict(
         ("bert-generation", "BertGenerationEncoder"),
         ("big_bird", "BigBirdModel"),
         ("bigbird_pegasus", "BigBirdPegasusModel"),
+        ("biogpt", "BioGptModel"),
         ("blenderbot", "BlenderbotModel"),
         ("blenderbot-small", "BlenderbotSmallModel"),
         ("bloom", "BloomModel"),
@@ -308,6 +309,7 @@ MODEL_FOR_CAUSAL_LM_MAPPING_NAMES = OrderedDict(
         ("bert-generation", "BertGenerationDecoder"),
         ("big_bird", "BigBirdForCausalLM"),
         ("bigbird_pegasus", "BigBirdPegasusForCausalLM"),
+        ("biogpt", "BioGptForCausalLM"),
         ("blenderbot", "BlenderbotForCausalLM"),
         ("blenderbot-small", "BlenderbotSmallForCausalLM"),
         ("bloom", "BloomForCausalLM"),

src/transformers/models/auto/tokenization_auto.py

@@ -74,6 +74,7 @@ else:
                 ),
             ),
             ("bigbird_pegasus", ("PegasusTokenizer", "PegasusTokenizerFast" if is_tokenizers_available() else None)),
+            ("biogpt", ("BioGptTokenizer", None)),
             ("blenderbot", ("BlenderbotTokenizer", "BlenderbotTokenizerFast")),
             ("blenderbot-small", ("BlenderbotSmallTokenizer", None)),
             ("bloom", (None, "BloomTokenizerFast" if is_tokenizers_available() else None)),

src/transformers/models/biogpt/__init__.py

+# flake8: noqa
+# There's no way to ignore "F401 '...' imported but unused" warnings in this
+# module, but to preserve other warnings. So, don't check this module at all.
+
+# Copyright 2022 The HuggingFace Team. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+from typing import TYPE_CHECKING
+
+# rely on isort to merge the imports
+from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_tokenizers_available, is_torch_available
+
+
+_import_structure = {
+    "configuration_biogpt": ["BIOGPT_PRETRAINED_CONFIG_ARCHIVE_MAP", "BioGptConfig"],
+    "tokenization_biogpt": ["BioGptTokenizer"],
+}
+
+try:
+    if not is_torch_available():
+        raise OptionalDependencyNotAvailable()
+except OptionalDependencyNotAvailable:
+    pass
+else:
+    _import_structure["modeling_biogpt"] = [
+        "BIOGPT_PRETRAINED_MODEL_ARCHIVE_LIST",
+        "BioGptForCausalLM",
+        "BioGptModel",
+        "BioGptPreTrainedModel",
+    ]
+
+
+if TYPE_CHECKING:
+    from .configuration_biogpt import BIOGPT_PRETRAINED_CONFIG_ARCHIVE_MAP, BioGptConfig
+    from .tokenization_biogpt import BioGptTokenizer
+
+    try:
+        if not is_torch_available():
+            raise OptionalDependencyNotAvailable()
+    except OptionalDependencyNotAvailable:
+        pass
+    else:
+        from .modeling_biogpt import (
+            BIOGPT_PRETRAINED_MODEL_ARCHIVE_LIST,
+            BioGptForCausalLM,
+            BioGptModel,
+            BioGptPreTrainedModel,
+        )
+
+
+else:
+    import sys
+
+    sys.modules[__name__] = _LazyModule(__name__, globals()["__file__"], _import_structure, module_spec=__spec__)

src/transformers/models/biogpt/configuration_biogpt.py

+# coding=utf-8
+# Copyright 2022 The HuggingFace Team and Microsoft Research AI4Science 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.
+""" BioGPT model configuration"""
+
+from ...configuration_utils import PretrainedConfig
+from ...utils import logging
+
+
+logger = logging.get_logger(__name__)
+
+BIOGPT_PRETRAINED_CONFIG_ARCHIVE_MAP = {
+    "microsoft/biogpt": "https://huggingface.co/microsoft/biogpt/resolve/main/config.json",
+    # See all BioGPT models at https://huggingface.co/models?filter=biogpt
+}
+
+
+class BioGptConfig(PretrainedConfig):
+    r"""
+    This is the configuration class to store the configuration of a [`BioGptModel`]. It is used to instantiate an
+    BioGPT model according to the specified arguments, defining the model architecture. Instantiating a configuration
+    with the defaults will yield a similar configuration to that of the BioGPT
+    [microsoft/biogpt](https://huggingface.co/microsoft/biogpt) architecture.
+
+    Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
+    documentation from [`PretrainedConfig`] for more information.
+
+
+    Args:
+        vocab_size (`int`, *optional*, defaults to 42384):
+            Vocabulary size of the BioGPT model. Defines the number of different tokens that can be represented by the
+            `inputs_ids` passed when calling [`BioGptModel`].
+        hidden_size (`int`, *optional*, defaults to 1024):
+            Dimension of the encoder layers and the pooler layer.
+        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.
+        intermediate_size (`int`, *optional*, defaults to 4096):
+            Dimension of the "intermediate" (i.e., feed-forward) layer in the Transformer encoder.
+        hidden_act (`str` or `function`, *optional*, defaults to `"gelu"`):
+            The non-linear activation function (function or string) in the encoder and pooler. If string, `"gelu"`,
+            `"relu"`, `"selu"` and `"gelu_new"` are supported.
+        hidden_dropout_prob (`float`, *optional*, defaults to 0.1):
+            The dropout probabilitiy for all fully connected layers in the embeddings, encoder, and pooler.
+        attention_probs_dropout_prob (`float`, *optional*, defaults to 0.1):
+            The dropout ratio for the attention probabilities.
+        max_position_embeddings (`int`, *optional*, defaults to 1024):
+            The maximum sequence length that this model might ever be used with. Typically set this to something large
+            just in case (e.g., 512 or 1024 or 2048).
+        initializer_range (`float`, *optional*, defaults to 0.02):
+            The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
+        layer_norm_eps (`float`, *optional*, defaults to 1e-12):
+            The epsilon used by the layer normalization layers.
+        scale_embedding (`bool`, *optional*, defaults to `True`):
+            Scale embeddings by diving by sqrt(d_model).
+        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`.
+        is_encoder_decoder (`bool`, *optional*, defaults to `False`):
+            Whether this is an encoder/decoder model.
+        layerdrop (`float`, *optional*, defaults to 0.0):
+            Please refer to the paper about LayerDrop: https://arxiv.org/abs/1909.11556 for further details
+        activation_dropout (`float`, *optional*, defaults to 0.0):
+            The dropout ratio for activations inside the fully connected layer.
+        pad_token_id (`int`, *optional*, defaults to 1)
+            Padding token id.
+        bos_token_id (`int`, *optional*, defaults to 0)
+            Beginning of stream token id.
+        eos_token_id (`int`, *optional*, defaults to 2)
+            End of stream token id.
+        Example:
+
+    ```python
+    >>> from transformers import BioGptModel, BioGptConfig
+
+    >>> # Initializing a BioGPT microsoft/biogpt style configuration
+    >>> configuration = BioGptConfig()
+
+    >>> # Initializing a model from the microsoft/biogpt style configuration
+    >>> model = BioGptModel(configuration)
+
+    >>> # Accessing the model configuration
+    >>> configuration = model.config
+    ```"""
+    model_type = "biogpt"
+
+    def __init__(
+        self,
+        vocab_size=42384,
+        hidden_size=1024,
+        num_hidden_layers=24,
+        num_attention_heads=16,
+        intermediate_size=4096,
+        hidden_act="gelu",
+        hidden_dropout_prob=0.1,
+        attention_probs_dropout_prob=0.1,
+        max_position_embeddings=1024,
+        initializer_range=0.02,
+        layer_norm_eps=1e-12,
+        scale_embedding=True,
+        use_cache=True,
+        is_encoder_decoder=False,
+        layerdrop=0.0,
+        activation_dropout=0.0,
+        pad_token_id=1,
+        bos_token_id=0,
+        eos_token_id=2,
+        **kwargs
+    ):
+        self.vocab_size = vocab_size
+        self.max_position_embeddings = max_position_embeddings
+        self.hidden_size = hidden_size
+        self.num_hidden_layers = num_hidden_layers
+        self.num_attention_heads = num_attention_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.initializer_range = initializer_range
+        self.layer_norm_eps = layer_norm_eps
+        self.scale_embedding = scale_embedding
+        self.use_cache = use_cache
+        self.is_encoder_decoder = is_encoder_decoder
+        self.layerdrop = layerdrop
+        self.activation_dropout = activation_dropout
+        super().__init__(pad_token_id=pad_token_id, bos_token_id=bos_token_id, eos_token_id=eos_token_id, **kwargs)

src/transformers/models/biogpt/convert_biogpt_original_pytorch_checkpoint_to_pytorch.py

+# coding=utf-8
+# Copyright 2022 The HuggingFace Inc. team.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+
+import argparse
+import json
+import os
+import re
+import shutil
+
+import torch
+
+from transformers import BioGptConfig, BioGptForCausalLM
+from transformers.models.biogpt.tokenization_biogpt import VOCAB_FILES_NAMES
+from transformers.tokenization_utils_base import TOKENIZER_CONFIG_FILE
+from transformers.utils import WEIGHTS_NAME, logging
+
+
+logging.set_verbosity_warning()
+
+json_indent = 2
+
+
+# modified from https://github.com/facebookresearch/fairseq/blob/dd74992d0d143155998e9ed4076826bcea80fb06/fairseq/data/dictionary.py#L18
+class Dictionary:
+    """A mapping from symbols to consecutive integers"""
+
+    def __init__(
+        self,
+        *,  # begin keyword-only arguments
+        bos="<s>",
+        pad="<pad>",
+        eos="</s>",
+        unk="<unk>",
+        extra_special_symbols=None,
+    ):
+        self.bos_word, self.unk_word, self.pad_word, self.eos_word = bos, unk, pad, eos
+        self.symbols = []
+        self.count = []
+        self.indices = {}
+        self.bos_index = self.add_symbol(bos)
+        self.pad_index = self.add_symbol(pad)
+        self.eos_index = self.add_symbol(eos)
+        self.unk_index = self.add_symbol(unk)
+        if extra_special_symbols:
+            for s in extra_special_symbols:
+                self.add_symbol(s)
+        self.nspecial = len(self.symbols)
+
+    def __eq__(self, other):
+        return self.indices == other.indices
+
+    def __getitem__(self, idx):
+        if idx < len(self.symbols):
+            return self.symbols[idx]
+        return self.unk_word
+
+    def __len__(self):
+        """Returns the number of symbols in the dictionary"""
+        return len(self.symbols)
+
+    def __contains__(self, sym):
+        return sym in self.indices
+
+    @classmethod
+    def load(cls, f):
+        """Loads the dictionary from a text file with the format:
+
+        ```
+        <symbol0> <count0>
+        <symbol1> <count1>
+        ...
+        ```
+        """
+        d = cls()
+        d.add_from_file(f)
+        return d
+
+    def add_symbol(self, word, n=1, overwrite=False):
+        """Adds a word to the dictionary"""
+        if word in self.indices and not overwrite:
+            idx = self.indices[word]
+            self.count[idx] = self.count[idx] + n
+            return idx
+        else:
+            idx = len(self.symbols)
+            self.indices[word] = idx
+            self.symbols.append(word)
+            self.count.append(n)
+            return idx
+
+    def _load_meta(self, lines):
+        return 0
+
+    def add_from_file(self, f):
+        """
+        Loads a pre-existing dictionary from a text file and adds its symbols to this instance.
+        """
+        if isinstance(f, str):
+            try:
+                with open(f, "r", encoding="utf-8") as fd:
+                    self.add_from_file(fd)
+            except FileNotFoundError as fnfe:
+                raise fnfe
+            except UnicodeError:
+                raise Exception("Incorrect encoding detected in {}, please rebuild the dataset".format(f))
+            return
+
+        lines = f.readlines()
+        indices_start_line = self._load_meta(lines)
+
+        for line in lines[indices_start_line:]:
+            try:
+                line, field = line.rstrip().rsplit(" ", 1)
+                if field == "#fairseq:overwrite":
+                    overwrite = True
+                    line, field = line.rsplit(" ", 1)
+                else:
+                    overwrite = False
+                count = int(field)
+                word = line
+                if word in self and not overwrite:
+                    raise RuntimeError(
+                        "Duplicate word found when loading Dictionary: '{}'. "
+                        "Duplicate words can overwrite earlier ones by adding the "
+                        "#fairseq:overwrite flag at the end of the corresponding row "
+                        "in the dictionary file. If using the Camembert model, please "
+                        "download an updated copy of the model file.".format(word)
+                    )
+                self.add_symbol(word, n=count, overwrite=overwrite)
+            except ValueError:
+                raise ValueError("Incorrect dictionary format, expected '<token> <cnt> [flags]'")
+
+
+def rewrite_dict_keys(d):
+    # (1) remove word breaking symbol, (2) add word ending symbol where the word is not broken up,
+    # e.g.: d = {'le@@': 5, 'tt@@': 6, 'er': 7} => {'le': 5, 'tt': 6, 'er</w>': 7}
+    d2 = dict((re.sub(r"@@$", "", k), v) if k.endswith("@@") else (re.sub(r"$", "</w>", k), v) for k, v in d.items())
+    keep_keys = "<s> <pad> </s> <unk>".split()
+    # restore the special tokens
+    for k in keep_keys:
+        del d2[f"{k}</w>"]
+        d2[k] = d[k]  # restore
+    return d2
+
+
+def convert_biogpt_checkpoint_to_pytorch(biogpt_checkpoint_path, pytorch_dump_folder_path):
+
+    # prep
+    if not os.path.exists(biogpt_checkpoint_path):
+        raise ValueError(f"path {biogpt_checkpoint_path} does not exist!")
+    os.makedirs(pytorch_dump_folder_path, exist_ok=True)
+    print(f"Writing results to {pytorch_dump_folder_path}")
+
+    # handle various types of models
+
+    checkpoint_file = os.path.join(biogpt_checkpoint_path, "checkpoint.pt")
+    if not os.path.isfile(checkpoint_file):
+        raise ValueError(f"path to the file {checkpoint_file} does not exist!")
+    chkpt = torch.load(checkpoint_file, map_location="cpu")
+
+    args = chkpt["cfg"]["model"]
+
+    # dicts
+    dict_file = os.path.join(biogpt_checkpoint_path, "dict.txt")
+    if not os.path.isfile(dict_file):
+        raise ValueError(f"path to the file {dict_file} does not exist!")
+    src_dict = Dictionary.load(dict_file)
+    src_vocab = rewrite_dict_keys(src_dict.indices)
+    src_vocab_size = len(src_vocab)
+    src_vocab_file = os.path.join(pytorch_dump_folder_path, VOCAB_FILES_NAMES["vocab_file"])
+    print(f"Generating {src_vocab_file} of {src_vocab_size} records")
+    with open(src_vocab_file, "w", encoding="utf-8") as f:
+        f.write(json.dumps(src_vocab, ensure_ascii=False, indent=json_indent))
+
+    # merges_file (bpecodes)
+    bpecodes_file = os.path.join(biogpt_checkpoint_path, "bpecodes")
+    if not os.path.isfile(bpecodes_file):
+        raise ValueError(f"path to the file {bpecodes_file} does not exist!")
+
+    merges_file = os.path.join(pytorch_dump_folder_path, VOCAB_FILES_NAMES["merges_file"])
+    shutil.copyfile(bpecodes_file, merges_file)
+
+    # model config
+    biogpt_model_config_file = os.path.join(pytorch_dump_folder_path, "config.json")
+
+    model_conf = {
+        "activation_dropout": args["activation_dropout"],
+        "architectures": ["BioGptForCausalLM"],
+        "attention_probs_dropout_prob": args["attention_dropout"],
+        "bos_token_id": 0,
+        "eos_token_id": 2,
+        "hidden_act": args["activation_fn"],
+        "hidden_dropout_prob": args["dropout"],
+        "hidden_size": args["decoder_embed_dim"],
+        "initializer_range": 0.02,
+        "intermediate_size": args["decoder_ffn_embed_dim"],
+        "layer_norm_eps": 1e-12,
+        "layerdrop": args["decoder_layerdrop"],
+        "max_position_embeddings": args["max_target_positions"],
+        "model_type": "biogpt",
+        "num_attention_heads": args["decoder_attention_heads"],
+        "num_hidden_layers": args["decoder_layers"],
+        "pad_token_id": 1,
+        "scale_embedding": not args["no_scale_embedding"],
+        "tie_word_embeddings": args["share_decoder_input_output_embed"],
+        "vocab_size": src_vocab_size,
+    }
+
+    # good hparam defaults to start with
+
+    print(f"Generating {biogpt_model_config_file}")
+    with open(biogpt_model_config_file, "w", encoding="utf-8") as f:
+        f.write(json.dumps(model_conf, ensure_ascii=False, indent=json_indent))
+
+    # tokenizer config
+    biogpt_tokenizer_config_file = os.path.join(pytorch_dump_folder_path, TOKENIZER_CONFIG_FILE)
+
+    tokenizer_conf = {
+        "bos_token": "<s>",
+        "eos_token": "</s>",
+        "model_max_length": 1024,
+        "pad_token": "<pad>",
+        "special_tokens_map_file": None,
+        "tokenizer_class": "BioGptTokenizer",
+        "unk_token": "<unk>",
+    }
+
+    print(f"Generating {biogpt_tokenizer_config_file}")
+    with open(biogpt_tokenizer_config_file, "w", encoding="utf-8") as f:
+        f.write(json.dumps(tokenizer_conf, ensure_ascii=False, indent=json_indent))
+
+    # model
+    model_state_dict = chkpt["model"]
+
+    # remove unneeded keys
+    ignore_keys = [
+        "decoder.version",
+    ]
+    for k in ignore_keys:
+        model_state_dict.pop(k, None)
+
+    layer_names = list(model_state_dict.keys())
+    for layer_name in layer_names:
+        if layer_name.endswith("output_projection.weight"):
+            model_state_dict[layer_name.replace("decoder.", "")] = model_state_dict.pop(layer_name)
+        else:
+            model_state_dict[layer_name.replace("decoder", "biogpt")] = model_state_dict.pop(layer_name)
+
+    config = BioGptConfig.from_pretrained(pytorch_dump_folder_path)
+    model_new = BioGptForCausalLM(config)
+
+    # check that it loads ok
+    model_new.load_state_dict(model_state_dict)
+
+    # save
+    pytorch_weights_dump_path = os.path.join(pytorch_dump_folder_path, WEIGHTS_NAME)
+    print(f"Generating {pytorch_weights_dump_path}")
+    torch.save(model_state_dict, pytorch_weights_dump_path)
+
+    print("Conversion is done!")
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    # Required parameters
+    parser.add_argument(
+        "--biogpt_checkpoint_path",
+        default=None,
+        type=str,
+        required=True,
+        help=(
+            "Path to the official PyTorch checkpoint file which is expected to reside in the dump dir with dicts,"
+            " bpecodes, etc."
+        ),
+    )
+    parser.add_argument(
+        "--pytorch_dump_folder_path", default=None, type=str, required=True, help="Path to the output PyTorch model."
+    )
+    args = parser.parse_args()
+    convert_biogpt_checkpoint_to_pytorch(args.biogpt_checkpoint_path, args.pytorch_dump_folder_path)

src/transformers/models/biogpt/tokenization_biogpt.py

+# coding=utf-8
+# Copyright 2022 The HuggingFace Team and Microsoft Research AI4Science. 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.
+"""Tokenization classes for BioGPT."""
+import json
+import os
+from typing import List, Optional, Tuple
+
+from ...tokenization_utils import PreTrainedTokenizer
+from ...utils import logging
+
+
+logger = logging.get_logger(__name__)
+
+VOCAB_FILES_NAMES = {
+    "vocab_file": "vocab.json",
+    "merges_file": "merges.txt",
+}
+
+PRETRAINED_VOCAB_FILES_MAP = {
+    "vocab_file": {
+        "microsoft/biogpt": "https://huggingface.co/microsoft/biogpt/resolve/main/vocab.json",
+    },
+    "merges_file": {"microsoft/biogpt": "https://huggingface.co/microsoft/biogpt/resolve/main/merges.txt"},
+}
+
+PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES = {
+    "microsoft/biogpt": 1024,
+}
+
+
+def get_pairs(word):
+    """
+    Return set of symbol pairs in a word. word is represented as tuple of symbols (symbols being variable-length
+    strings)
+    """
+    pairs = set()
+    prev_char = word[0]
+    for char in word[1:]:
+        pairs.add((prev_char, char))
+        prev_char = char
+    return pairs
+
+
+class BioGptTokenizer(PreTrainedTokenizer):
+    """
+    Construct an FAIRSEQ Transformer tokenizer. Moses tokenization followed by Byte-Pair Encoding.
+
+    This tokenizer inherits from [`PreTrainedTokenizer`] which contains most of the main methods. Users should refer to
+    this superclass for more information regarding those methods.
+
+    Args:
+        vocab_file (`str`):
+            Path to the vocabulary file.
+        merges_file (`str`):
+            Merges file.
+        unk_token (`str`, *optional*, defaults to `"<unk>"`):
+            The unknown token. A token that is not in the vocabulary cannot be converted to an ID and is set to be this
+            token instead.
+        bos_token (`str`, *optional*, defaults to `"<s>"`):
+            The beginning of sequence token that was used during pretraining. Can be used a sequence classifier token.
+
+            <Tip>
+
+            When building a sequence using special tokens, this is not the token that is used for the beginning of
+            sequence. The token used is the `cls_token`.
+
+            </Tip>
+
+        eos_token (`str`, *optional*, defaults to `"</s>"`):
+            The end of sequence token.
+
+            <Tip>
+
+            When building a sequence using special tokens, this is not the token that is used for the end of sequence.
+            The token used is the `sep_token`.
+
+            </Tip>
+
+        sep_token (`str`, *optional*, defaults to `"</s>"`):
+            The separator token, which is used when building a sequence from multiple sequences, e.g. two sequences for
+            sequence classification or for a text and a question for question answering. It is also used as the last
+            token of a sequence built with special tokens.
+        pad_token (`str`, *optional*, defaults to `"<pad>"`):
+            The token used for padding, for example when batching sequences of different lengths.
+    """
+
+    vocab_files_names = VOCAB_FILES_NAMES
+    pretrained_vocab_files_map = PRETRAINED_VOCAB_FILES_MAP
+    max_model_input_sizes = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES
+    model_input_names = ["input_ids", "attention_mask"]
+
+    def __init__(
+        self,
+        vocab_file,
+        merges_file,
+        unk_token="<unk>",
+        bos_token="<s>",
+        eos_token="</s>",
+        sep_token="</s>",
+        pad_token="<pad>",
+        **kwargs
+    ):
+        super().__init__(
+            bos_token=bos_token,
+            eos_token=eos_token,
+            sep_token=sep_token,
+            unk_token=unk_token,
+            pad_token=pad_token,
+            **kwargs,
+        )
+
+        try:
+            import sacremoses
+        except ImportError:
+            raise ImportError(
+                "You need to install sacremoses to use BioGptTokenizer. "
+                "See https://pypi.org/project/sacremoses/ for installation."
+            )
+
+        self.lang = "en"
+        self.sm = sacremoses
+        # cache of sm.MosesTokenizer instance
+        self.cache_moses_tokenizer = dict()
+        self.cache_moses_detokenizer = dict()
+
+        """ Initialisation"""
+        with open(vocab_file, encoding="utf-8") as vocab_handle:
+            self.encoder = json.load(vocab_handle)
+        self.decoder = {v: k for k, v in self.encoder.items()}
+        with open(merges_file, encoding="utf-8") as merges_handle:
+            merges = merges_handle.read().split("\n")[:-1]
+        merges = [tuple(merge.split()[:2]) for merge in merges]
+        self.bpe_ranks = dict(zip(merges, range(len(merges))))
+        self.cache = {}
+
+    @property
+    def vocab_size(self):
+        """Returns vocab size"""
+        return len(self.encoder)
+
+    def get_vocab(self):
+        return dict(self.encoder, **self.added_tokens_encoder)
+
+    def moses_tokenize(self, text, lang):
+        if lang not in self.cache_moses_tokenizer:
+            moses_tokenizer = self.sm.MosesTokenizer(lang=lang)
+            self.cache_moses_tokenizer[lang] = moses_tokenizer
+        return self.cache_moses_tokenizer[lang].tokenize(
+            text, aggressive_dash_splits=True, return_str=False, escape=True
+        )
+
+    def moses_detokenize(self, tokens, lang):
+        if lang not in self.cache_moses_detokenizer:
+            moses_detokenizer = self.sm.MosesDetokenizer(lang=lang)
+            self.cache_moses_detokenizer[lang] = moses_detokenizer
+        return self.cache_moses_detokenizer[lang].detokenize(tokens)
+
+    def bpe(self, token):
+        word = tuple(token[:-1]) + (token[-1] + "</w>",)
+        if token in self.cache:
+            return self.cache[token]
+        pairs = get_pairs(word)
+
+        if not pairs:
+            return token + "</w>"
+
+        while True:
+            bigram = min(pairs, key=lambda pair: self.bpe_ranks.get(pair, float("inf")))
+            if bigram not in self.bpe_ranks:
+                break
+            first, second = bigram
+            new_word = []
+            i = 0
+            while i < len(word):
+                try:
+                    j = word.index(first, i)
+                except ValueError:
+                    new_word.extend(word[i:])
+                    break
+                else:
+                    new_word.extend(word[i:j])
+                    i = j
+
+                if word[i] == first and i < len(word) - 1 and word[i + 1] == second:
+                    new_word.append(first + second)
+                    i += 2
+                else:
+                    new_word.append(word[i])
+                    i += 1
+            new_word = tuple(new_word)
+            word = new_word
+            if len(word) == 1:
+                break
+            else:
+                pairs = get_pairs(word)
+        word = " ".join(word)
+        if word == "\n  </w>":
+            word = "\n</w>"
+        self.cache[token] = word
+        return word
+
+    def _tokenize(self, text, bypass_tokenizer=False):
+        """Returns a tokenized string."""
+        if bypass_tokenizer:
+            text = text.split()
+        else:
+            text = self.moses_tokenize(text, self.lang)
+
+        split_tokens = []
+        for token in text:
+            if token:
+                split_tokens.extend([t for t in self.bpe(token).split(" ")])
+
+        return split_tokens
+
+    def _convert_token_to_id(self, token):
+        """Converts a token (str) in an id using the vocab."""
+        return self.encoder.get(token, self.encoder.get(self.unk_token))
+
+    def _convert_id_to_token(self, index):
+        """Converts an index (integer) in a token (str) using the vocab."""
+        return self.decoder.get(index, self.unk_token)
+
+    def convert_tokens_to_string(self, tokens):
+        """Converts a sequence of tokens (string) in a single string."""
+        # remove BPE
+        tokens = [t.replace(" ", "").replace("</w>", " ") for t in tokens]
+        tokens = "".join(tokens).split()
+        # detokenize
+        text = self.moses_detokenize(tokens, self.lang)
+        return text
+
+    def build_inputs_with_special_tokens(
+        self, token_ids_0: List[int], token_ids_1: Optional[List[int]] = None
+    ) -> List[int]:
+        """
+        Build model inputs from a sequence or a pair of sequence for sequence classification tasks by concatenating and
+        adding special tokens. A BioGPT sequence has the following format:
+
+        - single sequence: `</s> X `
+        - pair of sequences: `</s> A </s> B `
+
+        Args:
+            token_ids_0 (`List[int]`):
+                List of IDs to which the special tokens will be added.
+            token_ids_1 (`List[int]`, *optional*):
+                Optional second list of IDs for sequence pairs.
+
+        Returns:
+            `List[int]`: List of [input IDs](../glossary#input-ids) with the appropriate special tokens.
+        """
+        if token_ids_1 is None:
+            return [self.sep_token_id] + token_ids_0
+        sep = [self.sep_token_id]
+        return sep + token_ids_0 + sep + token_ids_1
+
+    def get_special_tokens_mask(
+        self, token_ids_0: List[int], token_ids_1: Optional[List[int]] = None, already_has_special_tokens: bool = False
+    ) -> List[int]:
+        """
+        Retrieve sequence ids from a token list that has no special tokens added. This method is called when adding
+        special tokens using the tokenizer `prepare_for_model` method.
+
+        Args:
+            token_ids_0 (`List[int]`):
+                List of IDs.
+            token_ids_1 (`List[int]`, *optional*):
+                Optional second list of IDs for sequence pairs.
+            already_has_special_tokens (`bool`, *optional*, defaults to `False`):
+                Whether or not the token list is already formatted with special tokens for the model.
+
+        Returns:
+            `List[int]`: A list of integers in the range [0, 1]: 1 for a special token, 0 for a sequence token.
+        """
+        if already_has_special_tokens:
+            return super().get_special_tokens_mask(
+                token_ids_0=token_ids_0, token_ids_1=token_ids_1, already_has_special_tokens=True
+            )
+        # no bos used in fairseq
+        if token_ids_1 is not None:
+            return [1] + ([0] * len(token_ids_0)) + [1] + ([0] * len(token_ids_1))
+        return [1] + ([0] * len(token_ids_0))
+
+    def create_token_type_ids_from_sequences(
+        self, token_ids_0: List[int], token_ids_1: Optional[List[int]] = None
+    ) -> List[int]:
+        """
+        Create a mask from the two sequences passed to be used in a sequence-pair classification task. A FAIRSEQ
+        Transformer sequence pair mask has the following format:
+
+        ```
+        0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1
+        | first sequence    | second sequence |
+        ```
+
+        If `token_ids_1` is `None`, this method only returns the first portion of the mask (0s).
+
+        Args:
+            token_ids_0 (`List[int]`):
+                List of IDs.
+            token_ids_1 (`List[int]`, *optional*):
+                Optional second list of IDs for sequence pairs.
+
+        Returns:
+            `List[int]`: List of [token type IDs](../glossary#token-type-ids) according to the given sequence(s).
+        """
+        sep = [self.sep_token_id]
+
+        # no bos used in fairseq
+        if token_ids_1 is None:
+            return len(token_ids_0 + sep) * [0]
+        return len(token_ids_0 + sep) * [0] + len(token_ids_1 + sep) * [1]
+
+    def save_vocabulary(self, save_directory: str, filename_prefix: Optional[str] = None) -> Tuple[str]:
+        if not os.path.isdir(save_directory):
+            logger.error(f"Vocabulary path ({save_directory}) should be a directory")
+            return
+        vocab_file = os.path.join(
+            save_directory, (filename_prefix + "-" if filename_prefix else "") + VOCAB_FILES_NAMES["vocab_file"]
+        )
+        merge_file = os.path.join(
+            save_directory, (filename_prefix + "-" if filename_prefix else "") + VOCAB_FILES_NAMES["merges_file"]
+        )
+
+        with open(vocab_file, "w", encoding="utf-8") as f:
+            f.write(json.dumps(self.encoder, indent=2, sort_keys=True, ensure_ascii=False) + "\n")
+
+        index = 0
+        with open(merge_file, "w", encoding="utf-8") as writer:
+            for bpe_tokens, token_index in sorted(self.bpe_ranks.items(), key=lambda kv: kv[1]):
+                if index != token_index:
+                    logger.warning(
+                        f"Saving vocabulary to {merge_file}: BPE merge indices are not consecutive."
+                        " Please check that the tokenizer is not corrupted!"
+                    )
+                    index = token_index
+                writer.write(" ".join(bpe_tokens) + "\n")
+                index += 1
+
+        return vocab_file, merge_file
+
+    def __getstate__(self):
+        state = self.__dict__.copy()
+        state["sm"] = None
+        return state
+
+    def __setstate__(self, d):
+        self.__dict__ = d
+
+        try:
+            import sacremoses
+        except ImportError:
+            raise ImportError(
+                "You need to install sacremoses to use XLMTokenizer. "
+                "See https://pypi.org/project/sacremoses/ for installation."
+            )
+
+        self.sm = sacremoses

src/transformers/utils/dummy_pt_objects.py

@@ -995,6 +995,30 @@ class BigBirdPegasusPreTrainedModel(metaclass=DummyObject):
         requires_backends(self, ["torch"])
 
 
+BIOGPT_PRETRAINED_MODEL_ARCHIVE_LIST = None
+
+
+class BioGptForCausalLM(metaclass=DummyObject):
+    _backends = ["torch"]
+
+    def __init__(self, *args, **kwargs):
+        requires_backends(self, ["torch"])
+
+
+class BioGptModel(metaclass=DummyObject):
+    _backends = ["torch"]
+
+    def __init__(self, *args, **kwargs):
+        requires_backends(self, ["torch"])
+
+
+class BioGptPreTrainedModel(metaclass=DummyObject):
+    _backends = ["torch"]
+
+    def __init__(self, *args, **kwargs):
+        requires_backends(self, ["torch"])
+
+
 BLENDERBOT_PRETRAINED_MODEL_ARCHIVE_LIST = None