Test model parallelization

Note: parellelism not yet introduced into t5. Just doing this to get past checks.

.github/PULL_REQUEST_TEMPLATE.md

@@ -37,25 +37,27 @@ members/contributors which may be interested in your PR.
  If you know how to use git blame, that is the easiest way, otherwise, here is a rough guide of **who to tag**.
  Please tag fewer than 3 people.
 
- albert, bert, GPT2, XLM: @LysandreJik 
+ albert, bert, XLM: @LysandreJik 
+ GPT2: @LysandreJik, @patrickvonplaten
  tokenizers: @mfuntowicz
  Trainer: @sgugger
- Speed and Memory Benchmarks: @patrickvonplaten
+ Benchmarks: @patrickvonplaten
  Model Cards: @julien-c
  Translation: @sshleifer
  Summarization: @sshleifer
- TextGeneration: @TevenLeScao 
  examples/distillation: @VictorSanh
  nlp datasets: [different repo](https://github.com/huggingface/nlp)
  rust tokenizers: [different repo](https://github.com/huggingface/tokenizers)
- Text Generation: @TevenLeScao
+ Text Generation: @patrickvonplaten, @TevenLeScao
  Blenderbot, Bart, Marian, Pegasus: @sshleifer
  T5: @patrickvonplaten
- Longformer/Reformer: @patrickvonplaten
- TransfoXL/XLNet: @TevenLeScao 
+ Rag: @patrickvonplaten, @lhoestq
+ EncoderDecoder: @patrickvonplaten
+ Longformer, Reformer: @patrickvonplaten
+ TransfoXL, XLNet: @TevenLeScao, @patrickvonplaten
  examples/seq2seq: @sshleifer
  examples/bert-loses-patience: @JetRunner
  tensorflow: @jplu
  examples/token-classification: @stefan-it
  documentation: @sgugger
- -->
+ -->

.github/workflows/github-torch-hub.yml

@@ -30,7 +30,8 @@ jobs:
       run: |
         pip install --upgrade pip
         pip install torch
-        pip install numpy tokenizers filelock requests tqdm regex sentencepiece sacremoses packaging
+        pip install numpy filelock protobuf requests tqdm regex sentencepiece sacremoses packaging
+        pip install tokenizers==0.9.0.rc2
 
     - name: Torch hub list
       run: |

.gitignore

@@ -12,6 +12,7 @@ __pycache__/
 tests/fixtures
 logs/
 lightning_logs/
+lang_code_data/
 
 # Distribution / packaging
 .Python

Makefile

@@ -5,18 +5,18 @@ check_dirs := examples templates tests src utils
 
 # get modified files since the branch was made
 fork_point_sha := $(shell git merge-base --fork-point master)
-joined_dirs    := $(shell echo $(check_dirs) | tr " " "|")
-modified_files := $(shell git diff --name-only $(fork_point_sha) | egrep '^($(joined_dirs))')
-#$(info modified files are: $(modified_files))
+joined_dirs := $(shell echo $(check_dirs) | tr " " "|")
+modified_py_files := $(shell git diff --name-only $(fork_point_sha) | egrep '^($(joined_dirs))' | egrep '\.py$$')
+#$(info modified files are: $(modified_py_files))
 
 modified_only_fixup:
-	@if [ -n "$(modified_files)" ]; then \
-		echo "Checking/fixing $(modified_files)"; \
-		black $(modified_files); \
-		isort $(modified_files); \
-		flake8 $(modified_files); \
+	@if [ -n "$(modified_py_files)" ]; then \
+		echo "Checking/fixing $(modified_py_files)"; \
+		black $(modified_py_files); \
+		isort $(modified_py_files); \
+		flake8 $(modified_py_files); \
 	else \
-		echo "No relevant files were modified"; \
+		echo "No library .py files were modified"; \
 	fi
 
 # Check that source code meets quality standards

README.md

@@ -161,31 +161,31 @@ If you'd like to play with the examples, you must [install the library from sour
 1. **[ALBERT](https://huggingface.co/transformers/model_doc/albert.html)** (from Google Research and the Toyota Technological Institute at Chicago) released with the paper [ALBERT: A Lite BERT for Self-supervised Learning of Language Representations](https://arxiv.org/abs/1909.11942), by Zhenzhong Lan, Mingda Chen, Sebastian Goodman, Kevin Gimpel, Piyush Sharma, Radu Soricut.
 1. **[BART](https://huggingface.co/transformers/model_doc/bart.html)** (from Facebook) released with the paper [BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension](https://arxiv.org/pdf/1910.13461.pdf) by Mike Lewis, Yinhan Liu, Naman Goyal, Marjan Ghazvininejad, Abdelrahman Mohamed, Omer Levy, Ves Stoyanov and Luke Zettlemoyer.
 1. **[BERT](https://huggingface.co/transformers/model_doc/bert.html)** (from Google) released with the paper [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://arxiv.org/abs/1810.04805) by Jacob Devlin, Ming-Wei Chang, Kenton Lee and Kristina Toutanova.
-1. **[BERT For Sequence Generation](https://tfhub.dev/s?module-type=text-generation&subtype=module,placeholder)** (from Google) released with the paper [Leveraging Pre-trained Checkpoints for Sequence Generation Tasks](https://arxiv.org/abs/1907.12461) by Sascha Rothe, Shashi Narayan, Aliaksei Severyn.
+1. **[BERT For Sequence Generation](https://huggingface.co/transformers/model_doc/bertgeneration.html)** (from Google) released with the paper [Leveraging Pre-trained Checkpoints for Sequence Generation Tasks](https://arxiv.org/abs/1907.12461) by Sascha Rothe, Shashi Narayan, Aliaksei Severyn.
+1. **[Blenderbot](https://huggingface.co/transformers/master/model_doc/blenderbot.html)** (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. **[CamemBERT](https://huggingface.co/transformers/model_doc/camembert.html)** (from Inria/Facebook/Sorbonne) released with the paper [CamemBERT: a Tasty French Language Model](https://arxiv.org/abs/1911.03894) by Louis Martin*, Benjamin Muller*, Pedro Javier Ortiz Suárez*, Yoann Dupont, Laurent Romary, Éric Villemonte de la Clergerie, Djamé Seddah and Benoît Sagot.
 1. **[CTRL](https://huggingface.co/transformers/model_doc/ctrl.html)** (from Salesforce) released with the paper [CTRL: A Conditional Transformer Language Model for Controllable Generation](https://arxiv.org/abs/1909.05858) by Nitish Shirish Keskar*, Bryan McCann*, Lav R. Varshney, Caiming Xiong and Richard Socher.
-1. **[DeBERTa](https://huggingface.co/transformers/model_doc/deberta.html)** (from Microsoft Research) released with the paper [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen.
+1. **[DeBERTa](https://huggingface.co/transformers/master/model_doc/deberta.html)** (from Microsoft Research) released with the paper [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen.
 1. **[DialoGPT](https://huggingface.co/transformers/model_doc/dialogpt.html)** (from Microsoft Research) released with the paper [DialoGPT: Large-Scale Generative Pre-training for Conversational Response Generation](https://arxiv.org/abs/1911.00536) by Yizhe Zhang, Siqi Sun, Michel Galley, Yen-Chun Chen, Chris Brockett, Xiang Gao, Jianfeng Gao, Jingjing Liu, Bill Dolan.
 1. **[DistilBERT](https://huggingface.co/transformers/model_doc/distilbert.html)** (from HuggingFace), released together with the paper [DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter](https://arxiv.org/abs/1910.01108) by Victor Sanh, Lysandre Debut and Thomas Wolf. The same method has been applied to compress GPT2 into [DistilGPT2](https://github.com/huggingface/transformers/tree/master/examples/distillation), RoBERTa into [DistilRoBERTa](https://github.com/huggingface/transformers/tree/master/examples/distillation), Multilingual BERT into [DistilmBERT](https://github.com/huggingface/transformers/tree/master/examples/distillation) and a German version of DistilBERT.
-1. **[DPR](https://github.com/facebookresearch/DPR)** (from Facebook) released with the paper [Dense Passage Retrieval
+1. **[DPR](https://huggingface.co/transformers/model_doc/dpr.html)** (from Facebook) released with the paper [Dense Passage Retrieval
 for Open-Domain Question Answering](https://arxiv.org/abs/2004.04906) by Vladimir Karpukhin, Barlas Oğuz, Sewon
 Min, Patrick Lewis, Ledell Wu, Sergey Edunov, Danqi Chen, and Wen-tau Yih.
 1. **[ELECTRA](https://huggingface.co/transformers/model_doc/electra.html)** (from Google Research/Stanford University) released with the paper [ELECTRA: Pre-training text encoders as discriminators rather than generators](https://arxiv.org/abs/2003.10555) by Kevin Clark, Minh-Thang Luong, Quoc V. Le, Christopher D. Manning.
 1. **[FlauBERT](https://huggingface.co/transformers/model_doc/flaubert.html)** (from CNRS) released with the paper [FlauBERT: Unsupervised Language Model Pre-training for French](https://arxiv.org/abs/1912.05372) by Hang Le, Loïc Vial, Jibril Frej, Vincent Segonne, Maximin Coavoux, Benjamin Lecouteux, Alexandre Allauzen, Benoît Crabbé, Laurent Besacier, Didier Schwab.
-1. **[Funnel Transformer](https://github.com/laiguokun/Funnel-Transformer)** (from CMU/Google Brain) released with the paper [Funnel-Transformer: Filtering out Sequential Redundancy for Efficient Language Processing](https://arxiv.org/abs/2006.03236) by Zihang Dai, Guokun Lai, Yiming Yang, Quoc V. Le.
+1. **[Funnel Transformer](https://huggingface.co/transformers/model_doc/funnel.html)** (from CMU/Google Brain) released with the paper [Funnel-Transformer: Filtering out Sequential Redundancy for Efficient Language Processing](https://arxiv.org/abs/2006.03236) by Zihang Dai, Guokun Lai, Yiming Yang, Quoc V. Le.
 1. **[GPT](https://huggingface.co/transformers/model_doc/gpt.html)** (from OpenAI) released with the paper [Improving Language Understanding by Generative Pre-Training](https://blog.openai.com/language-unsupervised/) by Alec Radford, Karthik Narasimhan, Tim Salimans and Ilya Sutskever.
 1. **[GPT-2](https://huggingface.co/transformers/model_doc/gpt2.html)** (from OpenAI) released with the paper [Language Models are Unsupervised Multitask Learners](https://blog.openai.com/better-language-models/) by Alec Radford*, Jeffrey Wu*, Rewon Child, David Luan, Dario Amodei** and Ilya Sutskever**.
-1. **[LayoutLM](https://github.com/microsoft/unilm/tree/master/layoutlm)** (from Microsoft Research Asia) released with the paper [LayoutLM: Pre-training of Text and Layout for Document Image Understanding](https://arxiv.org/abs/1912.13318) by Yiheng Xu, Minghao Li, Lei Cui, Shaohan Huang, Furu Wei, Ming Zhou.
+1. **[LayoutLM](https://huggingface.co/transformers/model_doc/layoutlm.html)** (from Microsoft Research Asia) released with the paper [LayoutLM: Pre-training of Text and Layout for Document Image Understanding](https://arxiv.org/abs/1912.13318) by Yiheng Xu, Minghao Li, Lei Cui, Shaohan Huang, Furu Wei, Ming Zhou.
 1. **[Longformer](https://huggingface.co/transformers/model_doc/longformer.html)** (from AllenAI) released with the paper [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) by Iz Beltagy, Matthew E. Peters, Arman Cohan.
-1. **[LXMERT](https://github.com/airsplay/lxmert)** (from UNC Chapel Hill) released with the paper [LXMERT: Learning Cross-Modality Encoder Representations from Transformers for Open-Domain Question Answering](https://arxiv.org/abs/1908.07490) by Hao Tan and Mohit Bansal.
+1. **[LXMERT](https://huggingface.co/transformers/model_doc/lxmert.html)** (from UNC Chapel Hill) released with the paper [LXMERT: Learning Cross-Modality Encoder Representations from Transformers for Open-Domain Question Answering](https://arxiv.org/abs/1908.07490) by Hao Tan and Mohit Bansal.
 1. **[MarianMT](https://huggingface.co/transformers/model_doc/marian.html)** Machine translation models trained using [OPUS](http://opus.nlpl.eu/) data by Jörg Tiedemann. The [Marian Framework](https://marian-nmt.github.io/) is being developed by the Microsoft Translator Team.
-1. **[MBart](https://github.com/pytorch/fairseq/tree/master/examples/mbart)** (from Facebook) released with the paper  [Multilingual Denoising Pre-training for Neural Machine Translation](https://arxiv.org/abs/2001.08210) by Yinhan Liu, Jiatao Gu, Naman Goyal, Xian Li, Sergey Edunov, Marjan Ghazvininejad, Mike Lewis, Luke Zettlemoyer.
-1. **[MMBT](https://github.com/facebookresearch/mmbt/)** (from Facebook), released together with the paper a [Supervised Multimodal Bitransformers for Classifying Images and Text](https://arxiv.org/pdf/1909.02950.pdf) by Douwe Kiela, Suvrat Bhooshan, Hamed Firooz, Davide Testuggine.
-1. **[Pegasus](https://github.com/google-research/pegasus)** (from Google) released with the paper [PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization](https://arxiv.org/abs/1912.08777)> by Jingqing Zhang, Yao Zhao, Mohammad Saleh and Peter J. Liu.
+1. **[MBart](https://huggingface.co/transformers/model_doc/mbart.html)** (from Facebook) released with the paper  [Multilingual Denoising Pre-training for Neural Machine Translation](https://arxiv.org/abs/2001.08210) by Yinhan Liu, Jiatao Gu, Naman Goyal, Xian Li, Sergey Edunov, Marjan Ghazvininejad, Mike Lewis, Luke Zettlemoyer.
+1. **[Pegasus](https://huggingface.co/transformers/model_doc/pegasus.html)** (from Google) released with the paper [PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization](https://arxiv.org/abs/1912.08777)> by Jingqing Zhang, Yao Zhao, Mohammad Saleh and Peter J. Liu.
 1. **[Reformer](https://huggingface.co/transformers/model_doc/reformer.html)** (from Google Research) released with the paper [Reformer: The Efficient Transformer](https://arxiv.org/abs/2001.04451) by Nikita Kitaev, Łukasz Kaiser, Anselm Levskaya.
 1. **[RoBERTa](https://huggingface.co/transformers/model_doc/roberta.html)** (from Facebook), released together with the paper a [Robustly Optimized BERT Pretraining Approach](https://arxiv.org/abs/1907.11692) by Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, Veselin Stoyanov.
 ultilingual BERT into [DistilmBERT](https://github.com/huggingface/transformers/tree/master/examples/distillation) and a German version of DistilBERT.
-1. **[SqueezeBert](https://huggingface.co/transformers/model_doc/squeezebert.html)** released with the paper [SqueezeBERT: What can computer vision teach NLP about efficient neural networks?](https://arxiv.org/abs/2006.11316) by Forrest N. Iandola, Albert E. Shaw, Ravi Krishna, and Kurt W. Keutzer.
+1. **[SqueezeBert](https://huggingface.co/transformers/master/model_doc/squeezebert.html)** released with the paper [SqueezeBERT: What can computer vision teach NLP about efficient neural networks?](https://arxiv.org/abs/2006.11316) by Forrest N. Iandola, Albert E. Shaw, Ravi Krishna, and Kurt W. Keutzer.
 1. **[T5](https://huggingface.co/transformers/model_doc/t5.html)** (from Google AI) released with the paper [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://arxiv.org/abs/1910.10683) by Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu.
 1. **[Transformer-XL](https://huggingface.co/transformers/model_doc/transformerxl.html)** (from Google/CMU) released with the paper [Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context](https://arxiv.org/abs/1901.02860) by Zihang Dai*, Zhilin Yang*, Yiming Yang, Jaime Carbonell, Quoc V. Le, Ruslan Salakhutdinov.
 1. **[XLM](https://huggingface.co/transformers/model_doc/xlm.html)** (from Facebook) released together with the paper [Cross-lingual Language Model Pretraining](https://arxiv.org/abs/1901.07291) by Guillaume Lample and Alexis Conneau.

docker/transformers-gpu/Dockerfile

@@ -1,4 +1,4 @@
-FROM nvidia/cuda:10.1-cudnn7-runtime-ubuntu18.04
+FROM nvidia/cuda:10.2-cudnn7-devel-ubuntu18.04
 LABEL maintainer="Hugging Face"
 LABEL repository="transformers"
 
@@ -18,9 +18,14 @@ RUN python3 -m pip install --no-cache-dir --upgrade pip && \
     tensorflow \
     torch
 
+RUN git clone https://github.com/NVIDIA/apex
+RUN cd apex && \
+    python3 setup.py install && \
+    pip install -v --no-cache-dir --global-option="--cpp_ext" --global-option="--cuda_ext" ./
+
 WORKDIR /workspace
 COPY . transformers/
 RUN cd transformers/ && \
     python3 -m pip install --no-cache-dir .
 
-CMD ["/bin/bash"]
+CMD ["/bin/bash"]

docker/transformers-pytorch-gpu/Dockerfile

@@ -1,4 +1,4 @@
-FROM nvidia/cuda:10.1-cudnn7-runtime-ubuntu18.04
+FROM nvidia/cuda:10.2-cudnn7-devel-ubuntu18.04
 LABEL maintainer="Hugging Face"
 LABEL repository="transformers"
 
@@ -17,9 +17,14 @@ RUN python3 -m pip install --no-cache-dir --upgrade pip && \
     mkl \
     torch
 
+RUN git clone https://github.com/NVIDIA/apex
+RUN cd apex && \
+    python3 setup.py install && \
+    pip install -v --no-cache-dir --global-option="--cpp_ext" --global-option="--cuda_ext" ./
+
 WORKDIR /workspace
 COPY . transformers/
 RUN cd transformers/ && \
     python3 -m pip install --no-cache-dir .
 
-CMD ["/bin/bash"]
+CMD ["/bin/bash"]

docs/source/index.rst

@@ -57,112 +57,103 @@ conversion utilities for the following models:
 ..
     This list is updated automatically from the README with `make fix-copies`. Do not update manually!
 
-1. `ALBERT <https://huggingface.co/transformers/model_doc/albert.html>`__ (from Google Research and the Toyota
-   Technological Institute at Chicago) released with the paper `ALBERT: A Lite BERT for Self-supervised Learning of
-   Language Representations <https://arxiv.org/abs/1909.11942>`__, by Zhenzhong Lan, Mingda Chen, Sebastian Goodman,
-   Kevin Gimpel, Piyush Sharma, Radu Soricut.
-2. `BART <https://huggingface.co/transformers/model_doc/bart.html>`__ (from Facebook) released with the paper `BART:
-   Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension
+1. :doc:`ALBERT <model_doc/albert>` (from Google Research and the Toyota Technological Institute at Chicago) released
+   with the paper `ALBERT: A Lite BERT for Self-supervised Learning of Language Representations
+   <https://arxiv.org/abs/1909.11942>`__, by Zhenzhong Lan, Mingda Chen, Sebastian Goodman, Kevin Gimpel, Piyush
+   Sharma, Radu Soricut.
+2. :doc:`BART <model_doc/bart>` (from Facebook) released with the paper `BART: Denoising Sequence-to-Sequence
+   Pre-training for Natural Language Generation, Translation, and Comprehension
    <https://arxiv.org/pdf/1910.13461.pdf>`__ by Mike Lewis, Yinhan Liu, Naman Goyal, Marjan Ghazvininejad, Abdelrahman
    Mohamed, Omer Levy, Ves Stoyanov and Luke Zettlemoyer.
-3. `BERT <https://huggingface.co/transformers/model_doc/bert.html>`__ (from Google) released with the paper `BERT:
-   Pre-training of Deep Bidirectional Transformers for Language Understanding <https://arxiv.org/abs/1810.04805>`__ by
-   Jacob Devlin, Ming-Wei Chang, Kenton Lee and Kristina Toutanova.
-4. `BERT For Sequence Generation <https://tfhub.dev/s?module-type=text-generation&subtype=module,placeholder>`__ (from
-   Google) released with the paper `Leveraging Pre-trained Checkpoints for Sequence Generation Tasks
-   <https://arxiv.org/abs/1907.12461>`__ by Sascha Rothe, Shashi Narayan, Aliaksei Severyn.
-5. `CamemBERT <https://huggingface.co/transformers/model_doc/camembert.html>`__ (from Inria/Facebook/Sorbonne) released
-   with the paper `CamemBERT: a Tasty French Language Model <https://arxiv.org/abs/1911.03894>`__ by Louis Martin*,
-   Benjamin Muller*, Pedro Javier Ortiz Suárez*, Yoann Dupont, Laurent Romary, Éric Villemonte de la Clergerie, Djamé
-   Seddah and Benoît Sagot.
-6. `CTRL <https://huggingface.co/transformers/model_doc/ctrl.html>`__ (from Salesforce) released with the paper `CTRL:
-   A Conditional Transformer Language Model for Controllable Generation <https://arxiv.org/abs/1909.05858>`__ by Nitish
-   Shirish Keskar*, Bryan McCann*, Lav R. Varshney, Caiming Xiong and Richard Socher.
-7. `DeBERTa <https://huggingface.co/transformers/model_doc/deberta.html>`__ (from Microsoft Research) released with the
-   paper `DeBERTa: Decoding-enhanced BERT with Disentangled Attention <https://arxiv.org/abs/2006.03654>`__ by
+3. :doc:`BERT <model_doc/bert>` (from Google) released with the paper `BERT: Pre-training of Deep Bidirectional
+   Transformers for Language Understanding <https://arxiv.org/abs/1810.04805>`__ by Jacob Devlin, Ming-Wei Chang,
+   Kenton Lee and Kristina Toutanova.
+4. :doc:`BERT For Sequence Generation <model_doc/bertgeneration>` (from Google) released with the paper `Leveraging
+   Pre-trained Checkpoints for Sequence Generation Tasks <https://arxiv.org/abs/1907.12461>`__ by Sascha Rothe, Shashi
+   Narayan, Aliaksei Severyn.
+5. `Blenderbot <https://huggingface.co/transformers/master/model_doc/blenderbot.html>`__ (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.
+6. :doc:`CamemBERT <model_doc/camembert>` (from Inria/Facebook/Sorbonne) released with the paper `CamemBERT: a Tasty
+   French Language Model <https://arxiv.org/abs/1911.03894>`__ by Louis Martin*, Benjamin Muller*, Pedro Javier Ortiz
+   Suárez*, Yoann Dupont, Laurent Romary, Éric Villemonte de la Clergerie, Djamé Seddah and Benoît Sagot.
+7. :doc:`CTRL <model_doc/ctrl>` (from Salesforce) released with the paper `CTRL: A Conditional Transformer Language
+   Model for Controllable Generation <https://arxiv.org/abs/1909.05858>`__ by Nitish Shirish Keskar*, Bryan McCann*,
+   Lav R. Varshney, Caiming Xiong and Richard Socher.
+8. `DeBERTa <https://huggingface.co/transformers/master/model_doc/deberta.html>`__ (from Microsoft Research) released
+   with the paper `DeBERTa: Decoding-enhanced BERT with Disentangled Attention <https://arxiv.org/abs/2006.03654>`__ by
    Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen.
-8. `DialoGPT <https://huggingface.co/transformers/model_doc/dialogpt.html>`__ (from Microsoft Research) released with
-   the paper `DialoGPT: Large-Scale Generative Pre-training for Conversational Response Generation
-   <https://arxiv.org/abs/1911.00536>`__ by Yizhe Zhang, Siqi Sun, Michel Galley, Yen-Chun Chen, Chris Brockett, Xiang
-   Gao, Jianfeng Gao, Jingjing Liu, Bill Dolan.
-9. `DistilBERT <https://huggingface.co/transformers/model_doc/distilbert.html>`__ (from HuggingFace), released together
-   with the paper `DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter
-   <https://arxiv.org/abs/1910.01108>`__ by Victor Sanh, Lysandre Debut and Thomas Wolf. The same method has been
-   applied to compress GPT2 into `DistilGPT2
-   <https://github.com/huggingface/transformers/tree/master/examples/distillation>`__, RoBERTa into `DistilRoBERTa
-   <https://github.com/huggingface/transformers/tree/master/examples/distillation>`__, Multilingual BERT into
-   `DistilmBERT <https://github.com/huggingface/transformers/tree/master/examples/distillation>`__ and a German version
-   of DistilBERT.
-10. `DPR <https://github.com/facebookresearch/DPR>`__ (from Facebook) released with the paper `Dense Passage Retrieval
-    for Open-Domain Question Answering <https://arxiv.org/abs/2004.04906>`__ by Vladimir Karpukhin, Barlas Oğuz, Sewon
-    Min, Patrick Lewis, Ledell Wu, Sergey Edunov, Danqi Chen, and Wen-tau Yih.
-11. `ELECTRA <https://huggingface.co/transformers/model_doc/electra.html>`__ (from Google Research/Stanford University)
-    released with the paper `ELECTRA: Pre-training text encoders as discriminators rather than generators
-    <https://arxiv.org/abs/2003.10555>`__ by Kevin Clark, Minh-Thang Luong, Quoc V. Le, Christopher D. Manning.
-12. `FlauBERT <https://huggingface.co/transformers/model_doc/flaubert.html>`__ (from CNRS) released with the paper
-    `FlauBERT: Unsupervised Language Model Pre-training for French <https://arxiv.org/abs/1912.05372>`__ by Hang Le,
-    Loïc Vial, Jibril Frej, Vincent Segonne, Maximin Coavoux, Benjamin Lecouteux, Alexandre Allauzen, Benoît Crabbé,
-    Laurent Besacier, Didier Schwab.
-13. `Funnel Transformer <https://github.com/laiguokun/Funnel-Transformer>`__ (from CMU/Google Brain) released with the
-    paper `Funnel-Transformer: Filtering out Sequential Redundancy for Efficient Language Processing
-    <https://arxiv.org/abs/2006.03236>`__ by Zihang Dai, Guokun Lai, Yiming Yang, Quoc V. Le.
-14. `GPT <https://huggingface.co/transformers/model_doc/gpt.html>`__ (from OpenAI) released with the paper `Improving
-    Language Understanding by Generative Pre-Training <https://blog.openai.com/language-unsupervised/>`__ by Alec
-    Radford, Karthik Narasimhan, Tim Salimans and Ilya Sutskever.
-15. `GPT-2 <https://huggingface.co/transformers/model_doc/gpt2.html>`__ (from OpenAI) released with the paper `Language
-    Models are Unsupervised Multitask Learners <https://blog.openai.com/better-language-models/>`__ by Alec Radford*,
-    Jeffrey Wu*, Rewon Child, David Luan, Dario Amodei** and Ilya Sutskever**.
-16. `LayoutLM <https://github.com/microsoft/unilm/tree/master/layoutlm>`__ (from Microsoft Research Asia) released with
-    the paper `LayoutLM: Pre-training of Text and Layout for Document Image Understanding
-    <https://arxiv.org/abs/1912.13318>`__ by Yiheng Xu, Minghao Li, Lei Cui, Shaohan Huang, Furu Wei, Ming Zhou.
-17. `Longformer <https://huggingface.co/transformers/model_doc/longformer.html>`__ (from AllenAI) released with the
-    paper `Longformer: The Long-Document Transformer <https://arxiv.org/abs/2004.05150>`__ by Iz Beltagy, Matthew E.
-    Peters, Arman Cohan.
-18. `LXMERT <https://github.com/airsplay/lxmert>`__ (from UNC Chapel Hill) released with the paper `LXMERT: Learning
-    Cross-Modality Encoder Representations from Transformers for Open-Domain Question Answering
-    <https://arxiv.org/abs/1908.07490>`__ by Hao Tan and Mohit Bansal.
-19. `MarianMT <https://huggingface.co/transformers/model_doc/marian.html>`__ Machine translation models trained using
-    `OPUS <http://opus.nlpl.eu/>`__ data by Jörg Tiedemann. The `Marian Framework <https://marian-nmt.github.io/>`__ is
-    being developed by the Microsoft Translator Team.
-20. `MBart <https://github.com/pytorch/fairseq/tree/master/examples/mbart>`__ (from Facebook) released with the paper 
-    `Multilingual Denoising Pre-training for Neural Machine Translation <https://arxiv.org/abs/2001.08210>`__ by Yinhan
-    Liu, Jiatao Gu, Naman Goyal, Xian Li, Sergey Edunov, Marjan Ghazvininejad, Mike Lewis, Luke Zettlemoyer.
-21. `MMBT <https://github.com/facebookresearch/mmbt/>`__ (from Facebook), released together with the paper a
-    `Supervised Multimodal Bitransformers for Classifying Images and Text <https://arxiv.org/pdf/1909.02950.pdf>`__ by
-    Douwe Kiela, Suvrat Bhooshan, Hamed Firooz, Davide Testuggine.
-22. `Pegasus <https://github.com/google-research/pegasus>`__ (from Google) released with the paper `PEGASUS:
-    Pre-training with Extracted Gap-sentences for Abstractive Summarization <https://arxiv.org/abs/1912.08777>`__> by
-    Jingqing Zhang, Yao Zhao, Mohammad Saleh and Peter J. Liu.
-23. `Reformer <https://huggingface.co/transformers/model_doc/reformer.html>`__ (from Google Research) released with the
-    paper `Reformer: The Efficient Transformer <https://arxiv.org/abs/2001.04451>`__ by Nikita Kitaev, Łukasz Kaiser,
-    Anselm Levskaya.
-24. `RoBERTa <https://huggingface.co/transformers/model_doc/roberta.html>`__ (from Facebook), released together with
-    the paper a `Robustly Optimized BERT Pretraining Approach <https://arxiv.org/abs/1907.11692>`__ by Yinhan Liu, Myle
-    Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, Veselin Stoyanov.
-    ultilingual BERT into `DistilmBERT
+9. :doc:`DialoGPT <model_doc/dialogpt>` (from Microsoft Research) released with the paper `DialoGPT: Large-Scale
+   Generative Pre-training for Conversational Response Generation <https://arxiv.org/abs/1911.00536>`__ by Yizhe Zhang,
+   Siqi Sun, Michel Galley, Yen-Chun Chen, Chris Brockett, Xiang Gao, Jianfeng Gao, Jingjing Liu, Bill Dolan.
+10. :doc:`DistilBERT <model_doc/distilbert>` (from HuggingFace), released together with the paper `DistilBERT, a
+    distilled version of BERT: smaller, faster, cheaper and lighter <https://arxiv.org/abs/1910.01108>`__ by Victor
+    Sanh, Lysandre Debut and Thomas Wolf. The same method has been applied to compress GPT2 into `DistilGPT2
+    <https://github.com/huggingface/transformers/tree/master/examples/distillation>`__, RoBERTa into `DistilRoBERTa
+    <https://github.com/huggingface/transformers/tree/master/examples/distillation>`__, Multilingual BERT into
+    `DistilmBERT <https://github.com/huggingface/transformers/tree/master/examples/distillation>`__ and a German
+    version of DistilBERT.
+11. :doc:`DPR <model_doc/dpr>` (from Facebook) released with the paper `Dense Passage Retrieval for Open-Domain
+    Question Answering <https://arxiv.org/abs/2004.04906>`__ by Vladimir Karpukhin, Barlas Oğuz, Sewon Min, Patrick
+    Lewis, Ledell Wu, Sergey Edunov, Danqi Chen, and Wen-tau Yih.
+12. :doc:`ELECTRA <model_doc/electra>` (from Google Research/Stanford University) released with the paper `ELECTRA:
+    Pre-training text encoders as discriminators rather than generators <https://arxiv.org/abs/2003.10555>`__ by Kevin
+    Clark, Minh-Thang Luong, Quoc V. Le, Christopher D. Manning.
+13. :doc:`FlauBERT <model_doc/flaubert>` (from CNRS) released with the paper `FlauBERT: Unsupervised Language Model
+    Pre-training for French <https://arxiv.org/abs/1912.05372>`__ by Hang Le, Loïc Vial, Jibril Frej, Vincent Segonne,
+    Maximin Coavoux, Benjamin Lecouteux, Alexandre Allauzen, Benoît Crabbé, Laurent Besacier, Didier Schwab.
+14. :doc:`Funnel Transformer <model_doc/funnel>` (from CMU/Google Brain) released with the paper `Funnel-Transformer:
+    Filtering out Sequential Redundancy for Efficient Language Processing <https://arxiv.org/abs/2006.03236>`__ by
+    Zihang Dai, Guokun Lai, Yiming Yang, Quoc V. Le.
+15. :doc:`GPT <model_doc/gpt>` (from OpenAI) released with the paper `Improving Language Understanding by Generative
+    Pre-Training <https://blog.openai.com/language-unsupervised/>`__ by Alec Radford, Karthik Narasimhan, Tim Salimans
+    and Ilya Sutskever.
+16. :doc:`GPT-2 <model_doc/gpt2>` (from OpenAI) released with the paper `Language Models are Unsupervised Multitask
+    Learners <https://blog.openai.com/better-language-models/>`__ by Alec Radford*, Jeffrey Wu*, Rewon Child, David
+    Luan, Dario Amodei** and Ilya Sutskever**.
+17. :doc:`LayoutLM <model_doc/layoutlm>` (from Microsoft Research Asia) released with the paper `LayoutLM: Pre-training
+    of Text and Layout for Document Image Understanding <https://arxiv.org/abs/1912.13318>`__ by Yiheng Xu, Minghao Li,
+    Lei Cui, Shaohan Huang, Furu Wei, Ming Zhou.
+18. :doc:`Longformer <model_doc/longformer>` (from AllenAI) released with the paper `Longformer: The Long-Document
+    Transformer <https://arxiv.org/abs/2004.05150>`__ by Iz Beltagy, Matthew E. Peters, Arman Cohan.
+19. :doc:`LXMERT <model_doc/lxmert>` (from UNC Chapel Hill) released with the paper `LXMERT: Learning Cross-Modality
+    Encoder Representations from Transformers for Open-Domain Question Answering <https://arxiv.org/abs/1908.07490>`__
+    by Hao Tan and Mohit Bansal.
+20. :doc:`MarianMT <model_doc/marian>` Machine translation models trained using `OPUS <http://opus.nlpl.eu/>`__ data by
+    Jörg Tiedemann. The `Marian Framework <https://marian-nmt.github.io/>`__ is being developed by the Microsoft
+    Translator Team.
+21. :doc:`MBart <model_doc/mbart>` (from Facebook) released with the paper  `Multilingual Denoising Pre-training for
+    Neural Machine Translation <https://arxiv.org/abs/2001.08210>`__ by Yinhan Liu, Jiatao Gu, Naman Goyal, Xian Li,
+    Sergey Edunov, Marjan Ghazvininejad, Mike Lewis, Luke Zettlemoyer.
+22. :doc:`Pegasus <model_doc/pegasus>` (from Google) released with the paper `PEGASUS: Pre-training with Extracted
+    Gap-sentences for Abstractive Summarization <https://arxiv.org/abs/1912.08777>`__> by Jingqing Zhang, Yao Zhao,
+    Mohammad Saleh and Peter J. Liu.
+23. :doc:`Reformer <model_doc/reformer>` (from Google Research) released with the paper `Reformer: The Efficient
+    Transformer <https://arxiv.org/abs/2001.04451>`__ by Nikita Kitaev, Łukasz Kaiser, Anselm Levskaya.
+24. :doc:`RoBERTa <model_doc/roberta>` (from Facebook), released together with the paper a `Robustly Optimized BERT
+    Pretraining Approach <https://arxiv.org/abs/1907.11692>`__ by Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar
+    Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, Veselin Stoyanov. ultilingual BERT into `DistilmBERT
     <https://github.com/huggingface/transformers/tree/master/examples/distillation>`__ and a German version of
     DistilBERT.
-25. `SqueezeBert <https://huggingface.co/transformers/model_doc/squeezebert.html>`__ released with the paper
+25. `SqueezeBert <https://huggingface.co/transformers/master/model_doc/squeezebert.html>`__ released with the paper
     `SqueezeBERT: What can computer vision teach NLP about efficient neural networks?
     <https://arxiv.org/abs/2006.11316>`__ by Forrest N. Iandola, Albert E. Shaw, Ravi Krishna, and Kurt W. Keutzer.
-26. `T5 <https://huggingface.co/transformers/model_doc/t5.html>`__ (from Google AI) released with the paper `Exploring
-    the Limits of Transfer Learning with a Unified Text-to-Text Transformer <https://arxiv.org/abs/1910.10683>`__ by
-    Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi
-    Zhou and Wei Li and Peter J. Liu.
-27. `Transformer-XL <https://huggingface.co/transformers/model_doc/transformerxl.html>`__ (from Google/CMU) released
-    with the paper `Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context
-    <https://arxiv.org/abs/1901.02860>`__ by Zihang Dai*, Zhilin Yang*, Yiming Yang, Jaime Carbonell, Quoc V. Le,
-    Ruslan Salakhutdinov.
-28. `XLM <https://huggingface.co/transformers/model_doc/xlm.html>`__ (from Facebook) released together with the paper
-    `Cross-lingual Language Model Pretraining <https://arxiv.org/abs/1901.07291>`__ by Guillaume Lample and Alexis
-    Conneau.
-29. `XLM-RoBERTa <https://huggingface.co/transformers/model_doc/xlmroberta.html>`__ (from Facebook AI), released
-    together with the paper `Unsupervised Cross-lingual Representation Learning at Scale
-    <https://arxiv.org/abs/1911.02116>`__ by Alexis Conneau*, Kartikay Khandelwal*, Naman Goyal, Vishrav Chaudhary,
-    Guillaume Wenzek, Francisco Guzmán, Edouard Grave, Myle Ott, Luke Zettlemoyer and Veselin Stoyanov.
-30. `XLNet <https://huggingface.co/transformers/model_doc/xlnet.html>`__ (from Google/CMU) released with the paper
-    `​XLNet: Generalized Autoregressive Pretraining for Language Understanding <https://arxiv.org/abs/1906.08237>`__ by
-    Zhilin Yang*, Zihang Dai*, Yiming Yang, Jaime Carbonell, Ruslan Salakhutdinov, Quoc V. Le.
+26. :doc:`T5 <model_doc/t5>` (from Google AI) released with the paper `Exploring the Limits of Transfer Learning with a
+    Unified Text-to-Text Transformer <https://arxiv.org/abs/1910.10683>`__ by Colin Raffel and Noam Shazeer and Adam
+    Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu.
+27. :doc:`Transformer-XL <model_doc/transformerxl>` (from Google/CMU) released with the paper `Transformer-XL:
+    Attentive Language Models Beyond a Fixed-Length Context <https://arxiv.org/abs/1901.02860>`__ by Zihang Dai*,
+    Zhilin Yang*, Yiming Yang, Jaime Carbonell, Quoc V. Le, Ruslan Salakhutdinov.
+28. :doc:`XLM <model_doc/xlm>` (from Facebook) released together with the paper `Cross-lingual Language Model
+    Pretraining <https://arxiv.org/abs/1901.07291>`__ by Guillaume Lample and Alexis Conneau.
+29. :doc:`XLM-RoBERTa <model_doc/xlmroberta>` (from Facebook AI), released together with the paper `Unsupervised
+    Cross-lingual Representation Learning at Scale <https://arxiv.org/abs/1911.02116>`__ by Alexis Conneau*, Kartikay
+    Khandelwal*, Naman Goyal, Vishrav Chaudhary, Guillaume Wenzek, Francisco Guzmán, Edouard Grave, Myle Ott, Luke
+    Zettlemoyer and Veselin Stoyanov.
+30. :doc:`XLNet <model_doc/xlnet>` (from Google/CMU) released with the paper `​XLNet: Generalized Autoregressive
+    Pretraining for Language Understanding <https://arxiv.org/abs/1906.08237>`__ by Zhilin Yang*, Zihang Dai*, Yiming
+    Yang, Jaime Carbonell, Ruslan Salakhutdinov, Quoc V. Le.
 31. `Other community models <https://huggingface.co/models>`__, contributed by the `community
     <https://huggingface.co/users>`__.
 
@@ -213,6 +204,7 @@ conversion utilities for the following models:
     :maxdepth: 2
     :caption: Main Classes
 
+    main_classes/callback
     main_classes/configuration
     main_classes/logging
     main_classes/model
@@ -232,6 +224,7 @@ conversion utilities for the following models:
     model_doc/bart
     model_doc/bert
     model_doc/bertgeneration
+    model_doc/blenderbot
     model_doc/camembert
     model_doc/ctrl
     model_doc/deberta
@@ -270,3 +263,4 @@ conversion utilities for the following models:
     internal/modeling_utils
     internal/pipelines_utils
     internal/tokenization_utils
+    internal/trainer_utils

docs/source/internal/trainer_utils.rst

@@ -0,0 +1,21 @@
+Utilities for Trainer
+-----------------------------------------------------------------------------------------------------------------------
+
+This page lists all the utility functions used by :class:`~transformers.Trainer`.
+
+Most of those are only useful if you are studying the code of the Trainer in the library.
+
+Utilities
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: transformers.EvalPrediction
+
+.. autofunction:: transformers.set_seed
+
+.. autofunction:: transformers.torch_distributed_zero_first
+
+
+Callbacks internals
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: transformers.trainer_callback.CallbackHandler

docs/source/main_classes/callback.rst

@@ -0,0 +1,68 @@
+Callbacks
+-----------------------------------------------------------------------------------------------------------------------
+
+Callbacks are objects that can customize the behavior of the training loop in the PyTorch
+:class:`~transformers.Trainer` (this feature is not yet implemented in TensorFlow) that can inspect the training loop
+state (for progress reporting, logging on TensorBoard or other ML platforms...) and take decisions (like early
+stopping).
+
+Callbacks are "read only" pieces of code, apart from the :class:`~transformers.TrainerControl` object they return, they
+cannot change anything in the training loop. For customizations that require changes in the training loop, you should
+subclass :class:`~transformers.Trainer` and override the methods you need (see :doc:`trainer` for examples).
+
+By default a :class:`~transformers.Trainer` will use the following callbacks:
+
+- :class:`~transformers.DefaultFlowCallback` which handles the default behavior for logging, saving and evaluation.
+- :class:`~transformers.PrinterCallback` or :class:`~transformers.ProrgressCallback` to display progress and print the
+  logs (the first one is used if you deactivate tqdm through the :class:`~transformers.TrainingArguments`, otherwise
+  it's the second one).
+- :class:`~transformers.integrations.TensorBoardCallback` if tensorboard is accessible (either through PyTorch >= 1.4
+  or tensorboardX).
+- :class:`~transformers.integrations.WandbCallback` if `wandb <https://www.wandb.com/>`__ is installed.
+- :class:`~transformers.integrations.CometCallback` if `comet_ml <https://www.comet.ml/site/>`__ is installed.
+
+The main class that implements callbacks is :class:`~transformers.TrainerCallback`. It gets the
+:class:`~transformers.TrainingArguments` used to instantiate the :class:`~transformers.Trainer`, can access that
+Trainer's internal state via :class:`~transformers.TrainerState`, and can take some actions on the training loop via
+:class:`~transformers.TrainerControl`.
+
+
+Available Callbacks
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Here is the list of the available :class:`~transformers.TrainerCallback` in the library:
+
+.. autoclass:: transformers.integrations.CometCallback
+    :members: setup
+
+.. autoclass:: transformers.DefaultFlowCallback
+
+.. autoclass:: transformers.PrinterCallback
+
+.. autoclass:: transformers.ProgressCallback
+
+.. autoclass:: transformers.integrations.TensorBoardCallback
+
+.. autoclass:: transformers.integrations.WandbCallback
+    :members: setup
+
+
+TrainerCallback
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: transformers.TrainerCallback
+    :members:
+
+
+TrainerState
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: transformers.TrainerState
+    :members:
+
+
+TrainerControl
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: transformers.TrainerControl
+    :members:

docs/source/main_classes/trainer.rst

@@ -15,10 +15,9 @@ Both :class:`~transformers.Trainer` and :class:`~transformers.TFTrainer` contain
 previous features. To inject custom behavior you can subclass them and override the following methods:
 
 - **get_train_dataloader**/**get_train_tfdataset** -- Creates the training DataLoader (PyTorch) or TF Dataset.
-- **get_eval_dataloader**/**get_eval_tfdataset** -- Creates the evaulation DataLoader (PyTorch) or TF Dataset.
+- **get_eval_dataloader**/**get_eval_tfdataset** -- Creates the evaluation DataLoader (PyTorch) or TF Dataset.
 - **get_test_dataloader**/**get_test_tfdataset** -- Creates the test DataLoader (PyTorch) or TF Dataset.
 - **log** -- Logs information on the various objects watching training.
-- **setup_wandb** -- Setups wandb (see `here <https://docs.wandb.com/huggingface>`__ for more information).
 - **create_optimizer_and_scheduler** -- Setups the optimizer and learning rate scheduler if they were not passed at
   init.
 - **compute_loss** - Computes the loss on a batch of training inputs.
@@ -40,6 +39,10 @@ Here is an example of how to customize :class:`~transformers.Trainer` using a cu
             logits = outputs[0]
             return my_custom_loss(logits, labels)
 
+Another way to customize the training loop behavior for the PyTorch :class:`~transformers.Trainer` is to use
+:doc:`callbacks <callback>` that can inspect the training loop state (for progress reporting, logging on TensorBoard or
+other ML platforms...) and take decisions (like early stopping).
+
 
 Trainer
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
@@ -47,29 +50,23 @@ Trainer
 .. autoclass:: transformers.Trainer
     :members:
 
+
 TFTrainer
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
 .. autoclass:: transformers.TFTrainer
     :members:
 
+
 TrainingArguments
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
 .. autoclass:: transformers.TrainingArguments
     :members:
 
+
 TFTrainingArguments
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
 .. autoclass:: transformers.TFTrainingArguments
     :members:
-
-Utilities
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-
-.. autoclass:: transformers.EvalPrediction
-
-.. autofunction:: transformers.set_seed
-
-.. autofunction:: transformers.torch_distributed_zero_first

docs/source/model_doc/blenderbot.rst

@@ -0,0 +1,75 @@
+Blenderbot
+-----------------------------------------------------------------------------------------------------------------------
+**DISCLAIMER:** If you see something strange,
+file a `Github Issue <https://github.com/huggingface/transformers/issues/new?assignees=&labels=&template=bug-report.md&title>`__ .
+
+Overview
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+The Blender chatbot model was proposed in `Recipes for building an open-domain chatbot <https://arxiv.org/pdf/2004.13637.pdf>`__ 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 on 30 Apr 2020.
+
+The abstract of the paper is the following:
+
+*Building open-domain chatbots is a challenging area for machine learning research. While prior work has shown that scaling neural models in the number of parameters and the size of the data they are trained on gives improved results, we show that other ingredients are important for a high-performing chatbot. Good conversation requires a number of skills that an expert conversationalist blends in a seamless way: providing engaging talking points and listening to their partners, and displaying knowledge, empathy and personality appropriately, while maintaining a consistent persona. We show that large scale models can learn these skills when given appropriate training data and choice of generation strategy. We build variants of these recipes with 90M, 2.7B and 9.4B parameter models, and make our models and code publicly available. Human evaluations show our best models are superior to existing approaches in multi-turn dialogue in terms of engagingness and humanness measurements. We then discuss the limitations of this work by analyzing failure cases of our models.*
+
+The authors' code can be found `here <https://github.com/facebookresearch/ParlAI>`__ .
+
+
+Implementation Notes
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+- Blenderbot uses a standard `seq2seq model transformer <https://arxiv.org/pdf/1706.03762.pdf>`__ based architecture.
+- It inherits completely from :class:`~transformers.BartForConditionalGeneration`
+- Even though blenderbot is one model, it uses two tokenizers :class:`~transformers.BlenderbotSmallTokenizer` for 90M checkpoint and :class:`~transformers.BlenderbotTokenizer` for all other checkpoints.
+- :class:`~transformers.BlenderbotSmallTokenizer` will always return :class:`~transformers.BlenderbotSmallTokenizer`, regardless of checkpoint. To use the 3B parameter checkpoint, you must call :class:`~transformers.BlenderbotTokenizer` directly.
+- Available checkpoints can be found in the `model hub <https://huggingface.co/models?search=blenderbot>`__.
+
+
+Usage
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Model Usage:
+
+        >>> from transformers import BlenderbotSmallTokenizer, BlenderbotForConditionalGeneration
+        >>> mname = 'facebook/blenderbot-90M'
+        >>> model = BlenderbotForConditionalGeneration.from_pretrained(mname)
+        >>> tokenizer = BlenderbotSmallTokenizer.from_pretrained(mname)
+        >>> UTTERANCE = "My friends are cool but they eat too many carbs."
+        >>> inputs = tokenizer([UTTERANCE], return_tensors='pt')
+        >>> reply_ids = model.generate(**inputs)
+        >>> print([tokenizer.decode(g, skip_special_tokens=True, clean_up_tokenization_spaces=False) for g in reply_ids])
+
+
+See Config Values:
+
+        >>> from transformers import BlenderbotConfig
+        >>> config_90 = BlenderbotConfig.from_pretrained("facebook/blenderbot-90M")
+        >>> config_90.to_diff_dict()  # show interesting Values.
+        >>> configuration_3B = BlenderbotConfig("facebook/blenderbot-3B")
+        >>> configuration_3B.to_diff_dict()
+
+
+BlenderbotConfig
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+.. autoclass:: transformers.BlenderbotConfig
+    :members:
+
+BlenderbotTokenizer
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: transformers.BlenderbotTokenizer
+    :members: build_inputs_with_special_tokens
+
+BlenderbotSmallTokenizer
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: transformers.BlenderbotSmallTokenizer
+    :members:
+
+
+BlenderbotForConditionalGeneration
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+See :obj:`transformers.BartForConditionalGeneration` for arguments to `forward` and `generate`
+
+.. autoclass:: transformers.BlenderbotForConditionalGeneration
+    :members:

docs/source/model_doc/gpt.rst

@@ -104,6 +104,13 @@ OpenAIGPTDoubleHeadsModel
     :members: forward
 
 
+OpenAIGPTForSequenceClassification
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: transformers.OpenAIGPTForSequenceClassification
+    :members: forward
+
+
 TFOpenAIGPTModel
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 

docs/source/model_doc/gpt2.rst

@@ -88,6 +88,13 @@ GPT2DoubleHeadsModel
     :members: forward
 
 
+GPT2ForSequenceClassification
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: transformers.GPT2ForSequenceClassification
+    :members: forward
+
+
 TFGPT2Model
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 

docs/source/model_doc/squeezebert.rst

@@ -1,8 +1,8 @@
 SqueezeBERT
-----------------------------------------------------
+-----------------------------------------------------------------------------------------------------------------------
 
 Overview
-~~~~~~~~~~~~~~~~~~~~~
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
 The SqueezeBERT model was proposed in
 `SqueezeBERT: What can computer vision teach NLP about efficient neural networks?
@@ -40,14 +40,14 @@ Tips:
   `squeezebert/squeezebert-mnli-headless` checkpoint.
 
 SqueezeBertConfig
-~~~~~~~~~~~~~~~~~~~~~~
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
 .. autoclass:: transformers.SqueezeBertConfig
     :members:
 
 
 SqueezeBertTokenizer
-~~~~~~~~~~~~~~~~~~~~~~~~
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
 .. autoclass:: transformers.SqueezeBertTokenizer
     :members: build_inputs_with_special_tokens, get_special_tokens_mask,
@@ -55,49 +55,49 @@ SqueezeBertTokenizer
 
 
 SqueezeBertTokenizerFast
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
 .. autoclass:: transformers.SqueezeBertTokenizerFast
     :members:
 
 
 SqueezeBertModel
-~~~~~~~~~~~~~~~~~~~~
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
 .. autoclass:: transformers.SqueezeBertModel
     :members:
 
 
 SqueezeBertForMaskedLM
-~~~~~~~~~~~~~~~~~~~~~~~~~~
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
 .. autoclass:: transformers.SqueezeBertForMaskedLM
     :members:
 
 
 SqueezeBertForSequenceClassification
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
 .. autoclass:: transformers.SqueezeBertForSequenceClassification
     :members:
 
 
 SqueezeBertForMultipleChoice
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
 .. autoclass:: transformers.SqueezeBertForMultipleChoice
     :members:
 
 
 SqueezeBertForTokenClassification
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
 .. autoclass:: transformers.SqueezeBertForTokenClassification
     :members:
 
 
 SqueezeBertForQuestionAnswering
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
 .. autoclass:: transformers.SqueezeBertForQuestionAnswering
     :members:

docs/source/philosophy.rst

@@ -66,7 +66,7 @@ The library is built around three types of classes for each model:
 All these classes can be instantiated from pretrained instances and saved locally using two methods:
 
 - :obj:`from_pretrained()` lets you instantiate a model/configuration/tokenizer from a pretrained version either
-  provided by the library itself (the suported models are provided in the list :doc:`here <pretrained_models>`
+  provided by the library itself (the supported models are provided in the list :doc:`here <pretrained_models>`
   or stored locally (or on a server) by the user,
 - :obj:`save_pretrained()` lets you save a model/configuration/tokenizer locally so that it can be reloaded using
   :obj:`from_pretrained()`.

docs/source/pretrained_models.rst

@@ -294,10 +294,10 @@ For a list that includes community-uploaded models, refer to `https://huggingfac
 |                    | ``t5-11B``                                                 | | ~11B parameters with 24-layers, 1024-hidden-state, 65536 feed-forward hidden-state, 128-heads,                                      |
 |                    |                                                            | | Trained on English text: the Colossal Clean Crawled Corpus (C4)                                                                     |
 +--------------------+------------------------------------------------------------+---------------------------------------------------------------------------------------------------------------------------------------+
-| XLM-RoBERTa        | ``xlm-roberta-base``                                       | | ~125M parameters with 12-layers, 768-hidden-state, 3072 feed-forward hidden-state, 8-heads,                                         |
+| XLM-RoBERTa        | ``xlm-roberta-base``                                       | | ~270M parameters with 12-layers, 768-hidden-state, 3072 feed-forward hidden-state, 8-heads,                                         |
 |                    |                                                            | | Trained on on 2.5 TB of newly created clean CommonCrawl data in 100 languages                                                       |
 |                    +------------------------------------------------------------+---------------------------------------------------------------------------------------------------------------------------------------+
-|                    | ``xlm-roberta-large``                                      | | ~355M parameters with 24-layers, 1027-hidden-state, 4096 feed-forward hidden-state, 16-heads,                                       |
+|                    | ``xlm-roberta-large``                                      | | ~550M parameters with 24-layers, 1024-hidden-state, 4096 feed-forward hidden-state, 16-heads,                                       |
 |                    |                                                            | | Trained on 2.5 TB of newly created clean CommonCrawl data in 100 languages                                                          |
 +--------------------+------------------------------------------------------------+---------------------------------------------------------------------------------------------------------------------------------------+
 | FlauBERT           | ``flaubert/flaubert_small_cased``                          | | 6-layer, 512-hidden, 8-heads, 54M parameters                                                                                        |

examples/language-modeling/run_language_modeling.py

@@ -24,8 +24,11 @@ import logging
 import math
 import os
 from dataclasses import dataclass, field
+from glob import glob
 from typing import Optional
 
+from torch.utils.data import ConcatDataset
+
 from transformers import (
     CONFIG_MAPPING,
     MODEL_WITH_LM_HEAD_MAPPING,
@@ -87,6 +90,13 @@ class DataTrainingArguments:
     train_data_file: Optional[str] = field(
         default=None, metadata={"help": "The input training data file (a text file)."}
     )
+    train_data_files: Optional[str] = field(
+        default=None,
+        metadata={
+            "help": "The input training data files (multiple files in glob format). "
+            "Very often splitting large files to smaller files can prevent tokenizer going out of memory"
+        },
+    )
     eval_data_file: Optional[str] = field(
         default=None,
         metadata={"help": "An optional input evaluation data file to evaluate the perplexity on (a text file)."},
@@ -131,17 +141,24 @@ def get_dataset(
     evaluate: bool = False,
     cache_dir: Optional[str] = None,
 ):
-    file_path = args.eval_data_file if evaluate else args.train_data_file
-    if args.line_by_line:
-        return LineByLineTextDataset(tokenizer=tokenizer, file_path=file_path, block_size=args.block_size)
+    def _dataset(file_path):
+        if args.line_by_line:
+            return LineByLineTextDataset(tokenizer=tokenizer, file_path=file_path, block_size=args.block_size)
+        else:
+            return TextDataset(
+                tokenizer=tokenizer,
+                file_path=file_path,
+                block_size=args.block_size,
+                overwrite_cache=args.overwrite_cache,
+                cache_dir=cache_dir,
+            )
+
+    if evaluate:
+        return _dataset(args.eval_data_file)
+    elif args.train_data_files:
+        return ConcatDataset([_dataset(f) for f in glob(args.train_data_files)])
     else:
-        return TextDataset(
-            tokenizer=tokenizer,
-            file_path=file_path,
-            block_size=args.block_size,
-            overwrite_cache=args.overwrite_cache,
-            cache_dir=cache_dir,
-        )
+        return _dataset(args.train_data_file)
 
 
 def main():

examples/lightning_base.py

@@ -119,7 +119,7 @@ class BaseTransformer(pl.LightningModule):
     def get_lr_scheduler(self):
         get_schedule_func = arg_to_scheduler[self.hparams.lr_scheduler]
         scheduler = get_schedule_func(
-            self.opt, num_warmup_steps=self.hparams.warmup_steps, num_training_steps=self.total_steps
+            self.opt, num_warmup_steps=self.hparams.warmup_steps, num_training_steps=self.total_steps()
         )
         scheduler = {"scheduler": scheduler, "interval": "step", "frequency": 1}
         return scheduler
@@ -159,19 +159,20 @@ class BaseTransformer(pl.LightningModule):
     def test_epoch_end(self, outputs):
         return self.validation_end(outputs)
 
-    @property
     def total_steps(self) -> int:
         """The number of total training steps that will be run. Used for lr scheduler purposes."""
         num_devices = max(1, self.hparams.gpus)  # TODO: consider num_tpu_cores
         effective_batch_size = self.hparams.train_batch_size * self.hparams.accumulate_grad_batches * num_devices
-        dataset_size = len(self.train_loader.dataset)
-        return (dataset_size / effective_batch_size) * self.hparams.max_epochs
+        return (self.dataset_size / effective_batch_size) * self.hparams.max_epochs
 
     def setup(self, mode):
-        if mode == "fit":
+        if mode == "test":
+            self.dataset_size = len(self.test_dataloader().dataset)
+        else:
             self.train_loader = self.get_dataloader("train", self.hparams.train_batch_size, shuffle=True)
+            self.dataset_size = len(self.train_loader.dataset)
 
-    def get_dataloader(self, type_path, batch_size, shuffle=False):
+    def get_dataloader(self, type_path: str, batch_size: int, shuffle: bool = False):
         raise NotImplementedError("You must implement this for your task")
 
     def train_dataloader(self):

examples/requirements.txt

@@ -5,7 +5,7 @@ psutil
 sacrebleu
 rouge-score
 tensorflow_datasets
-pytorch-lightning==0.8.5
+pytorch-lightning==0.9.0
 matplotlib
 git-python==1.0.3
 faiss-cpu

examples/seq2seq/README.md

@@ -12,14 +12,13 @@ For `bertabs` instructions, see [`bertabs/README.md`](bertabs/README.md).
 - `MBartForConditionalGeneration`
 - `FSMTForConditionalGeneration`
 - `T5ForConditionalGeneration`
-    
 
 ## Datasets
 
 #### XSUM:
 ```bash
 cd examples/seq2seq
-wget https://s3.amazonaws.com/datasets.huggingface.co/summarization/xsum.tar.gz
+wget https://cdn-datasets.huggingface.co/summarization/xsum.tar.gz
 tar -xzvf xsum.tar.gz
 export XSUM_DIR=${PWD}/xsum
 ```
@@ -29,7 +28,7 @@ To use your own data, copy that files format. Each article to be summarized is o
 #### CNN/DailyMail
 ```bash
 cd examples/seq2seq
-wget https://s3.amazonaws.com/datasets.huggingface.co/summarization/cnn_dm_v2.tgz
+wget https://cdn-datasets.huggingface.co/summarization/cnn_dm_v2.tgz
 tar -xzvf cnn_dm_v2.tgz  # empty lines removed
 mv cnn_cln cnn_dm
 export CNN_DIR=${PWD}/cnn_dm
@@ -39,7 +38,7 @@ this should make a directory called `cnn_dm/` with 6 files.
 #### WMT16 English-Romanian Translation Data:
 download with this command:
 ```bash
-wget https://s3.amazonaws.com/datasets.huggingface.co/translation/wmt_en_ro.tar.gz
+wget https://cdn-datasets.huggingface.co/translation/wmt_en_ro.tar.gz
 tar -xzvf wmt_en_ro.tar.gz
 export ENRO_DIR=${PWD}/wmt_en_ro
 ```
@@ -47,7 +46,7 @@ this should make a directory called `wmt_en_ro/` with 6 files.
 
 #### WMT English-German:
 ```bash
-wget https://s3.amazonaws.com/datasets.huggingface.co/translation/wmt_en_de.tgz
+wget https://cdn-datasets.huggingface.co/translation/wmt_en_de.tgz
 tar -xzvf wmt_en_de.tgz
 export DATA_DIR=${PWD}/wmt_en_de
 ```
@@ -100,7 +99,7 @@ All finetuning bash scripts call finetune.py (or distillation.py) with reasonabl
 To see all the possible command line options, run:
 
 ```bash
- ./finetune.py --help 
+./finetune.py --help
 ```
 
 ### Finetuning Training Params
@@ -192,7 +191,7 @@ model = AutoModelForSeq2SeqLM.from_pretrained(f'{output_dir}/best_tfmr')
 ### Fine-tuning using Seq2SeqTrainer
 To use `Seq2SeqTrainer` for fine-tuning you should use the `finetune_trainer.py` script. It subclasses `Trainer` to extend it for seq2seq training. Except the `Trainer` releated `TrainingArguments`, it shares the same argument names as that of `finetune.py` file. One notable difference is that, calculating generative metrics (BLEU, ROUGE) is optional and is controlled using the `--predict_with_generate` argument, set this argument to calculate BLEU and ROUGE metrics.
 
-With PyTorch 1.6+ it'll automatically use `native AMP` when `--fp16` is set. 
+With PyTorch 1.6+ it'll automatically use `native AMP` when `--fp16` is set.
 
 To see all the possible command line options, run:
 
@@ -265,6 +264,7 @@ export DATA_DIR=cnn_dm
     --fp16 \
     --bs 32
 ```
+
 ### Multi-GPU Evaluation
 here is a command to run xsum evaluation on 8 GPUS. It is more than linearly faster than run_eval.py in some cases 
 because it uses SortishSampler to minimize padding. You can also use it on 1 GPU. `data_dir` must have 
@@ -391,6 +391,17 @@ runtime: 13H on V-100 16GB GPU.
 pytest examples/seq2seq/
 ```
 
+### Converting pytorch-lightning checkpoints
+pytorch lightning ``-do_predict`` often fails, after you are done training, the best way to evaluate your model is to convert it.
+
+This should be done for you, with a file called `{save_dir}/best_tfmr`. 
+
+If that file doesn't exist but you have a lightning `.ckpt` file, you can run
+```bash
+python convert_pl_checkpoint_to_hf.py PATH_TO_CKPT  randomly_initialized_hf_model_path save_dir/best_tfmr
+```
+Then either `run_eval` or `run_distributed_eval` with `save_dir/best_tfmr` (see previous sections)
+
 
 ## Experimental Features 
 These features are harder to use and not always useful.
@@ -419,4 +430,3 @@ uses 12,723 batches of length 48 and takes slightly more time 9.5 minutes.
 The feature is still experimental, because:
 + we can make it much more robust if we have memory mapped/preprocessed datasets.
 + The speedup over sortish sampler is not that large at the moment.
-

examples/seq2seq/bertabs/README.md

@@ -39,7 +39,7 @@ python run_summarization.py \
     --compute_rouge true
 ```
 
-The scripts executes on GPU if one is available and if `no_cuda` is not set to `true`. Inference on multiple GPUs is not suported yet. The ROUGE scores will be displayed in the console at the end of evaluation and written in a `rouge_scores.txt` file. The script takes 30 hours to compute with a single Tesla V100 GPU and a batch size of 10 (300,000 texts to summarize).
+The scripts executes on GPU if one is available and if `no_cuda` is not set to `true`. Inference on multiple GPUs is not supported yet. The ROUGE scores will be displayed in the console at the end of evaluation and written in a `rouge_scores.txt` file. The script takes 30 hours to compute with a single Tesla V100 GPU and a batch size of 10 (300,000 texts to summarize).
 
 ## Summarize any text
 

examples/seq2seq/distillation.py

@@ -17,7 +17,7 @@ from finetune import main as ft_main
 from make_student import create_student_by_copying_alternating_layers, get_layers_to_supervise
 from transformers import AutoModelForSeq2SeqLM, MBartTokenizer, T5ForConditionalGeneration
 from transformers.modeling_bart import shift_tokens_right
-from utils import calculate_bleu, freeze_params, label_smoothed_nll_loss, pickle_load, use_task_specific_params
+from utils import calculate_bleu, freeze_params, label_smoothed_nll_loss, use_task_specific_params
 
 
 # need the parent dir module
@@ -28,7 +28,7 @@ from lightning_base import generic_train  # noqa
 class BartSummarizationDistiller(SummarizationModule):
     """Supports Bart, Pegasus and other models that inherit from Bart."""
 
-    loss_names = ["loss", "ce_loss", "mlm_loss", "enc_mse_loss", "hid_loss_enc", "hid_loss_dec"]
+    loss_names = ["loss", "ce_loss", "mlm_loss", "hid_loss_enc", "hid_loss_dec"]
 
     def __init__(self, hparams):
         assert Path(hparams.data_dir).exists()
@@ -46,9 +46,19 @@ class BartSummarizationDistiller(SummarizationModule):
         if hparams.length_penalty != -1:
             student.config.length_penalty = hparams.length_penalty
         super().__init__(hparams, model=student, config=student.config)
+        model_type = student.config.model_type
         self.e_layer_ids, self.d_layer_ids = e_layer_ids, d_layer_ids  # type: List[int], List[int]
-        self.different_encoder = hparams.student_encoder_layers != teacher.config.encoder_layers
-        self.different_decoder = hparams.student_decoder_layers != teacher.config.decoder_layers
+
+        if model_type == "t5":
+            teacher_encoder_layers = len(teacher.get_encoder().block)
+            teacher_decoder_layers = len(teacher.get_decoder().block)
+        else:
+            teacher_encoder_layers = teacher.config.encoder_layers
+            teacher_decoder_layers = teacher.config.decoder_layers
+
+        self.different_encoder = hparams.student_encoder_layers != teacher_encoder_layers
+        self.different_decoder = hparams.student_decoder_layers != teacher_decoder_layers
+
         self.teacher = teacher
         freeze_params(self.teacher)
 
@@ -59,17 +69,17 @@ class BartSummarizationDistiller(SummarizationModule):
                 del self.teacher.encoder
         # Intermediate supervision: Decide which layers to supervise
         if hparams.supervise_forward:
-            self.d_matches = get_layers_to_supervise(
-                n_student=len(self.d_layer_ids), n_teacher=self.teacher.config.decoder_layers
-            )
-        else:
+            self.e_matches = get_layers_to_supervise(n_student=len(self.e_layer_ids), n_teacher=teacher_encoder_layers)
+            self.d_matches = get_layers_to_supervise(n_student=len(self.d_layer_ids), n_teacher=teacher_decoder_layers)
+        else:  # student layer should emulate hidden states of the teacher layer it was copied from
+            self.e_matches = self.e_layer_ids
             self.d_matches = self.d_layer_ids
+
         self.ce_loss_fct = nn.KLDivLoss(reduction="batchmean")
         self.temperature = 2.0
         self.alpha_mlm = hparams.alpha_mlm
         self.alpha_ce = hparams.alpha_ce
         self.alpha_hid = hparams.alpha_hid
-        self.alpha_encoder_loss = hparams.alpha_encoder_loss
         gc.collect()
         torch.cuda.empty_cache()
 
@@ -129,7 +139,7 @@ class BartSummarizationDistiller(SummarizationModule):
             output_hidden_states=True,
             output_attentions=False,
             use_cache=False,
-        )  # TODO(@sshleifer): return_dict=True cleanup
+        )
 
         # Same cross entropy vs. label smoothing logic as finetune.py
         assert lm_logits.shape[-1] == self.model.config.vocab_size
@@ -146,30 +156,32 @@ class BartSummarizationDistiller(SummarizationModule):
         def zero_tensor():
             return torch.tensor(0.0).type_as(student_lm_loss)
 
-        loss_encoder, hid_loss_enc, hid_loss_dec = zero_tensor(), zero_tensor(), zero_tensor()
-        if self.different_encoder:
+        hid_loss_enc, hid_loss_dec = zero_tensor(), zero_tensor()
+        if self.different_encoder:  # compute encoder hidden state loss
             with torch.no_grad():
-                teacher_enc_outputs, teacher_enc_hid, _ = self.teacher.get_encoder()(
-                    input_ids, attention_mask=src_mask, output_hidden_states=True
-                )
-            # DEPRECATE THIS
-            if self.hparams.alpha_encoder_loss > 0:
-                loss_encoder = self.calc_mse_loss(enc_outputs, teacher_enc_outputs, src_mask)
+                teacher_enc_hid = self.teacher.get_encoder()(
+                    input_ids, attention_mask=src_mask, output_hidden_states=True, return_dict=True
+                ).hidden_states
 
-            hid_loss_enc = self.calc_hidden_loss(src_mask, enc_hidden_state, teacher_enc_hid, self.e_layer_ids)
-
-        teacher_enc_outputs = (enc_outputs,)
-        assert isinstance(teacher_enc_outputs, tuple), type(teacher_enc_outputs)
+            hid_loss_enc = self.calc_hidden_loss(
+                src_mask,
+                enc_hidden_state,
+                teacher_enc_hid,
+                self.e_matches,
+                normalize_hidden=self.hparams.normalize_hidden,
+            )
 
         with torch.no_grad():
-            tloss, tlogits, tdec_hidden, _ = self.teacher(
+            outputs = self.teacher(
                 input_ids,
                 attention_mask=src_mask,
-                encoder_outputs=teacher_enc_outputs,
+                encoder_outputs=(enc_outputs,),
                 decoder_input_ids=decoder_input_ids,
                 lm_labels=labels,
                 output_hidden_states=True,
+                return_dict=True,
             )
+            tlogits, tdec_hidden = outputs.logits, outputs.decoder_hidden_states
         dec_mask = decoder_input_ids.ne(pad_token_id)
         loss_ce = self.calc_ce_loss(dec_mask, lm_logits, tlogits)
         if self.alpha_hid > 0:  # Intermediate supervision of decoder hidden states
@@ -180,10 +192,9 @@ class BartSummarizationDistiller(SummarizationModule):
         blended_loss = (
             self.alpha_ce * loss_ce
             + self.alpha_mlm * student_lm_loss
-            + self.hparams.alpha_encoder_loss * loss_encoder
             + self.hparams.alpha_hid * (hid_loss_enc + hid_loss_dec)
         )
-        return blended_loss, loss_ce, student_lm_loss, loss_encoder, hid_loss_enc, hid_loss_dec
+        return blended_loss, loss_ce, student_lm_loss, hid_loss_enc, hid_loss_dec
 
     @staticmethod
     def calc_hidden_loss(attention_mask, hidden_states, hidden_states_T, matches, normalize_hidden):
@@ -207,7 +218,6 @@ def add_distill_args(parser):
     parser.add_argument("--teacher", type=str)
     parser.add_argument("--alpha_ce", default=0.8, type=float)
     parser.add_argument("--alpha_mlm", default=0.2, type=float)
-    parser.add_argument("--alpha_encoder_loss", default=0.0, type=float)
     parser.add_argument("--alpha_hid", default=0.0, type=float, required=False)
     parser.add_argument("--student_decoder_layers", default=12, type=int, required=False)
     parser.add_argument("--student_encoder_layers", default=12, type=int, required=False)
@@ -254,30 +264,6 @@ def create_module(args):
     return model
 
 
-def evaluate_checkpoint(ckpt_path: Path, dest_dir=None):
-    # TODO(SS): DELETE? Better to convert_pl_ckpt_to_hf and run_eval.py
-    exp_dir = ckpt_path.parent
-    if dest_dir is None:
-        dest_dir = exp_dir
-    clash = list(dest_dir.glob("test_generations*"))
-    if clash:
-        print(f"SKIPPING to avoid overwriting {clash}")
-    ckpt = torch.load(ckpt_path, map_location="cpu")
-    if "hparams" in ckpt:
-        args = argparse.Namespace(**ckpt["hparams"])
-    else:
-        args = argparse.Namespace(**pickle_load(exp_dir / "hparams.pkl"))
-    args.resume_from_checkpoint = str(ckpt_path)
-    args.do_train = False
-    args.output_dir = str(dest_dir)
-    args.n_gpu = 1
-    args.eval_batch_size = 16
-    Path(args.output_dir).mkdir(exist_ok=True)
-    model = create_module(args)
-    trainer: pl.Trainer = generic_train(model, args, early_stopping_callback=False)
-    trainer.test(model)
-
-
 def distill_main(args):
     Path(args.output_dir).mkdir(exist_ok=True)
     if len(os.listdir(args.output_dir)) > 3 and args.do_train:

examples/seq2seq/finetune.py

@@ -33,6 +33,7 @@ from utils import (
     lmap,
     pickle_save,
     save_git_info,
+    save_json,
     use_task_specific_params,
 )
 
@@ -105,6 +106,7 @@ class SummarizationModule(BaseTransformer):
         self.dataset_class = (
             Seq2SeqDataset if hasattr(self.tokenizer, "prepare_seq2seq_batch") else LegacySeq2SeqDataset
         )
+        self.already_saved_batch = False
         self.eval_beams = self.model.config.num_beams if self.hparams.eval_beams is None else self.hparams.eval_beams
         if self.hparams.eval_max_gen_length is not None:
             self.eval_max_length = self.hparams.eval_max_gen_length
@@ -112,6 +114,17 @@ class SummarizationModule(BaseTransformer):
             self.eval_max_length = self.model.config.max_length
         self.val_metric = self.default_val_metric if self.hparams.val_metric is None else self.hparams.val_metric
 
+    def save_readable_batch(self, batch: Dict[str, torch.Tensor]) -> Dict[str, List[str]]:
+        """A debugging utility"""
+        readable_batch = {
+            k: self.tokenizer.batch_decode(v.tolist()) if "mask" not in k else v.shape for k, v in batch.items()
+        }
+        save_json(readable_batch, Path(self.output_dir) / "text_batch.json")
+        save_json({k: v.tolist() for k, v in batch.items()}, Path(self.output_dir) / "tok_batch.json")
+
+        self.already_saved_batch = True
+        return readable_batch
+
     def forward(self, input_ids, **kwargs):
         return self.model(input_ids, **kwargs)
 
@@ -129,6 +142,9 @@ class SummarizationModule(BaseTransformer):
             decoder_input_ids = self.model._shift_right(tgt_ids)
         else:
             decoder_input_ids = shift_tokens_right(tgt_ids, pad_token_id)
+        if not self.already_saved_batch:  # This would be slightly better if it only happened on rank zero
+            batch["decoder_input_ids"] = decoder_input_ids
+            self.save_readable_batch(batch)
 
         outputs = self(src_ids, attention_mask=src_mask, decoder_input_ids=decoder_input_ids, use_cache=False)
         lm_logits = outputs[0]
@@ -165,6 +181,7 @@ class SummarizationModule(BaseTransformer):
         return self._generative_step(batch)
 
     def validation_epoch_end(self, outputs, prefix="val") -> Dict:
+
         self.step_count += 1
         losses = {k: torch.stack([x[k] for x in outputs]).mean() for k in self.loss_names}
         loss = losses["loss"]

examples/seq2seq/finetune_bart_tiny.sh

@@ -1,7 +1,7 @@
 # Script for verifying that run_bart_sum can be invoked from its directory
 
 # Get tiny dataset with cnn_dm format (4 examples for train, val, test)
-wget https://s3.amazonaws.com/datasets.huggingface.co/summarization/cnn_tiny.tgz
+wget https://cdn-datasets.huggingface.co/summarization/cnn_tiny.tgz
 tar -xzvf cnn_tiny.tgz
 rm cnn_tiny.tgz
 

examples/seq2seq/finetune_trainer.py

@@ -4,7 +4,7 @@ import sys
 from dataclasses import dataclass, field
 from typing import Optional
 
-from seq2seq_trainer import Seq2SeqTrainer
+from seq2seq_trainer import Seq2SeqTrainer, arg_to_scheduler_choices
 from transformers import (
     AutoConfig,
     AutoModelForSeq2SeqLM,
@@ -63,6 +63,9 @@ class Seq2SeqTrainingArguments(TrainingArguments):
     attention_dropout: Optional[float] = field(
         default=None, metadata={"help": "Attention dropout probability. Goes into model.config."}
     )
+    lr_scheduler: Optional[str] = field(
+        default="linear", metadata={"help": f"Which lr scheduler to use. Selected in {arg_to_scheduler_choices}"}
+    )
 
 
 @dataclass

examples/seq2seq/make_student.py

@@ -13,7 +13,7 @@ logger = logging.get_logger(__name__)
 
 
 def copy_layers(src_layers: nn.ModuleList, dest_layers: nn.ModuleList, layers_to_copy: List[int]) -> None:
-    layers_to_copy = nn.ModuleList([l for i, l in enumerate(src_layers) if i in layers_to_copy])
+    layers_to_copy = nn.ModuleList([src_layers[i] for i in layers_to_copy])
     assert len(dest_layers) == len(layers_to_copy), f"{len(dest_layers)} != {len(layers_to_copy)}"
     dest_layers.load_state_dict(layers_to_copy.state_dict())
 
@@ -81,6 +81,8 @@ def create_student_by_copying_alternating_layers(
     e: Union[int, None] = None,
     d: Union[int, None] = None,
     copy_first_teacher_layers=False,
+    e_layers_to_copy=None,
+    d_layers_to_copy=None,
     **extra_config_kwargs
 ) -> Tuple[PreTrainedModel, List[int], List[int]]:
     """Make a student by copying alternating layers from a teacher, save it to save_path.
@@ -142,8 +144,10 @@ def create_student_by_copying_alternating_layers(
         return student, e_layers_to_copy, d_layers_to_copy
 
     # Decide which layers of the teacher to copy. Not exactly alternating -- we try to keep first and last layer.
-    e_layers_to_copy: List[int] = pick_layers_to_copy(e, teacher_e)
-    d_layers_to_copy: List[int] = pick_layers_to_copy(d, teacher_d)
+    if e_layers_to_copy is None:
+        e_layers_to_copy: List[int] = pick_layers_to_copy(e, teacher_e)
+    if d_layers_to_copy is None:
+        d_layers_to_copy: List[int] = pick_layers_to_copy(d, teacher_d)
 
     try:
         copy_layers(teacher.model.encoder.layers, student.model.encoder.layers, e_layers_to_copy)

examples/seq2seq/precomputed_pseudo_labels.md

@@ -0,0 +1,43 @@
+### Saved Pseudo-Labels
+These are the generations of various large models on various large **training** sets. All in all they took about 200 GPU hours to produce.
+
+### Available Pseudo-labels
+| Dataset | Model                       | Link                                                                                   | Rouge Scores       | Notes                                                                                                       
+|---------|-----------------------------|----------------------------------------------------------------------------------------|--------------------|-------------------------------------------------------------------------------------------------------------
+| XSUM    | `facebook/bart-large-xsum`    | [download](https://cdn-datasets.huggingface.co/pseudo/xsum/bart_xsum_pl.tgz)          | 49.8/28.0/42.5     |                                                                                                             
+| XSUM    | `google/pegasus-xsum`         | [download](https://cdn-datasets.huggingface.co/pseudo/xsum/pegasus_xsum.tgz)          | 53.3/32.7/46.5     |                                                                                                             
+| XSUM    | `facebook/bart-large-xsum`    | [download](https://cdn-datasets.huggingface.co/pseudo/xsum/xsum_pl2_bart.tgz)         |                   | Bart pseudolabels filtered to those with Rouge2 > 10.0 w GT.                                                 
+| CNN/DM  | `sshleifer/pegasus-cnn-ft-v2` | [download](https://cdn-datasets.huggingface.co/pseudo/cnn_dm/pegasus_cnn_cnn_pls.tgz) | 47.316/26.65/44.56 | do not worry about the fact that train.source is one line shorter.                                          
+| CNN/DM  | `facebook/bart-large-cnn`     | [download](https://cdn-datasets.huggingface.co/pseudo/cnn_dm/cnn_bart_pl.tgz)         |                    | 5K (2%) are missing, there should be 282173                                                                 
+| CNN/DM  | `google/pegasus-xsum`         | [download](https://cdn-datasets.huggingface.co/pseudo/cnn_dm/pegasus_xsum_on_cnn.tgz) | 21.5/6.76/25       | extra labels for xsum distillation  Used max_source_length=512, (and all other pegasus-xsum configuration). 
+| EN-RO   | `Helsinki-NLP/opus-mt-en-ro`  | [download](https://cdn-datasets.huggingface.co/pseudo/wmt_en_ro/opus_mt_en_ro.tgz) |       |  
+| EN-RO   | `facebook/mbart-large-en-ro`  | [download](https://cdn-datasets.huggingface.co/pseudo/wmt_en_ro/mbart_large_en_ro.tgz) |       |  
+
+
+(EN_RO = WMT 2016 English-Romanian).
+
+Example Download Command:
+```bash
+curl -S https://cdn-datasets.huggingface.co/pseudo/xsum/bart_xsum_pl.tgz | tar -xvz -C .
+```
+### Generating New Pseudolabels
+Here is the command I used to generate the pseudolabels in the second row of the table, after downloading XSUM from [here](https://cdn-datasets.huggingface.co/summarization/xsum.tar.gz). 
+
+```bash                                                                         
+python -m torch.distributed.launch --nproc_per_node=8 run_distributed_eval.py \
+    --model_name google/pegasus-xsum \ 
+    --save_dir pegasus_xsum \ 
+    --data_dir xsum \
+    --bs 8 --sync_timeout 60000 \
+    --max_source_length 512 \
+    --type_path train
+```
+
++ These command takes a while to run. For example,  pegasus_cnn_cnn_pls.tgz took 8 hours on 8 GPUs.
++ Pegasus does not work in fp16 :(, Bart, mBART and Marian do.
++ Even if you have 1 GPU, `run_distributed_eval.py` is 10-20% faster than `run_eval.py` because it uses `SortishSampler` to minimize padding computation.
+
+### Contributions
+Feel free to contribute your own pseudolabels via PR. Add a row to this table with a new google drive link (or other command line downloadable link).
+
+

examples/seq2seq/seq2seq_trainer.py

@@ -8,8 +8,17 @@ from torch.utils.data import DistributedSampler, RandomSampler
 from transformers import Trainer
 from transformers.configuration_fsmt import FSMTConfig
 from transformers.file_utils import is_torch_tpu_available
-from transformers.optimization import Adafactor, AdamW, get_linear_schedule_with_warmup
-from transformers.trainer import get_tpu_sampler
+from transformers.optimization import (
+    Adafactor,
+    AdamW,
+    get_constant_schedule,
+    get_constant_schedule_with_warmup,
+    get_cosine_schedule_with_warmup,
+    get_cosine_with_hard_restarts_schedule_with_warmup,
+    get_linear_schedule_with_warmup,
+    get_polynomial_decay_schedule_with_warmup,
+)
+from transformers.trainer_pt_utils import get_tpu_sampler
 
 
 try:
@@ -20,6 +29,16 @@ except ImportError:
 
 logger = logging.getLogger(__name__)
 
+arg_to_scheduler = {
+    "linear": get_linear_schedule_with_warmup,
+    "cosine": get_cosine_schedule_with_warmup,
+    "cosine_w_restarts": get_cosine_with_hard_restarts_schedule_with_warmup,
+    "polynomial": get_polynomial_decay_schedule_with_warmup,
+    "constant": get_constant_schedule,
+    "constant_w_warmup": get_constant_schedule_with_warmup,
+}
+arg_to_scheduler_choices = sorted(arg_to_scheduler.keys())
+
 
 class Seq2SeqTrainer(Trainer):
     def __init__(self, config, data_args, *args, **kwargs):
@@ -62,9 +81,21 @@ class Seq2SeqTrainer(Trainer):
                 )
 
         if self.lr_scheduler is None:
-            self.lr_scheduler = get_linear_schedule_with_warmup(
+            self.lr_scheduler = self._get_lr_scheduler(num_training_steps)
+        else:  # ignoring --lr_scheduler
+            logger.warn("scheduler is passed to `Seq2SeqTrainer`, `--lr_scheduler` arg is ignored.")
+
+    def _get_lr_scheduler(self, num_training_steps):
+        schedule_func = arg_to_scheduler[self.args.lr_scheduler]
+        if self.args.lr_scheduler == "constant":
+            scheduler = schedule_func(self.optimizer)
+        elif self.args.lr_scheduler == "constant_w_warmup":
+            scheduler = schedule_func(self.optimizer, num_warmup_steps=self.args.warmup_steps)
+        else:
+            scheduler = schedule_func(
                 self.optimizer, num_warmup_steps=self.args.warmup_steps, num_training_steps=num_training_steps
             )
+        return scheduler
 
     def _get_train_sampler(self) -> Optional[torch.utils.data.sampler.Sampler]:
         if isinstance(self.train_dataset, torch.utils.data.IterableDataset):

examples/seq2seq/test_finetune_trainer.py

@@ -4,7 +4,8 @@ import tempfile
 from unittest.mock import patch
 
 from transformers.testing_utils import slow
-from transformers.trainer_utils import TrainerState, set_seed
+from transformers.trainer_callback import TrainerState
+from transformers.trainer_utils import set_seed
 
 from .finetune_trainer import main
 from .test_seq2seq_examples import MBART_TINY

examples/seq2seq/test_seq2seq_examples.py

@@ -13,7 +13,7 @@ import torch
 
 import lightning_base
 from convert_pl_checkpoint_to_hf import convert_pl_to_hf
-from distillation import distill_main, evaluate_checkpoint
+from distillation import distill_main
 from finetune import SummarizationModule, main
 from run_eval import generate_summaries_or_translations, run_generate
 from run_eval_search import run_search
@@ -86,7 +86,6 @@ CHEAP_ARGS = {
     "n_val": -1,
     "n_test": -1,
     "student_encoder_layers": 1,
-    "alpha_encoder_loss": 0.0,
     "freeze_encoder": False,
     "auto_scale_batch_size": False,
 }
@@ -179,7 +178,6 @@ class TestSummarizationDistiller(unittest.TestCase):
         generate_summaries_or_translations(examples, out_path, str(model.output_dir / "best_tfmr"))
         self.assertTrue(Path(out_path).exists())
 
-        evaluate_checkpoint(ckpts[0], dest_dir=Path(tempfile.mkdtemp()))
         out_path_new = tempfile.mkdtemp()
         convert_pl_to_hf(ckpts[0], transformer_ckpts[0].parent, out_path_new)
         assert os.path.exists(os.path.join(out_path_new, "pytorch_model.bin"))
@@ -228,9 +226,6 @@ class TestSummarizationDistiller(unittest.TestCase):
         assert len(all_files) > 2
         self.assertEqual(len(transformer_ckpts), 2)
 
-        evaluate_checkpoint(ckpts[0], dest_dir=Path(tempfile.mkdtemp()))
-
-    @unittest.skip("T5 distillation is broken at the moment")
     def test_distill_t5(self):
         updates = dict(
             student_encoder_layers=1,
@@ -255,7 +250,6 @@ class TestSummarizationDistiller(unittest.TestCase):
             model_name_or_path="sshleifer/tinier_bart",
             teacher=CHEAP_ARGS["model_name_or_path"],
             val_check_interval=0.5,
-            alpha_encoder_loss=0.4,
         )
         default_updates.update(updates)
         args_d: dict = CHEAP_ARGS.copy()
@@ -425,6 +419,10 @@ def test_finetune(model):
         assert bart.decoder.embed_tokens == bart.encoder.embed_tokens
         assert bart.decoder.embed_tokens == bart.shared
 
+    example_batch = load_json(module.output_dir / "text_batch.json")
+    assert isinstance(example_batch, dict)
+    assert len(example_batch) >= 4
+
 
 def test_finetune_extra_model_args():
     args_d: dict = CHEAP_ARGS.copy()

examples/seq2seq/test_tatoeba_conversion.py

@@ -0,0 +1,22 @@
+import tempfile
+import unittest
+
+from transformers.convert_marian_tatoeba_to_pytorch import TatoebaConverter
+from transformers.file_utils import cached_property
+from transformers.testing_utils import slow
+
+
+class TatoebaConversionTester(unittest.TestCase):
+    @cached_property
+    def resolver(self):
+        tmp_dir = tempfile.mkdtemp()
+        return TatoebaConverter(save_dir=tmp_dir)
+
+    @slow
+    def test_resolver(self):
+        self.resolver.convert_models(["heb-eng"])
+
+    @slow
+    def test_model_card(self):
+        content, mmeta = self.resolver.write_model_card("opus-mt-he-en", dry_run=True)
+        assert mmeta["long_pair"] == "heb-eng"

examples/test_examples.py

@@ -116,8 +116,8 @@ class ExamplesTests(TestCasePlus):
             testargs.append("--fp16")
 
         with patch.object(sys, "argv", testargs):
-            result = run_pl_glue.main()
-            # for now just testing that the script can run to a completion
+            result = run_pl_glue.main()[0]
+            # for now just testing that the script can run to completion
             self.assertGreater(result["acc"], 0.25)
             #
             # TODO: this fails on CI - doesn't get acc/f1>=0.75:

examples/text-classification/run_tf_text_classification.py

@@ -60,7 +60,7 @@ def get_tfds(
         for k in files.keys():
             transformed_ds[k] = ds[k].map(
                 lambda example: tokenizer.batch_encode_plus(
-                    (example[features_name[0]], features_name[1]),
+                    (example[features_name[0]], example[features_name[1]]),
                     truncation=True,
                     max_length=max_seq_length,
                     padding="max_length",
@@ -96,6 +96,9 @@ def get_tfds(
         else None
     )
 
+    if train_ds is not None:
+        train_ds = train_ds.apply(tf.data.experimental.assert_cardinality(len(ds[datasets.Split.TRAIN])))
+
     val_ds = (
         tf.data.Dataset.from_generator(
             gen_val,
@@ -106,6 +109,9 @@ def get_tfds(
         else None
     )
 
+    if val_ds is not None:
+        val_ds = val_ds.apply(tf.data.experimental.assert_cardinality(len(ds[datasets.Split.VALIDATION])))
+
     test_ds = (
         tf.data.Dataset.from_generator(
             gen_test,
@@ -116,6 +122,9 @@ def get_tfds(
         else None
     )
 
+    if test_ds is not None:
+        test_ds = test_ds.apply(tf.data.experimental.assert_cardinality(len(ds[datasets.Split.TEST])))
+
     return train_ds, val_ds, test_ds, label2id
 
 

model_cards/Rostlab/prot_t5_xl_bfd/README.md

@@ -0,0 +1,125 @@
+---
+language: protein
+tags:
+- protein language model
+datasets:
+- BFD
+---
+
+# ProtT5-XL-BFD model
+
+Pretrained model on protein sequences using a masked language modeling (MLM) objective. It was introduced in
+[this paper](https://doi.org/10.1101/2020.07.12.199554) and first released in
+[this repository](https://github.com/agemagician/ProtTrans). This model is trained on uppercase amino acids: it only works with capital letter amino acids.
+
+
+## Model description
+
+ProtT5-XL-BFD is based on the `t5-3b` model and was pretrained on a large corpus of protein sequences in a self-supervised fashion.
+This means it was pretrained on the raw protein sequences only, with no humans labelling them in any way (which is why it can use lots of
+publicly available data) with an automatic process to generate inputs and labels from those protein sequences.
+
+One important difference between this T5 model and the original T5 version is the denosing objective.
+The original T5-3B model was pretrained using a span denosing objective, while this model was pre-trained with a Bart-like MLM denosing objective.
+The masking probability is consistent with the original T5 training by randomly masking 15% of the amino acids in the input.
+
+It has been shown that the features extracted from this self-supervised model (LM-embeddings) captured important biophysical properties governing protein shape.
+shape.
+This implied learning some of the grammar of the language of life realized in protein sequences.
+
+## Intended uses & limitations
+
+The model could be used for protein feature extraction or to be fine-tuned on downstream tasks.
+We have noticed in some tasks on can gain more accuracy by fine-tuning the model rather than using it as a feature extractor.
+We have also noticed that for feature extraction, its better to use the feature extracted from the encoder not from the decoder.
+
+### How to use
+
+Here is how to use this model to extract the features of a given protein sequence in PyTorch:
+
+```python
+from transformers import T5Tokenizer, T5Model
+import re
+import torch
+
+tokenizer = T5Tokenizer.from_pretrained('Rostlab/prot_t5_xl_bfd', do_lower_case=False)
+
+model = T5Model.from_pretrained("Rostlab/prot_t5_xl_bfd")
+
+sequences_Example = ["A E T C Z A O","S K T Z P"]
+
+sequences_Example = [re.sub(r"[UZOB]", "X", sequence) for sequence in sequences_Example]
+
+ids = tokenizer.batch_encode_plus(sequences_Example, add_special_tokens=True, padding=True)
+
+input_ids = torch.tensor(ids['input_ids'])
+attention_mask = torch.tensor(ids['attention_mask'])
+
+with torch.no_grad():
+    embedding = model(input_ids=input_ids,attention_mask=attention_mask,decoder_input_ids=None)
+
+# For feature extraction we recommend to use the encoder embedding
+encoder_embedding = embedding[2].cpu().numpy()
+decoder_embedding = embedding[0].cpu().numpy()
+```
+
+## Training data
+
+The ProtT5-XL-BFD model was pretrained on [BFD](https://bfd.mmseqs.com/), a dataset consisting of 2.1 billion protein sequences.
+
+## Training procedure
+
+### Preprocessing
+
+The protein sequences are uppercased and tokenized using a single space and a vocabulary size of 21. The rare amino acids "U,Z,O,B" were mapped to "X".
+The inputs of the model are then of the form:
+
+```
+Protein Sequence [EOS]
+```
+
+The preprocessing step was performed on the fly, by cutting and padding the protein sequences up to 512 tokens.
+
+The details of the masking procedure for each sequence are as follows:
+- 15% of the amino acids are masked.
+- In 90% of the cases, the masked amino acids are replaced by `[MASK]` token.
+- In 10% of the cases, the masked amino acids are replaced by a random amino acid (different) from the one they replace.
+
+### Pretraining
+
+The model was trained on a single TPU Pod V3-1024 for 1.2 million steps in total, using sequence length 512 (batch size 4k).
+It has a total of approximately 3B parameters and was trained using the encoder-decoder architecture.
+The optimizer used is AdaFactor with inverse square root learning rate schedule for pre-training.
+
+
+## Evaluation results
+
+When the model is used for feature etraction, this model achieves the following results:
+
+Test results :
+
+| Task/Dataset | secondary structure (3-states) | secondary structure (8-states)  |  Localization | Membrane  |
+|:-----:|:-----:|:-----:|:-----:|:-----:|
+|   CASP12  | 77 | 66 |    |    |
+|   TS115   | 85 | 74 |    |    | 
+|   CB513   | 84 | 71 |    |    |
+|  DeepLoc  |    |    | 77 | 91 |
+
+### BibTeX entry and citation info
+
+```bibtex
+@article {Elnaggar2020.07.12.199554,
+	author = {Elnaggar, Ahmed and Heinzinger, Michael and Dallago, Christian and Rehawi, Ghalia and Wang, Yu and Jones, Llion and Gibbs, Tom and Feher, Tamas and Angerer, Christoph and Steinegger, Martin and BHOWMIK, DEBSINDHU and Rost, Burkhard},
+	title = {ProtTrans: Towards Cracking the Language of Life{\textquoteright}s Code Through Self-Supervised Deep Learning and High Performance Computing},
+	elocation-id = {2020.07.12.199554},
+	year = {2020},
+	doi = {10.1101/2020.07.12.199554},
+	publisher = {Cold Spring Harbor Laboratory},
+	abstract = {Computational biology and bioinformatics provide vast data gold-mines from protein sequences, ideal for Language Models (LMs) taken from Natural Language Processing (NLP). These LMs reach for new prediction frontiers at low inference costs. Here, we trained two auto-regressive language models (Transformer-XL, XLNet) and two auto-encoder models (Bert, Albert) on data from UniRef and BFD containing up to 393 billion amino acids (words) from 2.1 billion protein sequences (22- and 112 times the entire English Wikipedia). The LMs were trained on the Summit supercomputer at Oak Ridge National Laboratory (ORNL), using 936 nodes (total 5616 GPUs) and one TPU Pod (V3-512 or V3-1024). We validated the advantage of up-scaling LMs to larger models supported by bigger data by predicting secondary structure (3-states: Q3=76-84, 8 states: Q8=65-73), sub-cellular localization for 10 cellular compartments (Q10=74) and whether a protein is membrane-bound or water-soluble (Q2=89). Dimensionality reduction revealed that the LM-embeddings from unlabeled data (only protein sequences) captured important biophysical properties governing protein shape. This implied learning some of the grammar of the language of life realized in protein sequences. The successful up-scaling of protein LMs through HPC to larger data sets slightly reduced the gap between models trained on evolutionary information and LMs. Availability ProtTrans: \<a href="https://github.com/agemagician/ProtTrans"\>https://github.com/agemagician/ProtTrans\</a\>Competing Interest StatementThe authors have declared no competing interest.},
+	URL = {https://www.biorxiv.org/content/early/2020/07/21/2020.07.12.199554},
+	eprint = {https://www.biorxiv.org/content/early/2020/07/21/2020.07.12.199554.full.pdf},
+	journal = {bioRxiv}
+}
+```
+
+> Created by [Ahmed Elnaggar/@Elnaggar_AI](https://twitter.com/Elnaggar_AI) | [LinkedIn](https://www.linkedin.com/in/prof-ahmed-elnaggar/)

model_cards/abhilash1910/french-roberta/README.md

@@ -0,0 +1,131 @@
+# Roberta Trained Model For Masked Language Model On French Corpus :robot:
+
+
+This is a Masked Language Model trained with [Roberta](https://huggingface.co/transformers/model_doc/roberta.html) on a small French News Corpus(Leipzig corpora).
+The model is built using Huggingface transformers.
+The model can be found at :[French-Roberta](https://huggingface.co/abhilash1910/french-roberta)
+
+
+## Specifications
+
+
+The corpus for training is taken from Leipzig Corpora (French News) , and is trained on a small set of the corpus (300K). 
+
+
+## Model Specification
+
+
+The model chosen for training is [Roberta](https://arxiv.org/abs/1907.11692) with the following specifications:
+ 1. vocab_size=32000
+ 2. max_position_embeddings=514
+ 3. num_attention_heads=12
+ 4. num_hidden_layers=6
+ 5. type_vocab_size=1
+
+
+This is trained by using  RobertaConfig from transformers package.The total training parameters :68124416
+The model is trained for 100 epochs with a gpu batch size of 64 units. 
+More details for building custom models can be found at the [HuggingFace Blog](https://huggingface.co/blog/how-to-train)
+
+
+
+## Usage Specifications
+
+
+For using this model, we have to first import AutoTokenizer and AutoModelWithLMHead Modules from transformers
+After that we have to specify, the pre-trained model,which in this case is 'abhilash1910/french-roberta' for the tokenizers and the model.
+
+
+```python
+from transformers import AutoTokenizer, AutoModelWithLMHead
+
+tokenizer = AutoTokenizer.from_pretrained("abhilash1910/french-roberta")
+
+model = AutoModelWithLMHead.from_pretrained("abhilash1910/french-roberta")
+```
+
+
+After this the model will be downloaded, it will take some time to download all the model files.
+For testing the model, we have to import  pipeline module from transformers and create a masked output model for inference as follows:
+
+
+```python
+from transformers import pipeline
+model_mask = pipeline('fill-mask', model='abhilash1910/french-roberta')
+model_mask("Le tweet <mask>.")
+```
+
+
+Some of the examples are also provided with generic French sentences:
+
+Example 1:
+
+
+```python
+model_mask("À ce jour, <mask> projet a entraîné")
+```
+
+
+Output:
+
+
+```bash
+[{'sequence': '<s>À ce jour, belles projet a entraîné</s>',
+  'score': 0.18685665726661682,
+  'token': 6504,
+  'token_str': 'Ġbelles'},
+ {'sequence': '<s>À ce jour,- projet a entraîné</s>',
+  'score': 0.0005200508167035878,
+  'token': 17,
+  'token_str': '-'},
+ {'sequence': '<s>À ce jour, de projet a entraîné</s>',
+  'score': 0.00045729897101409733,
+  'token': 268,
+  'token_str': 'Ġde'},
+ {'sequence': '<s>À ce jour, du projet a entraîné</s>',
+  'score': 0.0004307595663703978,
+  'token': 326,
+  'token_str': 'Ġdu'},
+ {'sequence': '<s>À ce jour," projet a entraîné</s>',
+  'score': 0.0004219160182401538,
+  'token': 6,
+  'token_str': '"'}]
+  ```
+ 
+ Example 2:
+ 
+```python
+ model_mask("C'est un <mask>")
+```
+
+Output:
+
+```bash
+[{'sequence': "<s>C'est un belles</s>",
+  'score': 0.16440927982330322,
+  'token': 6504,
+  'token_str': 'Ġbelles'},
+ {'sequence': "<s>C'est un de</s>",
+  'score': 0.0005495127406902611,
+  'token': 268,
+  'token_str': 'Ġde'},
+ {'sequence': "<s>C'est un du</s>",
+  'score': 0.00044988933950662613,
+  'token': 326,
+  'token_str': 'Ġdu'},
+ {'sequence': "<s>C'est un-</s>",
+  'score': 0.00044542422983795404,
+  'token': 17,
+  'token_str': '-'},
+ {'sequence': "<s>C'est un\t</s>",
+  'score': 0.00037563967634923756,
+  'token': 202,
+  'token_str': 'ĉ'}]
+  ```
+  
+
+## Resources
+
+For all resources , please look into the [HuggingFace](https://huggingface.co/) Site and the [Repositories](https://github.com/huggingface).
+
+

model_cards/akhooli/xlm-r-large-arabic-toxic/README.md

@@ -0,0 +1,12 @@
+---
+
+language:
+- ar
+- en
+
+license: mit
+---
+### xlm-r-large-arabic-toxic (toxic/hate speech classifier) 
+Toxic (hate speech) classification (Label_0: non-toxic, Label_1: toxic) of Arabic comments by fine-tuning XLM-Roberta-Large. 
+Zero shot classification of other languages (also works in mixed languages - ex. Arabic & English).  
+Usage and further info: see last section in this [Colab notebook](https://lnkd.in/d3bCFyZ)

model_cards/amine/bert-base-5lang-cased/README.md

@@ -0,0 +1,64 @@
+---
+language: 
+- en
+- fr
+- es
+- de
+- zh
+
+tags:
+- pytorch
+- bert
+- multilingual
+- en
+- fr
+- es
+- de
+- zh
+
+datasets: wikipedia
+
+license: apache-2.0
+
+inference: false
+---
+
+# bert-base-5lang-cased
+This is a smaller version of [bert-base-multilingual-cased](https://huggingface.co/bert-base-multilingual-cased) that handles only 5 languages (en, fr, es, de and zh) instead of 104.
+The model is therefore 30% smaller than the original one (124M parameters instead of 178M) but gives exactly the same representations for the above cited languages. 
+Starting from `bert-base-5lang-cased` will facilitate the deployment of your model on public cloud platforms while keeping similar results. 
+For instance, Google Cloud Platform requires that the model size on disk should be lower than 500 MB for serveless deployments (Cloud Functions / Cloud ML) which is not the case of the original `bert-base-multilingual-cased`.
+
+For more information about the models size, memory footprint and loading time please refer to the table below:
+
+|            Model             | Num parameters |   Size   |  Memory  | Loading time |
+| ---------------------------- | -------------- | -------- | -------- | ------------ |
+| bert-base-multilingual-cased |   178 million  |  714 MB  | 1400 MB  |    4.2 sec   |
+| bert-base-5lang-cased        |   124 million  |  495 MB  |  950 MB  |    3.6 sec   |
+
+These measurements have been computed on a [Google Cloud n1-standard-1 machine (1 vCPU, 3.75 GB)](https://cloud.google.com/compute/docs/machine-types\#n1_machine_type).
+
+## How to use
+
+```python
+from transformers import AutoTokenizer, AutoModel
+
+tokenizer = AutoTokenizer.from_pretrained("amine/bert-base-5lang-cased")
+model = AutoModel.from_pretrained("amine/bert-base-5lang-cased")
+
+```
+
+### How to cite
+
+```bibtex
+@inproceedings{smallermbert,
+  title={Load What You Need: Smaller Versions of Mutlilingual BERT},
+  author={Abdaoui, Amine and Pradel, Camille and Sigel, Grégoire},
+  booktitle={SustaiNLP / EMNLP},
+  year={2020}
+}
+```
+
+## Contact 
+
+Please contact amine@geotrend.fr for any question, feedback or request.

model_cards/cedpsam/chatbot_fr/README.md

@@ -13,6 +13,9 @@ trained with this notebook
 https://colab.research.google.com/drive/1pfCV3bngAmISNZVfDvBMyEhQKuYw37Rl#scrollTo=AyImj9qZYLRi&uniqifier=3
 
 config from microsoft/DialoGPT-medium
+dataset generated from 2018 opensubtitle from opus folowing these guidelines
+https://github.com/PolyAI-LDN/conversational-datasets/tree/master/opensubtitles with this notebook
+https://colab.research.google.com/drive/1uyh3vJ9nEjqOHI68VD73qxt4olJzODxi#scrollTo=deaacv4XfLMk
 ### How to use
 
 Now we are ready to try out how the model works as a chatting partner!

model_cards/cimm-kzn/rudr-bert/README.md

@@ -8,16 +8,22 @@ RuDR-BERT - Multilingual, Cased, which pretrained on the raw part of the RuDReC
 
 If you find this repository helpful, feel free to cite our publication:
 
-[1] https://arxiv.org/abs/2004.03659
+[1] Tutubalina E, Alimova I, Miftahutdinov Z, et al. The Russian Drug Reaction Corpus and Neural Models for Drug Reactions and Effectiveness Detection in User Reviews. 
+   
+   preprint: https://arxiv.org/abs/2004.03659
 ```
-@misc{tutubalina2020russian,
-    title={The Russian Drug Reaction Corpus and Neural Models for Drug Reactions and Effectiveness Detection in User Reviews},
-    author={Elena Tutubalina and Ilseyar Alimova and Zulfat Miftahutdinov and Andrey Sakhovskiy and Valentin Malykh and Sergey Nikolenko},
-    year={2020},
-    eprint={2004.03659},
-    archivePrefix={arXiv},
-    primaryClass={cs.CL}
-}
+@article{10.1093/bioinformatics/btaa675,
+    author = {Tutubalina, Elena and Alimova, Ilseyar and Miftahutdinov, Zulfat and Sakhovskiy, Andrey and Malykh, Valentin and Nikolenko, Sergey},
+    title = "{The Russian Drug Reaction Corpus and Neural Models for Drug Reactions and Effectiveness Detection in User Reviews}",
+    journal = {Bioinformatics},
+    year = {2020},
+    month = {07},
+    issn = {1367-4803},
+    doi = {10.1093/bioinformatics/btaa675},
+    url = {https://doi.org/10.1093/bioinformatics/btaa675},
+    note = {btaa675},
+    eprint = {https://academic.oup.com/bioinformatics/advance-article-pdf/doi/10.1093/bioinformatics/btaa675/33539752/btaa675.pdf},
+} 
 ```
 [2] Tutubalina, EV and Miftahutdinov, Z Sh and Nugmanov, RI and Madzhidov, TI and Nikolenko, SI and Alimova, IS and Tropsha, AE Using semantic analysis of texts for the identification of drugs with similar therapeutic effects.
    [link to paper](https://www.researchgate.net/profile/Elena_Tutubalina/publication/323751823_Using_semantic_analysis_of_texts_for_the_identification_of_drugs_with_similar_therapeutic_effects/links/5bf7cfc3299bf1a0202cbc1f/Using-semantic-analysis-of-texts-for-the-identification-of-drugs-with-similar-therapeutic-effects.pdf)

model_cards/distilbert-base-cased-README.md

@@ -0,0 +1,3 @@
+---
+license: apache-2.0
+---

model_cards/distilbert-base-cased-distilled-squad-README.md

@@ -4,4 +4,5 @@ datasets:
 - squad
 metrics:
 - squad
+license: apache-2.0
 ---

model_cards/distilbert-base-german-cased-README.md

@@ -1,4 +1,5 @@
 ---
 language: de
+license: apache-2.0
 ---
 ## distilbert-base-german-cased

model_cards/distilbert-base-uncased-distilled-squad-README.md

@@ -6,4 +6,5 @@ widget:
   context: "The Amazon rainforest (Portuguese: Floresta Amazônica or Amazônia; Spanish: Selva Amazónica, Amazonía or usually Amazonia; French: Forêt amazonienne; Dutch: Amazoneregenwoud), also known in English as Amazonia or the Amazon Jungle, is a moist broadleaf forest that covers most of the Amazon basin of South America. This basin encompasses 7,000,000 square kilometres (2,700,000 sq mi), of which 5,500,000 square kilometres (2,100,000 sq mi) are covered by the rainforest. This region includes territory belonging to nine nations. The majority of the forest is contained within Brazil, with 60% of the rainforest, followed by Peru with 13%, Colombia with 10%, and with minor amounts in Venezuela, Ecuador, Bolivia, Guyana, Suriname and French Guiana. States or departments in four nations contain \"Amazonas\" in their names. The Amazon represents over half of the planet's remaining rainforests, and comprises the largest and most biodiverse tract of tropical rainforest in the world, with an estimated 390 billion individual trees divided into 16,000 species."
 - text: "How many square kilometers of rainforest is covered in the basin?"
   context: "The Amazon rainforest (Portuguese: Floresta Amazônica or Amazônia; Spanish: Selva Amazónica, Amazonía or usually Amazonia; French: Forêt amazonienne; Dutch: Amazoneregenwoud), also known in English as Amazonia or the Amazon Jungle, is a moist broadleaf forest that covers most of the Amazon basin of South America. This basin encompasses 7,000,000 square kilometres (2,700,000 sq mi), of which 5,500,000 square kilometres (2,100,000 sq mi) are covered by the rainforest. This region includes territory belonging to nine nations. The majority of the forest is contained within Brazil, with 60% of the rainforest, followed by Peru with 13%, Colombia with 10%, and with minor amounts in Venezuela, Ecuador, Bolivia, Guyana, Suriname and French Guiana. States or departments in four nations contain \"Amazonas\" in their names. The Amazon represents over half of the planet's remaining rainforests, and comprises the largest and most biodiverse tract of tropical rainforest in the world, with an estimated 390 billion individual trees divided into 16,000 species."
+license: apache-2.0
 ---

model_cards/distilbert-base-uncased-finetuned-sst-2-english-README.md

@@ -0,0 +1,3 @@
+---
+license: apache-2.0
+---

model_cards/facebook/bart-large-mnli/README.md

@@ -0,0 +1,3 @@
+---
+license: mit
+---

model_cards/jcblaise/bert-tagalog-base-cased-WWM/README.md

@@ -0,0 +1,62 @@
+---
+language: tl
+tags:
+- bert
+- tagalog
+- filipino
+license: gpl-3.0
+inference: false
+---
+
+# BERT Tagalog Base Cased (Whole Word Masking)
+Tagalog version of BERT trained on a large preprocessed text corpus scraped and sourced from the internet. This model is part of a larger research project. We open-source the model to allow greater usage within the Filipino NLP community. This particular version uses whole word masking.
+
+## Usage
+The model can be loaded and used in both PyTorch and TensorFlow through the HuggingFace Transformers package.
+
+```python
+from transformers import TFAutoModel, AutoModel, AutoTokenizer
+
+# TensorFlow
+model = TFAutoModel.from_pretrained('jcblaise/bert-tagalog-base-cased-WWM', from_pt=True)
+tokenizer = AutoTokenizer.from_pretrained('jcblaise/bert-tagalog-base-cased-WWM', do_lower_case=False)
+
+# PyTorch
+model = AutoModel.from_pretrained('jcblaise/bert-tagalog-base-cased-WWM')
+tokenizer = AutoTokenizer.from_pretrained('jcblaise/bert-tagalog-base-cased-WWM', do_lower_case=False)
+```
+Finetuning scripts and other utilities we use for our projects can be found in our centralized repository at https://github.com/jcblaisecruz02/Filipino-Text-Benchmarks
+
+## Citations
+All model details and training setups can be found in our papers. If you use our model or find it useful in your projects, please cite our work:
+
+```
+@inproceedings{localization2020cruz,
+  title={{Localization of Fake News Detection via Multitask Transfer Learning}},
+  author={Cruz, Jan Christian Blaise and Tan, Julianne Agatha and Cheng, Charibeth},
+  booktitle={Proceedings of The 12th Language Resources and Evaluation Conference},
+  pages={2589--2597},
+  year={2020},
+  url={https://www.aclweb.org/anthology/2020.lrec-1.315}
+}
+
+@article{cruz2020establishing,
+  title={Establishing Baselines for Text Classification in Low-Resource Languages},
+  author={Cruz, Jan Christian Blaise and Cheng, Charibeth},
+  journal={arXiv preprint arXiv:2005.02068},
+  year={2020}
+}
+
+@article{cruz2019evaluating,
+  title={Evaluating Language Model Finetuning Techniques for Low-resource Languages},
+  author={Cruz, Jan Christian Blaise and Cheng, Charibeth},
+  journal={arXiv preprint arXiv:1907.00409},
+  year={2019}
+}
+```
+
+## Data and Other Resources
+Data used to train this model as well as other benchmark datasets in Filipino can be found in my website at https://blaisecruz.com
+
+## Contact
+If you have questions, concerns, or if you just want to chat about NLP and low-resource languages in general, you may reach me through my work email at jan_christian_cruz@dlsu.edu.ph

model_cards/jcblaise/bert-tagalog-base-cased/README.md

@@ -0,0 +1,62 @@
+---
+language: tl
+tags:
+- bert
+- tagalog
+- filipino
+license: gpl-3.0
+inference: false
+---
+
+# BERT Tagalog Base Cased
+Tagalog version of BERT trained on a large preprocessed text corpus scraped and sourced from the internet. This model is part of a larger research project. We open-source the model to allow greater usage within the Filipino NLP community.
+
+## Usage
+The model can be loaded and used in both PyTorch and TensorFlow through the HuggingFace Transformers package.
+
+```python
+from transformers import TFAutoModel, AutoModel, AutoTokenizer
+
+# TensorFlow
+model = TFAutoModel.from_pretrained('jcblaise/bert-tagalog-base-cased', from_pt=True)
+tokenizer = AutoTokenizer.from_pretrained('jcblaise/bert-tagalog-base-cased', do_lower_case=False)
+
+# PyTorch
+model = AutoModel.from_pretrained('jcblaise/bert-tagalog-base-cased')
+tokenizer = AutoTokenizer.from_pretrained('jcblaise/bert-tagalog-base-cased', do_lower_case=False)
+```
+Finetuning scripts and other utilities we use for our projects can be found in our centralized repository at https://github.com/jcblaisecruz02/Filipino-Text-Benchmarks
+
+## Citations
+All model details and training setups can be found in our papers. If you use our model or find it useful in your projects, please cite our work:
+
+```
+@inproceedings{localization2020cruz,
+  title={{Localization of Fake News Detection via Multitask Transfer Learning}},
+  author={Cruz, Jan Christian Blaise and Tan, Julianne Agatha and Cheng, Charibeth},
+  booktitle={Proceedings of The 12th Language Resources and Evaluation Conference},
+  pages={2589--2597},
+  year={2020},
+  url={https://www.aclweb.org/anthology/2020.lrec-1.315}
+}
+
+@article{cruz2020establishing,
+  title={Establishing Baselines for Text Classification in Low-Resource Languages},
+  author={Cruz, Jan Christian Blaise and Cheng, Charibeth},
+  journal={arXiv preprint arXiv:2005.02068},
+  year={2020}
+}
+
+@article{cruz2019evaluating,
+  title={Evaluating Language Model Finetuning Techniques for Low-resource Languages},
+  author={Cruz, Jan Christian Blaise and Cheng, Charibeth},
+  journal={arXiv preprint arXiv:1907.00409},
+  year={2019}
+}
+```
+
+## Data and Other Resources
+Data used to train this model as well as other benchmark datasets in Filipino can be found in my website at https://blaisecruz.com
+
+## Contact
+If you have questions, concerns, or if you just want to chat about NLP and low-resource languages in general, you may reach me through my work email at jan_christian_cruz@dlsu.edu.ph

model_cards/jcblaise/bert-tagalog-base-uncased-WWM/README.md

@@ -0,0 +1,62 @@
+---
+language: tl
+tags:
+- bert
+- tagalog
+- filipino
+license: gpl-3.0
+inference: false
+---
+
+# BERT Tagalog Base Uncased (Whole Word Masking)
+Tagalog version of BERT trained on a large preprocessed text corpus scraped and sourced from the internet. This model is part of a larger research project. We open-source the model to allow greater usage within the Filipino NLP community. This particular version uses whole word masking.
+
+## Usage
+The model can be loaded and used in both PyTorch and TensorFlow through the HuggingFace Transformers package.
+
+```python
+from transformers import TFAutoModel, AutoModel, AutoTokenizer
+
+# TensorFlow
+model = TFAutoModel.from_pretrained('jcblaise/bert-tagalog-base-uncased-WWM', from_pt=True)
+tokenizer = AutoTokenizer.from_pretrained('jcblaise/bert-tagalog-base-uncased-WWM', do_lower_case=True)
+
+# PyTorch
+model = AutoModel.from_pretrained('jcblaise/bert-tagalog-base-uncased-WWM')
+tokenizer = AutoTokenizer.from_pretrained('jcblaise/bert-tagalog-base-uncased-WWM', do_lower_case=True)
+```
+Finetuning scripts and other utilities we use for our projects can be found in our centralized repository at https://github.com/jcblaisecruz02/Filipino-Text-Benchmarks
+
+## Citations
+All model details and training setups can be found in our papers. If you use our model or find it useful in your projects, please cite our work:
+
+```
+@inproceedings{localization2020cruz,
+  title={{Localization of Fake News Detection via Multitask Transfer Learning}},
+  author={Cruz, Jan Christian Blaise and Tan, Julianne Agatha and Cheng, Charibeth},
+  booktitle={Proceedings of The 12th Language Resources and Evaluation Conference},
+  pages={2589--2597},
+  year={2020},
+  url={https://www.aclweb.org/anthology/2020.lrec-1.315}
+}
+
+@article{cruz2020establishing,
+  title={Establishing Baselines for Text Classification in Low-Resource Languages},
+  author={Cruz, Jan Christian Blaise and Cheng, Charibeth},
+  journal={arXiv preprint arXiv:2005.02068},
+  year={2020}
+}
+
+@article{cruz2019evaluating,
+  title={Evaluating Language Model Finetuning Techniques for Low-resource Languages},
+  author={Cruz, Jan Christian Blaise and Cheng, Charibeth},
+  journal={arXiv preprint arXiv:1907.00409},
+  year={2019}
+}
+```
+
+## Data and Other Resources
+Data used to train this model as well as other benchmark datasets in Filipino can be found in my website at https://blaisecruz.com
+
+## Contact
+If you have questions, concerns, or if you just want to chat about NLP and low-resource languages in general, you may reach me through my work email at jan_christian_cruz@dlsu.edu.ph

model_cards/jcblaise/bert-tagalog-base-uncased/README.md

@@ -0,0 +1,62 @@
+---
+language: tl
+tags:
+- bert
+- tagalog
+- filipino
+license: gpl-3.0
+inference: false
+---
+
+# BERT Tagalog Base Uncased
+Tagalog version of BERT trained on a large preprocessed text corpus scraped and sourced from the internet. This model is part of a larger research project. We open-source the model to allow greater usage within the Filipino NLP community.
+
+## Usage
+The model can be loaded and used in both PyTorch and TensorFlow through the HuggingFace Transformers package.
+
+```python
+from transformers import TFAutoModel, AutoModel, AutoTokenizer
+
+# TensorFlow
+model = TFAutoModel.from_pretrained('jcblaise/bert-tagalog-base-uncased', from_pt=True)
+tokenizer = AutoTokenizer.from_pretrained('jcblaise/bert-tagalog-base-uncased', do_lower_case=True)
+
+# PyTorch
+model = AutoModel.from_pretrained('jcblaise/bert-tagalog-base-uncased')
+tokenizer = AutoTokenizer.from_pretrained('jcblaise/bert-tagalog-base-uncased', do_lower_case=True)
+```
+Finetuning scripts and other utilities we use for our projects can be found in our centralized repository at https://github.com/jcblaisecruz02/Filipino-Text-Benchmarks
+
+## Citations
+All model details and training setups can be found in our papers. If you use our model or find it useful in your projects, please cite our work:
+
+```
+@inproceedings{localization2020cruz,
+  title={{Localization of Fake News Detection via Multitask Transfer Learning}},
+  author={Cruz, Jan Christian Blaise and Tan, Julianne Agatha and Cheng, Charibeth},
+  booktitle={Proceedings of The 12th Language Resources and Evaluation Conference},
+  pages={2589--2597},
+  year={2020},
+  url={https://www.aclweb.org/anthology/2020.lrec-1.315}
+}
+
+@article{cruz2020establishing,
+  title={Establishing Baselines for Text Classification in Low-Resource Languages},
+  author={Cruz, Jan Christian Blaise and Cheng, Charibeth},
+  journal={arXiv preprint arXiv:2005.02068},
+  year={2020}
+}
+
+@article{cruz2019evaluating,
+  title={Evaluating Language Model Finetuning Techniques for Low-resource Languages},
+  author={Cruz, Jan Christian Blaise and Cheng, Charibeth},
+  journal={arXiv preprint arXiv:1907.00409},
+  year={2019}
+}
+```
+
+## Data and Other Resources
+Data used to train this model as well as other benchmark datasets in Filipino can be found in my website at https://blaisecruz.com
+
+## Contact
+If you have questions, concerns, or if you just want to chat about NLP and low-resource languages in general, you may reach me through my work email at jan_christian_cruz@dlsu.edu.ph

model_cards/jcblaise/distilbert-tagalog-base-cased/README.md

@@ -0,0 +1,63 @@
+---
+language: tl
+tags:
+- distilbert
+- bert
+- tagalog
+- filipino
+license: gpl-3.0
+inference: false
+---
+
+# DistilBERT Tagalog Base Cased
+Tagalog version of DistilBERT, distilled from [`bert-tagalog-base-cased`](https://huggingface.co/jcblaise/bert-tagalog-base-cased). This model is part of a larger research project. We open-source the model to allow greater usage within the Filipino NLP community.
+
+## Usage
+The model can be loaded and used in both PyTorch and TensorFlow through the HuggingFace Transformers package.
+
+```python
+from transformers import TFAutoModel, AutoModel, AutoTokenizer
+
+# TensorFlow
+model = TFAutoModel.from_pretrained('jcblaise/distilbert-tagalog-base-cased', from_pt=True)
+tokenizer = AutoTokenizer.from_pretrained('jcblaise/distilbert-tagalog-base-cased', do_lower_case=False)
+
+# PyTorch
+model = AutoModel.from_pretrained('jcblaise/distilbert-tagalog-base-cased')
+tokenizer = AutoTokenizer.from_pretrained('jcblaise/distilbert-tagalog-base-cased', do_lower_case=False)
+```
+Finetuning scripts and other utilities we use for our projects can be found in our centralized repository at https://github.com/jcblaisecruz02/Filipino-Text-Benchmarks
+
+## Citations
+All model details and training setups can be found in our papers. If you use our model or find it useful in your projects, please cite our work:
+
+```
+@inproceedings{localization2020cruz,
+  title={{Localization of Fake News Detection via Multitask Transfer Learning}},
+  author={Cruz, Jan Christian Blaise and Tan, Julianne Agatha and Cheng, Charibeth},
+  booktitle={Proceedings of The 12th Language Resources and Evaluation Conference},
+  pages={2589--2597},
+  year={2020},
+  url={https://www.aclweb.org/anthology/2020.lrec-1.315}
+}
+
+@article{cruz2020establishing,
+  title={Establishing Baselines for Text Classification in Low-Resource Languages},
+  author={Cruz, Jan Christian Blaise and Cheng, Charibeth},
+  journal={arXiv preprint arXiv:2005.02068},
+  year={2020}
+}
+
+@article{cruz2019evaluating,
+  title={Evaluating Language Model Finetuning Techniques for Low-resource Languages},
+  author={Cruz, Jan Christian Blaise and Cheng, Charibeth},
+  journal={arXiv preprint arXiv:1907.00409},
+  year={2019}
+}
+```
+
+## Data and Other Resources
+Data used to train this model as well as other benchmark datasets in Filipino can be found in my website at https://blaisecruz.com
+
+## Contact
+If you have questions, concerns, or if you just want to chat about NLP and low-resource languages in general, you may reach me through my work email at jan_christian_cruz@dlsu.edu.ph

model_cards/joeddav/xlm-roberta-large-xnli/README.md

@@ -9,6 +9,7 @@ datasets:
 - xnli
 widget:
 - text: "За кого вы голосуете в 2020 году? <sep> This text is about politique."
+license: mit
 ---
 
 # xlm-roberta-large-xnli

model_cards/keshan/SinhalaBERTo/README.md

@@ -0,0 +1,37 @@
+---
+language: si
+tags:
+- SinhalaBERTo
+- Sinhala
+- roberta
+datasets:
+- oscar
+---
+### Overview
+
+This is a slightly smaller model trained on [OSCAR](https://oscar-corpus.com/) Sinhala dedup dataset. As Sinhala is one of those low resource languages, there are only a handful of models been trained. So, this would be a great place to start training for more downstream tasks. 
+
+## Model Specification
+
+
+The model chosen for training is [Roberta](https://arxiv.org/abs/1907.11692) with the following specifications:
+ 1. vocab_size=52000
+ 2. max_position_embeddings=514
+ 3. num_attention_heads=12
+ 4. num_hidden_layers=6
+ 5. type_vocab_size=1
+ 
+## How to Use
+You can use this model directly with a pipeline for masked language modeling:
+
+```py
+from transformers import AutoTokenizer, AutoModelWithLMHead, pipeline
+
+model = BertForMaskedLM.from_pretrained("keshan/SinhalaBERTo")
+tokenizer = BertTokenizer.from_pretrained("keshan/SinhalaBERTo")
+
+fill_mask = pipeline('fill-mask', model=model, tokenizer=tokenizer)
+
+fill_mask("මම ගෙදර <mask>.")
+
+```

model_cards/neurocode/IsRoBERTa/README.md

@@ -0,0 +1,74 @@
+---
+language: is
+datasets:
+- Icelandic portion of the OSCAR corpus from INRIA
+- oscar
+---
+
+# IsRoBERTa a RoBERTa-like masked language model
+
+Probably the first icelandic transformer language model!
+
+## Overview
+**Language:** Icelandic  
+**Downstream-task:** masked-lm 
+**Training data:** OSCAR corpus 
+**Code:**  See [here](https://github.com/neurocode-io/icelandic-language-model)
+**Infrastructure**: 1x Nvidia K80
+
+## Hyperparameters
+
+```
+per_device_train_batch_size = 48
+n_epochs = 1
+vocab_size = 52.000
+max_position_embeddings = 514
+num_attention_heads = 12
+num_hidden_layers = 6
+type_vocab_size = 1
+learning_rate=0.00005
+``` 
+
+
+## Usage
+
+### In Transformers
+```python
+from transformers import (
+  pipeline,
+  AutoTokenizer,
+  AutoModelWithLMHead
+)
+
+model_name = "neurocode/IsRoBERTa"
+
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+model = AutoModelWithLMHead.from_pretrained(model_name)
+>>> fill_mask = pipeline(
+...     "fill-mask",
+...     model=model,
+...     tokenizer=tokenizer
+... )
+>>> result = fill_mask("Hann fór út að <mask>.")
+>>> result
+[
+  {'sequence': '<s>Hann fór út að nýju.</s>', 'score': 0.03395755589008331, 'token': 2219, 'token_str': 'Ġnýju'},
+  {'sequence': '<s>Hann fór út að undanförnu.</s>', 'score': 0.029087543487548828, 'token': 7590, 'token_str': 'Ġundanförnu'},
+  {'sequence': '<s>Hann fór út að lokum.</s>', 'score': 0.024420788511633873, 'token': 4384, 'token_str': 'Ġlokum'},
+  {'sequence': '<s>Hann fór út að þessu.</s>', 'score': 0.021231256425380707, 'token': 921, 'token_str': 'Ġþessu'},
+  {'sequence': '<s>Hann fór út að honum.</s>', 'score': 0.0205782949924469, 'token': 1136, 'token_str': 'Ġhonum'}
+]
+```
+
+
+## Authors
+Bobby Donchev: `contact [at] donchev.is`
+Elena Cramer: `elena.cramer [at] neurocode.io`
+
+## About us
+
+We bring AI software for our customers live
+Our focus: AI software development
+ 
+Get in touch:
+[LinkedIn](https://de.linkedin.com/company/neurocodeio) | [Website](https://neurocode.io)

model_cards/nikokons/gpt2-greek/README.md

@@ -0,0 +1,5 @@
+---
+language: el
+---
+
+## gpt2-greek

model_cards/nlpaueb/legal-bert-base-uncased/README.md

@@ -0,0 +1,101 @@
+---
+language: en
+tags:
+- legal
+---
+
+# LEGAL-BERT: The Muppets straight out of Law School
+
+<img align="left" src="https://i.ibb.co/p3kQ7Rw/Screenshot-2020-10-06-at-12-16-36-PM.png" width="100"/> 
+
+LEGAL-BERT is a family of BERT models for the legal domain, intended to assist legal NLP research, computational law, and legal technology applications.  To pre-train the different variations of LEGAL-BERT, we collected 12 GB of diverse English legal text from several fields (e.g., legislation, court cases,  contracts) scraped from publicly available resources. Sub-domains variants (CONTRACTS-, EURLEX-, ECHR-) and/or general LEGAL-BERT perform better than using BERT out of the box for domain-specific tasks. A light-weight model (33% the size of BERT-BASE) pre-trained from scratch on legal data with competitive perfomance is also available.
+<br/><br/><br/><br/>
+
+---
+
+I. Chalkidis, M. Fergadiotis, P. Malakasiotis, N. Aletras and I. Androutsopoulos. "LEGAL-BERT: The Muppets straight out of Law School". In Findings of Empirical Methods in Natural Language Processing (EMNLP 2020) (Short Papers), to be held online, 2020. (https://arxiv.org/abs/2010.02559)
+
+---
+
+## Pre-training corpora
+
+The pre-training corpora of LEGAL-BERT include:
+
+* 116,062 documents of EU legislation, publicly available from EURLEX (http://eur-lex.europa.eu), the repository of EU Law running under the EU Publication Office.
+    
+* 61,826 documents of UK legislation, publicly available from the UK legislation portal (http://www.legislation.gov.uk).
+    
+* 19,867 cases from European Court of Justice (ECJ), also available from EURLEX.
+    
+* 12,554 cases from HUDOC, the repository of the European Court of Human Rights (ECHR) (http://hudoc.echr.coe.int/eng).
+    
+* 164,141 cases from various courts across the USA, hosted in the Case Law Access Project portal (https://case.law).
+    
+* 76,366 US contracts from EDGAR, the database of US Securities and Exchange Commission (SECOM) (https://www.sec.gov/edgar.shtml).
+
+## Pre-training details
+
+* We trained BERT using the official code provided in Google BERT's github repository (https://github.com/google-research/bert).
+* We released a model similar to the English BERT-BASE model (12-layer, 768-hidden, 12-heads, 110M parameters).
+* We chose to follow the same training set-up: 1 million training steps with batches of 256 sequences of length 512 with an initial learning rate 1e-4.
+* We were able to use a single Google Cloud TPU v3-8 provided for free from [TensorFlow Research Cloud (TFRC)](https://www.tensorflow.org/tfrc), while also utilizing [GCP research credits](https://edu.google.com/programs/credits/research). Huge thanks to both Google programs for supporting us!
+* Part of LEGAL-BERT is a light-weight model pre-trained from scratch on legal data, which achieves comparable performance to larger models, while being much more efficient (approximately 4 times faster) with a smaller environmental footprint.
+## Models list
+
+| Model name          | Model Path                            | Training corpora    |
+| ------------------- | ------------------------------------  | ------------------- |
+| CONTRACTS-BERT-BASE | `nlpaueb/bert-base-uncased-contracts` | US contracts        |
+| EURLEX-BERT-BASE    | `nlpaueb/bert-base-uncased-eurlex`    | EU legislation      |
+| ECHR-BERT-BASE      | `nlpaueb/bert-base-uncased-echr`      | ECHR cases          |
+| LEGAL-BERT-BASE     | `nlpaueb/legal-bert-base-uncased`     | All                 |
+| LEGAL-BERT-SMALL    | `nlpaueb/legal-bert-small-uncased`    | All                 |
+
+## Load Pretrained Model
+
+```python
+from transformers import AutoTokenizer, AutoModel
+
+tokenizer = AutoTokenizer.from_pretrained("nlpaueb/legal-bert-base-uncased")
+model = AutoModel.from_pretrained("nlpaueb/legal-bert-base-uncased")
+```
+
+## Use LEBAL-BERT variants as Language Models
+
+| Corpus                             | Model                               | Masked token | Predictions  |
+| --------------------------------- | ---------------------------------- | ------------ | ------------ |
+|  | **BERT-BASE-UNCASED**                 |
+| (Contracts) | This [MASK] Agreement is between General Motors and John Murray . | employment | ('new', '0.09'), ('current', '0.04'), ('proposed', '0.03'), ('marketing', '0.03'), ('joint', '0.02')
+| (ECHR) | The applicant submitted that her husband was subjected to treatment amounting to [MASK] whilst in the custody of Adana Security Directorate | torture | ('torture', '0.32'), ('rape', '0.22'), ('abuse', '0.14'), ('death', '0.04'), ('violence', '0.03')
+| (EURLEX) | Establishing a system for the identification and registration of [MASK] animals and regarding the labelling of beef and beef products . | bovine | ('farm', '0.25'), ('livestock', '0.08'), ('draft', '0.06'), ('domestic', '0.05'), ('wild', '0.05')
+|  | **CONTRACTS-BERT-BASE**                 |
+| (Contracts) | This [MASK] Agreement is between General Motors and John Murray . | employment | ('letter', '0.38'), ('dealer', '0.04'), ('employment', '0.03'), ('award', '0.03'), ('contribution', '0.02')
+| (ECHR) | The applicant submitted that her husband was subjected to treatment amounting to [MASK] whilst in the custody of Adana Security Directorate | torture | ('death', '0.39'), ('imprisonment', '0.07'), ('contempt', '0.05'), ('being', '0.03'), ('crime', '0.02')
+| (EURLEX) | Establishing a system for the identification and registration of [MASK] animals and regarding the labelling of beef and beef products . | bovine | (('domestic', '0.18'), ('laboratory', '0.07'), ('household', '0.06'), ('personal', '0.06'), ('the', '0.04')
+|  | **EURLEX-BERT-BASE**                 |
+| (Contracts) | This [MASK] Agreement is between General Motors and John Murray . | employment | ('supply', '0.11'), ('cooperation', '0.08'), ('service', '0.07'), ('licence', '0.07'), ('distribution', '0.05')
+| (ECHR) | The applicant submitted that her husband was subjected to treatment amounting to [MASK] whilst in the custody of Adana Security Directorate | torture | ('torture', '0.66'), ('death', '0.07'), ('imprisonment', '0.07'), ('murder', '0.04'), ('rape', '0.02')
+| (EURLEX) | Establishing a system for the identification and registration of [MASK] animals and regarding the labelling of beef and beef products . | bovine | ('live', '0.43'), ('pet', '0.28'), ('certain', '0.05'), ('fur', '0.03'), ('the', '0.02')
+|  | **ECHR-BERT-BASE**                 |
+| (Contracts) | This [MASK] Agreement is between General Motors and John Murray . | employment | ('second', '0.24'), ('latter', '0.10'), ('draft', '0.05'), ('bilateral', '0.05'), ('arbitration', '0.04')
+| (ECHR) | The applicant submitted that her husband was subjected to treatment amounting to [MASK] whilst in the custody of Adana Security Directorate | torture | ('torture', '0.99'), ('death', '0.01'), ('inhuman', '0.00'), ('beating', '0.00'), ('rape', '0.00')
+| (EURLEX) | Establishing a system for the identification and registration of [MASK] animals and regarding the labelling of beef and beef products . | bovine | ('pet', '0.17'), ('all', '0.12'), ('slaughtered', '0.10'), ('domestic', '0.07'), ('individual', '0.05')
+|  | **LEGAL-BERT-BASE**                |
+| (Contracts) | This [MASK] Agreement is between General Motors and John Murray . | employment | ('settlement', '0.26'), ('letter', '0.23'), ('dealer', '0.04'), ('master', '0.02'), ('supplemental', '0.02')
+| (ECHR) | The applicant submitted that her husband was subjected to treatment amounting to [MASK] whilst in the custody of Adana Security Directorate | torture | ('torture', '1.00'), ('detention', '0.00'), ('arrest', '0.00'), ('rape', '0.00'), ('death', '0.00')
+| (EURLEX) | Establishing a system for the identification and registration of [MASK] animals and regarding the labelling of beef and beef products . | bovine | ('live', '0.67'), ('beef', '0.17'), ('farm', '0.03'), ('pet', '0.02'), ('dairy', '0.01')
+|  | **LEGAL-BERT-SMALL**                |
+| (Contracts) | This [MASK] Agreement is between General Motors and John Murray . | employment | ('license', '0.09'), ('transition', '0.08'), ('settlement', '0.04'), ('consent', '0.03'), ('letter', '0.03')
+| (ECHR) | The applicant submitted that her husband was subjected to treatment amounting to [MASK] whilst in the custody of Adana Security Directorate | torture | ('torture', '0.59'), ('pain', '0.05'), ('ptsd', '0.05'), ('death', '0.02'), ('tuberculosis', '0.02')
+| (EURLEX) | Establishing a system for the identification and registration of [MASK] animals and regarding the labelling of beef and beef products . | bovine | ('all', '0.08'), ('live', '0.07'), ('certain', '0.07'), ('the', '0.07'), ('farm', '0.05')
+
+
+
+## Evaluation on downstream tasks
+
+Consider the experiments in the article "LEGAL-BERT: The Muppets straight out of Law School". Chalkidis et al., 2018, (https://arxiv.org/abs/2010.02559)
+
+## Author
+
+Ilias Chalkidis on behalf of [AUEB's Natural Language Processing Group](http://nlp.cs.aueb.gr)
+
+| Github: [@ilias.chalkidis](https://github.com/seolhokim) | Twitter: [@KiddoThe2B](https://twitter.com/KiddoThe2B) |

model_cards/nlptown/bert-base-multilingual-uncased-sentiment/README.md

@@ -6,6 +6,8 @@ language:
 - fr
 - it
 - es
+
+license: mit
 ---
 
 # bert-base-multilingual-uncased-sentiment

notebooks/05-benchmark.ipynb

@@ -1658,7 +1658,7 @@
               " 'add_final_layer_norm': False,\n",
               " 'attention_dropout': 0.0,\n",
               " 'bos_token_id': 0,\n",
-              " 'classif_dropout': 0.0,\n",
+              " 'classifier_dropout': 0.0,\n",
               " 'd_model': 1024,\n",
               " 'decoder_attention_heads': 16,\n",
               " 'decoder_ffn_dim': 4096,\n",

scripts/pegasus/build_test_sample_spm_no_bos.py

@@ -0,0 +1,20 @@
+#!/usr/bin/env python
+
+# this script builds a small sample spm file tests/fixtures/test_sentencepiece_no_bos.model, with features needed by pegasus 
+
+# 1. pip install sentencepiece
+# 
+# 2. wget https://raw.githubusercontent.com/google/sentencepiece/master/data/botchan.txt
+
+# 3. build
+import sentencepiece as spm
+
+# pegasus:
+# 1. no bos
+# 2. eos_id is 1
+# 3. unk_id is 2
+# build a sample spm file accordingly
+spm.SentencePieceTrainer.train('--input=botchan.txt --model_prefix=test_sentencepiece_no_bos --bos_id=-1 --unk_id=2  --eos_id=1  --vocab_size=1000')
+
+# 4. now update the fixture
+# mv test_sentencepiece_no_bos.model ../../tests/fixtures/

scripts/tatoeba/README.md

@@ -0,0 +1,44 @@
+Setup transformers following instructions in README.md, (I would fork first).
+```bash
+git clone git@github.com:huggingface/transformers.git
+cd transformers
+pip install -e .
+pip install pandas
+```
+
+Get required metadata
+```
+curl https://cdn-datasets.huggingface.co/language_codes/language-codes-3b2.csv  > language-codes-3b2.csv
+curl https://cdn-datasets.huggingface.co/language_codes/iso-639-3.csv > iso-639-3.csv
+```
+
+Install Tatoeba-Challenge repo inside transformers
+```bash
+git clone git@github.com:Helsinki-NLP/Tatoeba-Challenge.git
+```
+
+To convert a few models, call the conversion script from command line:
+```bash
+python src/transformers/convert_marian_tatoeba_to_pytorch.py --models heb-eng eng-heb --save_dir converted
+```
+
+To convert lots of models you can pass your list of Tatoeba model names to `resolver.convert_models` in a python client or script.
+
+```python
+from transformers.convert_marian_tatoeba_to_pytorch import TatoebaConverter
+resolver = TatoebaConverter(save_dir='converted')
+resolver.convert_models(['heb-eng', 'eng-heb'])
+```
+
+
+### Upload converted models
+```bash
+cd converted
+transformers-cli login
+for FILE in *; do transformers-cli upload $FILE; done
+```
+
+
+### Modifications
+- To change naming logic, change the code near `os.rename`. The model card creation code may also need to change.
+- To change model card content, you must modify `TatoebaCodeResolver.write_model_card`

setup.py

@@ -3,7 +3,10 @@ Simple check list from AllenNLP repo: https://github.com/allenai/allennlp/blob/m
 
 To create the package for pypi.
 
-1. Change the version in __init__.py, setup.py as well as docs/source/conf.py.
+1. Change the version in __init__.py, setup.py as well as docs/source/conf.py. Remove the master from the links in
+   the new models of the README:
+   (https://huggingface.co/transformers/master/model_doc/ -> https://huggingface.co/transformers/model_doc/)
+   then run `make fix-copies` to fix the index of the documentation.
 
 2. Unpin specific versions from setup.py that use a git install.
 
@@ -111,7 +114,7 @@ setup(
     packages=find_packages("src"),
     install_requires=[
         "numpy",
-        "tokenizers == 0.9.0.rc1",
+        "tokenizers == 0.9.0.rc2",
         # dataclasses for Python versions that don't have it
         "dataclasses;python_version<'3.7'",
         # utilities from PyPA to e.g. compare versions
@@ -131,9 +134,7 @@ setup(
         "sacremoses",
     ],
     extras_require=extras,
-    entry_points={
-        "console_scripts": ["transformers-cli=transformers.commands.transformers_cli:main"]
-    },
+    entry_points={"console_scripts": ["transformers-cli=transformers.commands.transformers_cli:main"]},
     python_requires=">=3.6.0",
     classifiers=[
         "Development Status :: 5 - Production/Stable",

src/transformers/__init__.py

@@ -33,6 +33,7 @@ from .configuration_auto import ALL_PRETRAINED_CONFIG_ARCHIVE_MAP, CONFIG_MAPPIN
 from .configuration_bart import BartConfig
 from .configuration_bert import BERT_PRETRAINED_CONFIG_ARCHIVE_MAP, BertConfig
 from .configuration_bert_generation import BertGenerationConfig
+from .configuration_blenderbot import BLENDERBOT_PRETRAINED_CONFIG_ARCHIVE_MAP, BlenderbotConfig
 from .configuration_camembert import CAMEMBERT_PRETRAINED_CONFIG_ARCHIVE_MAP, CamembertConfig
 from .configuration_ctrl import CTRL_PRETRAINED_CONFIG_ARCHIVE_MAP, CTRLConfig
 from .configuration_deberta import DEBERTA_PRETRAINED_CONFIG_ARCHIVE_MAP, DebertaConfig
@@ -107,8 +108,6 @@ from .file_utils import (
     is_tf_available,
     is_torch_available,
     is_torch_tpu_available,
-    is_sentencepiece_available,
-    is_tokenizers_available,
 )
 from .hf_argparser import HfArgumentParser
 
@@ -160,6 +159,7 @@ from .tokenization_bert import BasicTokenizer, BertTokenizer, BertTokenizerFast,
 from .tokenization_bert_generation import BertGenerationTokenizer
 from .tokenization_bert_japanese import BertJapaneseTokenizer, CharacterTokenizer, MecabTokenizer
 from .tokenization_bertweet import BertweetTokenizer
+from .tokenization_blenderbot import BlenderbotSmallTokenizer, BlenderbotTokenizer
 from .tokenization_camembert import CamembertTokenizer, CamembertTokenizerFast
 from .tokenization_ctrl import CTRLTokenizer
 from .tokenization_deberta import DebertaTokenizer
@@ -207,7 +207,15 @@ from .tokenization_xlm_roberta import XLMRobertaTokenizer, XLMRobertaTokenizerFa
 from .tokenization_xlnet import SPIECE_UNDERLINE, XLNetTokenizer, XLNetTokenizerFast
 
 # Trainer
-from .trainer_utils import EvalPrediction, TrainerState, set_seed
+from .trainer_callback import (
+    DefaultFlowCallback,
+    PrinterCallback,
+    ProgressCallback,
+    TrainerCallback,
+    TrainerControl,
+    TrainerState,
+)
+from .trainer_utils import EvalPrediction, EvaluationStrategy, set_seed
 from .training_args import TrainingArguments
 from .training_args_tf import TFTrainingArguments
 from .utils import logging
@@ -303,6 +311,7 @@ if is_torch_available():
         BertGenerationEncoder,
         load_tf_weights_in_bert_generation,
     )
+    from .modeling_blenderbot import BLENDERBOT_PRETRAINED_MODEL_ARCHIVE_LIST, BlenderbotForConditionalGeneration
     from .modeling_camembert import (
         CAMEMBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
         CamembertForCausalLM,
@@ -377,6 +386,7 @@ if is_torch_available():
     from .modeling_gpt2 import (
         GPT2_PRETRAINED_MODEL_ARCHIVE_LIST,
         GPT2DoubleHeadsModel,
+        GPT2ForSequenceClassification,
         GPT2LMHeadModel,
         GPT2Model,
         GPT2PreTrainedModel,
@@ -427,6 +437,7 @@ if is_torch_available():
     from .modeling_openai import (
         OPENAI_GPT_PRETRAINED_MODEL_ARCHIVE_LIST,
         OpenAIGPTDoubleHeadsModel,
+        OpenAIGPTForSequenceClassification,
         OpenAIGPTLMHeadModel,
         OpenAIGPTModel,
         OpenAIGPTPreTrainedModel,
@@ -530,7 +541,8 @@ if is_torch_available():
     from .tokenization_marian import MarianTokenizer
 
     # Trainer
-    from .trainer import EvalPrediction, Trainer, set_seed, torch_distributed_zero_first
+    from .trainer import Trainer
+    from .trainer_pt_utils import torch_distributed_zero_first
 else:
     from .utils.dummy_pt_objects import *
 

src/transformers/configuration_auto.py

@@ -21,6 +21,7 @@ from .configuration_albert import ALBERT_PRETRAINED_CONFIG_ARCHIVE_MAP, AlbertCo
 from .configuration_bart import BART_PRETRAINED_CONFIG_ARCHIVE_MAP, BartConfig
 from .configuration_bert import BERT_PRETRAINED_CONFIG_ARCHIVE_MAP, BertConfig
 from .configuration_bert_generation import BertGenerationConfig
+from .configuration_blenderbot import BLENDERBOT_PRETRAINED_CONFIG_ARCHIVE_MAP, BlenderbotConfig
 from .configuration_camembert import CAMEMBERT_PRETRAINED_CONFIG_ARCHIVE_MAP, CamembertConfig
 from .configuration_ctrl import CTRL_PRETRAINED_CONFIG_ARCHIVE_MAP, CTRLConfig
 from .configuration_deberta import DEBERTA_PRETRAINED_CONFIG_ARCHIVE_MAP, DebertaConfig
@@ -58,6 +59,7 @@ ALL_PRETRAINED_CONFIG_ARCHIVE_MAP = dict(
     for pretrained_map in [
         BERT_PRETRAINED_CONFIG_ARCHIVE_MAP,
         BART_PRETRAINED_CONFIG_ARCHIVE_MAP,
+        BLENDERBOT_PRETRAINED_CONFIG_ARCHIVE_MAP,
         MBART_PRETRAINED_CONFIG_ARCHIVE_MAP,
         OPENAI_GPT_PRETRAINED_CONFIG_ARCHIVE_MAP,
         TRANSFO_XL_PRETRAINED_CONFIG_ARCHIVE_MAP,
@@ -101,6 +103,7 @@ CONFIG_MAPPING = OrderedDict(
         ("marian", MarianConfig),
         ("mbart", MBartConfig),
         ("bart", BartConfig),
+        ("blenderbot", BlenderbotConfig),
         ("reformer", ReformerConfig),
         ("longformer", LongformerConfig),
         ("roberta", RobertaConfig),
@@ -136,6 +139,7 @@ MODEL_NAMES_MAPPING = OrderedDict(
         ("camembert", "CamemBERT"),
         ("xlm-roberta", "XLM-RoBERTa"),
         ("pegasus", "Pegasus"),
+        ("blenderbot", "Blenderbot"),
         ("marian", "Marian"),
         ("mbart", "mBART"),
         ("bart", "BART"),

src/transformers/configuration_bart.py

@@ -84,6 +84,8 @@ class BartConfig(PretrainedConfig):
             Don't learn positional embeddings, use sinusoidal.
         add_final_layer_norm (:obj:`bool`, `optional`, defaults to :obj:`False`):
             Why not add another layernorm?
+        do_blenderbot_90_layernorm (:obj:`bool`, `optional`, defaults to :obj:`False`):
+            Blenderbot-90m checkpoint uses `layernorm_embedding` one line earlier in the decoder.
         scale_embedding (:obj:`bool`, `optional`, defaults to :obj:`False`):
             Scale embeddings by diving by sqrt(d_model).
         eos_token_id (:obj:`int`, `optional`, defaults to 2)
@@ -113,7 +115,7 @@ class BartConfig(PretrainedConfig):
     def __init__(
         self,
         activation_dropout=0.0,
-        extra_pos_embeddings=2,  # FIXME(@sshleifer): delete?
+        extra_pos_embeddings=2,
         activation_function="gelu",
         vocab_size=50265,
         d_model=1024,
@@ -137,6 +139,7 @@ class BartConfig(PretrainedConfig):
         eos_token_id=2,
         normalize_before=False,
         add_final_layer_norm=False,
+        do_blenderbot_90_layernorm=False,
         scale_embedding=False,
         normalize_embedding=True,
         static_position_embeddings=False,
@@ -195,13 +198,16 @@ class BartConfig(PretrainedConfig):
         self.dropout = dropout
 
         # Classifier stuff
-        self.classif_dropout = classifier_dropout
+        self.classifier_dropout = classifier_dropout
 
         # pos embedding offset
-        self.extra_pos_embeddings = self.pad_token_id + 1
+        self.extra_pos_embeddings = extra_pos_embeddings
+        # bart has a hack that offsets positional embeddings by 2, other models don't do this
 
         self.force_bos_token_to_be_generated = force_bos_token_to_be_generated
 
+        self.do_blenderbot_90_layernorm = do_blenderbot_90_layernorm
+
     @property
     def num_attention_heads(self) -> int:
         return self.encoder_attention_heads

src/transformers/configuration_blenderbot.py

@@ -0,0 +1,178 @@
+#!/usr/bin/env python3
+# coding=utf-8
+# Copyright (c) Facebook, Inc. and Huggingface, 2020
+#
+# This source code is licensed under the MIT license found in the;
+# 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.
+# LICENSE file in the root directory of this source tree.
+"""BlenderbotConfig has the same signature as BartConfig. We only rewrite the signature in order to document blenderbot-90M defaults."""
+from .configuration_bart import BartConfig
+
+
+BLENDERBOT_PRETRAINED_CONFIG_ARCHIVE_MAP = {
+    "facebook/blenderbot-3B": "https://cdn.huggingface.co/facebook/blenderbot-3B/config.json",
+    "facebook/blenderbot-90M": "https://cdn.huggingface.co/facebook/blenderbot-90M/config.json",
+}
+
+
+class BlenderbotConfig(BartConfig):
+    r"""
+    This is the configuration class to store the configuration of a :class:`~transformers.BlenderbotForConditionalGeneration`.
+    It inherits from :class:`~transformers.BartConfig` and has the same signature with different defaults.
+
+    Configuration objects inherit from  :class:`~transformers.PretrainedConfig` and can be used
+    to control the model outputs. Read the documentation from  :class:`~transformers.PretrainedConfig`
+    for more information.
+
+    Args:
+        vocab_size (:obj:`int`, `optional`, defaults to 54944):
+            Vocabulary size of the BERT model. Defines the number of different tokens that can be represented by the
+            :obj:`inputs_ids` passed when calling :class:`~transformers.BlenderbotForConditionalGeneration`.
+        d_model (:obj:`int`, `optional`, defaults to 512):
+            Dimensionality of the layers and the pooler layer.
+        encoder_layers (:obj:`int`, `optional`, defaults to 8):
+            Number of encoder layers, 6 are used for the `blenderbot-90M` model.
+        decoder_layers (:obj:`int`, `optional`, defaults to 8):
+            Number of decoder layers, 6 are used for the `blenderbot-90M` model.
+        encoder_attention_heads (:obj:`int`, `optional`, defaults to 16):
+            Number of attention heads for each attention layer in the Transformer encoder.
+        decoder_attention_heads (:obj:`int`, `optional`, defaults to 16):
+            Number of attention heads for each attention layer in the Transformer decoder.
+        decoder_ffn_dim (:obj:`int`, `optional`, defaults to 2048):
+            Dimensionality of the "intermediate" (often named feed-forward) layer in decoder.
+        encoder_ffn_dim (:obj:`int`, `optional`, defaults to 2048):
+            Dimensionality of the "intermediate" (often named feed-forward) layer in decoder.
+        activation_function (:obj:`str` or :obj:`function`, `optional`, defaults to :obj:`"gelu"`):
+            The non-linear activation function (function or string) in the encoder and pooler.
+            If string, :obj:`"gelu"`, :obj:`"relu"`, :obj:`"swish"` and :obj:`"gelu_new"` are supported.
+        dropout (:obj:`float`, `optional`, defaults to 0.1):
+            The dropout probabilitiy for all fully connected layers in the embeddings, encoder, and pooler.
+        attention_dropout (:obj:`float`, `optional`, defaults to 0.0):
+            The dropout ratio for the attention probabilities.
+        activation_dropout (:obj:`float`, `optional`, defaults to 0.0):
+            The dropout ratio for activations inside the fully connected layer.
+        classifier_dropout (:obj:`float`, `optional`, defaults to 0.0):
+            The dropout ratio for classifier.
+        max_position_embeddings (:obj:`int`, `optional`, defaults to 512):
+            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).
+        init_std (:obj:`float`, `optional`, defaults to 0.02):
+            The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
+        add_bias_logits (:obj:`bool`, `optional`, defaults to :obj:`False`):
+            This should be completed, specific to marian.
+        normalize_before (:obj:`bool`, `optional`, defaults to :obj:`False`):
+            Call layernorm before attention ops.
+        normalize_embedding (:obj:`bool`, `optional`, defaults to :obj:`True`):
+            Call layernorm after embeddings.
+        static_position_embeddings (:obj:`bool`, `optional`, defaults to :obj:`False`):
+            Don't learn positional embeddings, use sinusoidal.
+        add_final_layer_norm (:obj:`bool`, `optional`, defaults to :obj:`False`):
+            Why not add another layernorm?
+        do_blenderbot_90_layernorm (:obj:`bool`, `optional`, defaults to :obj:`True`):
+            Blenderbot-90m checkpoint uses `layernorm_embedding` one line earlier in the decoder.
+        scale_embedding (:obj:`bool`, `optional`, defaults to :obj:`False`):
+            Scale embeddings by diving by sqrt(d_model).
+        eos_token_id (:obj:`int`, `optional`, defaults to 2)
+            End of stream token id.
+        pad_token_id (:obj:`int`, `optional`, defaults to 1)
+            Padding token id.
+        bos_token_id (:obj:`int`, `optional`, defaults to 0)
+            Beginning of stream token id.
+        encoder_layerdrop: (:obj:`float`, `optional`, defaults to 0.0):
+            The LayerDrop probability for the encoder. See the `LayerDrop paper
+            <see https://arxiv.org/abs/1909.11556>`__ for more details.
+        decoder_layerdrop: (:obj:`float`, `optional`, defaults to 0.0):
+            The LayerDrop probability for the decoder. See the `LayerDrop paper
+            <see https://arxiv.org/abs/1909.11556>`__ for more details.
+        extra_pos_embeddings: (:obj:`int`, `optional`, defaults to 2):
+            How many extra learned positional embeddings to use. Should be set to :obj:`pad_token_id+1`.
+        is_encoder_decoder (:obj:`bool`, `optional`, defaults to :obj:`True`):
+            Whether this is an encoder/decoder model.
+        force_bos_token_to_be_generated (:obj:`bool`, `optional`, defaults to :obj:`False`):
+            Whether or not to force BOS token to be generated at step 1 (after ``decoder_start_token_id``),
+    """
+    model_type = "blenderbot"
+
+    def __init__(
+        self,
+        activation_dropout=0.0,
+        extra_pos_embeddings=0,
+        activation_function="gelu",
+        vocab_size=54944,
+        d_model=512,
+        encoder_ffn_dim=2048,
+        encoder_layers=8,
+        encoder_attention_heads=16,
+        decoder_ffn_dim=2048,
+        decoder_layers=8,
+        decoder_attention_heads=16,
+        encoder_layerdrop=0.0,
+        decoder_layerdrop=0.0,
+        attention_dropout=0.0,
+        dropout=0.1,
+        max_position_embeddings=512,
+        classifier_dropout=0.0,
+        is_encoder_decoder=True,
+        pad_token_id=1,
+        bos_token_id=0,
+        eos_token_id=2,
+        normalize_before=False,
+        add_final_layer_norm=False,
+        do_blenderbot_90_layernorm=True,
+        scale_embedding=False,
+        normalize_embedding=True,
+        static_position_embeddings=False,
+        add_bias_logits=False,
+        force_bos_token_to_be_generated=False,
+        **common_kwargs
+    ):
+        r"""
+        Examples::
+
+            >>> from transformers import BlenderbotConfig
+            >>> config = BlenderbotConfig.from_pretrained('facebook/blenderbot-90M')
+
+        """
+        if "hidden_size" in common_kwargs:
+            raise ValueError("hidden size is called d_model")
+        super().__init__(
+            pad_token_id=pad_token_id,
+            bos_token_id=bos_token_id,
+            eos_token_id=eos_token_id,
+            is_encoder_decoder=is_encoder_decoder,
+            vocab_size=vocab_size,
+            d_model=d_model,
+            encoder_ffn_dim=encoder_ffn_dim,
+            encoder_layers=encoder_layers,
+            encoder_layerdrop=encoder_layerdrop,
+            encoder_attention_heads=encoder_attention_heads,
+            decoder_layerdrop=decoder_layerdrop,
+            decoder_ffn_dim=decoder_ffn_dim,
+            decoder_layers=decoder_layers,
+            normalize_before=normalize_before,
+            normalize_embedding=normalize_embedding,
+            static_position_embeddings=static_position_embeddings,
+            add_bias_logits=add_bias_logits,
+            force_bos_token_to_be_generated=force_bos_token_to_be_generated,
+            do_blenderbot_90_layernorm=do_blenderbot_90_layernorm,
+            add_final_layer_norm=add_final_layer_norm,
+            scale_embedding=scale_embedding,
+            attention_dropout=attention_dropout,
+            dropout=dropout,
+            classifier_dropout=classifier_dropout,
+            activation_dropout=activation_dropout,
+            max_position_embeddings=max_position_embeddings,
+            extra_pos_embeddings=extra_pos_embeddings,
+            activation_function=activation_function,
+            decoder_attention_heads=decoder_attention_heads,
+            **common_kwargs,
+        )

src/transformers/configuration_mmbt.py

@@ -31,7 +31,7 @@ class MMBTConfig(object):
             Config of the underlying Transformer models. Its values are copied over to use a single config.
         num_labels (:obj:`int`, `optional`):
             Size of final Linear layer for classification.
-        modal_hidden_size (:obj:`int`, `optional`, defautls to 2048):
+        modal_hidden_size (:obj:`int`, `optional`, defaults to 2048):
             Embedding dimension of the non-text modality encoder.
     """
 

src/transformers/configuration_squeezebert.py

@@ -39,52 +39,51 @@ class SqueezeBertConfig(PretrainedConfig):
 
 
     Args:
-        vocab_size (:obj:`int`, optional, defaults to 30522):
-            Vocabulary size of the SqueezeBERT model. Defines the different tokens that
-            can be represented by the :obj:`inputs_ids` passed to the forward method of :class:`~transformers.SqueezeBertModel`.
-        hidden_size (:obj:`int`, optional, defaults to 512):
+        vocab_size (:obj:`int`, `optional`, defaults to 30522):
+            Vocabulary size of the SqueezeBERT model. Defines the number of different tokens that can be
+            represented by the :obj:`inputs_ids` passed when calling :class:`~transformers.SqueezeBertModel`.
+        hidden_size (:obj:`int`, `optional`, defaults to 768):
             Dimensionality of the encoder layers and the pooler layer.
-        num_hidden_layers (:obj:`int`, optional, defaults to 24):
+        num_hidden_layers (:obj:`int`, `optional`, defaults to 12):
             Number of hidden layers in the Transformer encoder.
-        num_attention_heads (:obj:`int`, optional, defaults to 4):
+        num_attention_heads (:obj:`int`, `optional`, defaults to 12):
             Number of attention heads for each attention layer in the Transformer encoder.
-        intermediate_size (:obj:`int`, optional, defaults to 512):
-            Dimensionality of the "intermediate" (i.e., feed-forward) layer in the Transformer encoder.
-        hidden_act (:obj:`str` or :obj:`function`, optional, defaults to :obj:`"gelu"`):
+        intermediate_size (:obj:`int`, `optional`, defaults to 3072):
+            Dimensionality of the "intermediate" (often named feed-forward) layer in the Transformer encoder.
+        hidden_act (:obj:`str` or :obj:`Callable`, `optional`, defaults to :obj:`"gelu"`):
             The non-linear activation function (function or string) in the encoder and pooler.
-            If string, :obj:`"gelu"`, :obj:`"relu"`, :obj:`"swish"`, :obj:`"gelu_new"`, :obj:`"gelu_fast"`,
-            and :obj:`"mish"` are supported.
-        hidden_dropout_prob (:obj:`float`, optional, defaults to 0.0):
-            The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
-        attention_probs_dropout_prob (:obj:`float`, optional, defaults to 0.1):
+            If string, :obj:`"gelu"`, :obj:`"relu"`, :obj:`"swish"` and :obj:`"gelu_new"` are supported.
+        hidden_dropout_prob (:obj:`float`, `optional`, defaults to 0.1):
+            The dropout probabilitiy for all fully connected layers in the embeddings, encoder, and pooler.
+        attention_probs_dropout_prob (:obj:`float`, `optional`, defaults to 0.1):
             The dropout ratio for the attention probabilities.
-        max_position_embeddings (:obj:`int`, optional, defaults to 512):
+        max_position_embeddings (:obj:`int`, `optional`, defaults to 512):
             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).
-        type_vocab_size (:obj:`int`, optional, defaults to 2):
-            The vocabulary size of the :obj:`token_type_ids` passed into :class:`~transformers.SqueezeBertModel`.
-        initializer_range (:obj:`float`, optional, defaults to 0.02):
+        type_vocab_size (:obj:`int`, `optional`, defaults to 2):
+            The vocabulary size of the :obj:`token_type_ids` passed when calling :class:`~transformers.BertModel` or
+            :class:`~transformers.TFBertModel`.
+        initializer_range (:obj:`float`, `optional`, defaults to 0.02):
             The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
-        layer_norm_eps (:obj:`float`, optional, defaults to 1e-12):
-            The epsilon used by the layer normalization layers.
+        layer_norm_eps (:obj:`float`, `optional`, defaults to 1e-12):
 
-        pad_token_id (:obj:`int`, optional, defaults to 0):
+        pad_token_id (:obj:`int`, `optional`, defaults to 0):
             The ID of the token in the word embedding to use as padding.
-        embedding_size (:obj:`int`, optional, defaults to 128):
+        embedding_size (:obj:`int`, `optional`, defaults to 768):
             The dimension of the word embedding vectors.
 
-        q_groups (:obj:`int`, optional, defaults to 4):
-            The number of groups in Q layer
-        k_groups (:obj:`int`, optional, defaults to 4):
-            The number of groups in K layer
-        v_groups (:obj:`int`, optional, defaults to 4):
-            The number of groups in V layer
-        post_attention_groups (:obj:`int`, optional, defaults to 1):
-            The number of groups in the first feed forward network layer
-        intermediate_groups (:obj:`int`, optional, defaults to 4):
-            The number of groups in the second feed forward network layer
-        output_groups (:obj:`int`, optional, defaults to 4):
-            The number of groups in the third feed forward network layer
+        q_groups (:obj:`int`, `optional`, defaults to 4):
+            The number of groups in Q layer.
+        k_groups (:obj:`int`, `optional`, defaults to 4):
+            The number of groups in K layer.
+        v_groups (:obj:`int`, `optional`, defaults to 4):
+            The number of groups in V layer.
+        post_attention_groups (:obj:`int`, `optional`, defaults to 1):
+            The number of groups in the first feed forward network layer.
+        intermediate_groups (:obj:`int`, `optional`, defaults to 4):
+            The number of groups in the second feed forward network layer.
+        output_groups (:obj:`int`, `optional`, defaults to 4):
+            The number of groups in the third feed forward network layer.
 
     Example:
 

src/transformers/configuration_utils.py

@@ -274,7 +274,7 @@ class PretrainedConfig(object):
                 Path to a directory in which a downloaded pretrained model configuration should be cached if the
                 standard cache should not be used.
             force_download (:obj:`bool`, `optional`, defaults to :obj:`False`):
-                Wheter or not to force to (re-)download the configuration files and override the cached versions if they
+                Whether or not to force to (re-)download the configuration files and override the cached versions if they
                 exist.
             resume_download (:obj:`bool`, `optional`, defaults to :obj:`False`):
                 Whether or not to delete incompletely received file. Attempts to resume the download if such a file

src/transformers/configuration_xlnet.py

@@ -204,7 +204,7 @@ class XLNetConfig(PretrainedConfig):
         if mem_len is None or mem_len == 0:
             warnings.warn(
                 "This config doesn't use attention memories, a core feature of XLNet."
-                " Consider setting `men_len` to a non-zero value, for example "
+                " Consider setting `mem_len` to a non-zero value, for example "
                 "`xlnet = XLNetLMHeadModel.from_pretrained('xlnet-base-cased'', mem_len=1024)`,"
                 " for accurate training performance as well as an order of magnitude faster inference."
                 " Starting from version 3.5.0, the default parameter will be 1024, following"

src/transformers/convert_blenderbot_original_pytorch_checkpoint_to_pytorch.py

@@ -0,0 +1,114 @@
+# coding=utf-8
+# Copyright 2020 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.
+"""Convert Blenderbot checkpoint."""
+
+import argparse
+
+import torch
+
+from transformers import BartConfig, BartForConditionalGeneration
+from transformers.utils import logging
+
+
+logging.set_verbosity_info()
+logger = logging.get_logger(__name__)
+
+PATTERNS = [
+    ["attention", "attn"],
+    ["encoder_attention", "encoder_attn"],
+    ["q_lin", "q_proj"],
+    ["k_lin", "k_proj"],
+    ["v_lin", "v_proj"],
+    ["out_lin", "out_proj"],
+    ["norm_embeddings", "layernorm_embedding"],
+    ["position_embeddings", "embed_positions"],
+    ["embeddings", "embed_tokens"],
+    ["ffn.lin", "fc"],
+]
+
+
+def rename_state_dict_key(k):
+    if k == "embeddings.weight":
+        return "shared.weight"
+
+    for parlai_name, hf_name in PATTERNS:
+        k = k.replace(parlai_name, hf_name)
+
+    if k.startswith("encoder"):
+        k = k.replace(".attn", ".self_attn")
+        k = k.replace("norm1", "self_attn_layer_norm")
+        k = k.replace("norm2", "final_layer_norm")
+    elif k.startswith("decoder"):
+        k = k.replace("norm1", "self_attn_layer_norm")
+        k = k.replace("norm2", "encoder_attn_layer_norm")
+        k = k.replace("norm3", "final_layer_norm")
+    return k
+
+
+def rename_layernorm_keys(sd):
+    keys = [
+        "model.encoder.layernorm_embedding.weight",
+        "model.encoder.layernorm_embedding.bias",
+        "model.decoder.layernorm_embedding.weight",
+        "model.decoder.layernorm_embedding.bias",
+    ]
+    for k in keys:
+        v = sd.pop(k)
+        new_k = k.replace("layernorm_embedding", "layer_norm")
+        assert new_k not in sd
+        sd[new_k] = v
+
+
+IGNORE_KEYS = ["START"]
+
+
+@torch.no_grad()
+def convert_parlai_checkpoint(checkpoint_path, pytorch_dump_folder_path, config_json_path):
+    """
+    Copy/paste/tweak model's weights to our BERT structure.
+    """
+    model = torch.load(checkpoint_path, map_location="cpu")
+    sd = model["model"]
+    cfg = BartConfig.from_json_file(config_json_path)
+    m = BartForConditionalGeneration(cfg)
+    valid_keys = m.model.state_dict().keys()
+    failures = []
+    mapping = {}
+    for k, v in sd.items():
+        if k in IGNORE_KEYS:
+            continue
+
+        new_k = rename_state_dict_key(k)
+        if new_k not in valid_keys:
+            failures.append([k, new_k])
+        else:
+            mapping[new_k] = v
+    if cfg.normalize_before:  # Blenderbot-3B checkpoints. Rename layernorm_embedding -> layer_norm
+        rename_layernorm_keys(sd)
+    m.model.load_state_dict(mapping, strict=True)
+    m.half()
+    m.save_pretrained(pytorch_dump_folder_path)
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    # Required parameters
+    parser.add_argument("--src_path", type=str, help="like blenderbot-model.bin")
+    parser.add_argument("--save_dir", default="hf_blenderbot", type=str, help="Where to save converted model.")
+    parser.add_argument(
+        "--hf_config_json", default="blenderbot-3b-config.json", type=str, help="Path to config to use"
+    )
+    args = parser.parse_args()
+    convert_parlai_checkpoint(args.src_path, args.save_dir, args.hf_config_json)

src/transformers/convert_graph_to_onnx.py

@@ -211,7 +211,7 @@ def load_graph_from_args(pipeline_name: str, framework: str, model: str, tokeniz
         pipeline_name: The kind of pipeline to use (ner, question-answering, etc.)
         framework: The actual model to convert the pipeline from ("pt" or "tf")
         model: The model name which will be loaded by the pipeline
-        tokenizer: The tokenizer name which will be loaded by the pipeline, defaut to the model's value
+        tokenizer: The tokenizer name which will be loaded by the pipeline, default to the model's value
 
     Returns: Pipeline object
 

src/transformers/convert_marian_to_pytorch.py

@@ -1,12 +1,11 @@
 import argparse
 import json
 import os
-import shutil
 import socket
 import time
 import warnings
 from pathlib import Path
-from typing import Dict, List, Tuple, Union
+from typing import Dict, List, Union
 from zipfile import ZipFile
 
 import numpy as np
@@ -23,85 +22,6 @@ def remove_suffix(text: str, suffix: str):
     return text  # or whatever
 
 
-def _process_benchmark_table_row(x):
-    fields = lmap(str.strip, x.replace("\t", "").split("|")[1:-1])
-    assert len(fields) == 3
-    return (fields[0], float(fields[1]), float(fields[2]))
-
-
-def process_last_benchmark_table(readme_path) -> List[Tuple[str, float, float]]:
-    md_content = Path(readme_path).open().read()
-    entries = md_content.split("## Benchmarks")[-1].strip().split("\n")[2:]
-    data = lmap(_process_benchmark_table_row, entries)
-    return data
-
-
-def check_if_models_are_dominated(old_repo_path="OPUS-MT-train/models", new_repo_path="Tatoeba-Challenge/models/"):
-    """Make a blacklist for models where we have already ported the same language pair, and the ported model has higher BLEU score."""
-    import pandas as pd
-
-    newest_released, old_reg, released = get_released_df(new_repo_path, old_repo_path)
-
-    short_to_new_bleu = newest_released.set_index("short_pair").bleu
-
-    assert released.groupby("short_pair").pair.nunique().max() == 1
-
-    short_to_long = released.groupby("short_pair").pair.first().to_dict()
-
-    overlap_short = old_reg.index.intersection(released.short_pair.unique())
-    overlap_long = [short_to_long[o] for o in overlap_short]
-    new_reported_bleu = [short_to_new_bleu[o] for o in overlap_short]
-
-    def get_old_bleu(o) -> float:
-        pat = old_repo_path + "/{}/README.md"
-        bm_data = process_last_benchmark_table(pat.format(o))
-        tab = pd.DataFrame(bm_data, columns=["testset", "bleu", "chr-f"])
-        tato_bleu = tab.loc[lambda x: x.testset.str.startswith("Tato")].bleu
-        if tato_bleu.shape[0] > 0:
-            return tato_bleu.iloc[0]
-        else:
-            return np.nan
-
-    old_bleu = [get_old_bleu(o) for o in overlap_short]
-    cmp_df = pd.DataFrame(
-        dict(short=overlap_short, long=overlap_long, old_bleu=old_bleu, new_bleu=new_reported_bleu)
-    ).fillna(-1)
-
-    dominated = cmp_df[cmp_df.old_bleu > cmp_df.new_bleu]
-    whitelist_df = cmp_df[cmp_df.old_bleu <= cmp_df.new_bleu]
-    blacklist = dominated.long.unique().tolist()  # 3 letter codes
-    return whitelist_df, dominated, blacklist
-
-
-def get_released_df(new_repo_path, old_repo_path):
-    import pandas as pd
-
-    released_cols = [
-        "url_base",
-        "pair",  # (ISO639-3/ISO639-5 codes),
-        "short_pair",  # (reduced codes),
-        "chrF2_score",
-        "bleu",
-        "brevity_penalty",
-        "ref_len",
-        "src_name",
-        "tgt_name",
-    ]
-    released = pd.read_csv(f"{new_repo_path}/released-models.txt", sep="\t", header=None).iloc[:-1]
-    released.columns = released_cols
-    old_reg = make_registry(repo_path=old_repo_path)
-    old_reg = pd.DataFrame(old_reg, columns=["id", "prepro", "url_model", "url_test_set"])
-    assert old_reg.id.value_counts().max() == 1
-    old_reg = old_reg.set_index("id")
-    released["fname"] = released["url_base"].apply(
-        lambda x: remove_suffix(remove_prefix(x, "https://object.pouta.csc.fi/Tatoeba-Challenge/opus"), ".zip")
-    )
-    released["2m"] = released.fname.str.startswith("2m")
-    released["date"] = pd.to_datetime(released["fname"].apply(lambda x: remove_prefix(remove_prefix(x, "2m-"), "-")))
-    newest_released = released.dsort("date").drop_duplicates(["short_pair"], keep="first")
-    return newest_released, old_reg, released
-
-
 def remove_prefix(text: str, prefix: str):
     if text.startswith(prefix):
         return text[len(prefix) :]
@@ -183,7 +103,11 @@ def find_model_file(dest_dir):  # this one better
 
 
 # Group Names Logic: change long opus model names to something shorter, like opus-mt-en-ROMANCE
-ROM_GROUP = "fr+fr_BE+fr_CA+fr_FR+wa+frp+oc+ca+rm+lld+fur+lij+lmo+es+es_AR+es_CL+es_CO+es_CR+es_DO+es_EC+es_ES+es_GT+es_HN+es_MX+es_NI+es_PA+es_PE+es_PR+es_SV+es_UY+es_VE+pt+pt_br+pt_BR+pt_PT+gl+lad+an+mwl+it+it_IT+co+nap+scn+vec+sc+ro+la"
+ROM_GROUP = (
+    "fr+fr_BE+fr_CA+fr_FR+wa+frp+oc+ca+rm+lld+fur+lij+lmo+es+es_AR+es_CL+es_CO+es_CR+es_DO+es_EC+es_ES+es_GT"
+    "+es_HN+es_MX+es_NI+es_PA+es_PE+es_PR+es_SV+es_UY+es_VE+pt+pt_br+pt_BR+pt_PT+gl+lad+an+mwl+it+it_IT+co"
+    "+nap+scn+vec+sc+ro+la"
+)
 GROUPS = [
     ("cmn+cn+yue+ze_zh+zh_cn+zh_CN+zh_HK+zh_tw+zh_TW+zh_yue+zhs+zht+zh", "ZH"),
     (ROM_GROUP, "ROMANCE"),
@@ -221,13 +145,15 @@ ORG_NAME = "Helsinki-NLP/"
 
 
 def convert_opus_name_to_hf_name(x):
+    """For OPUS-MT-Train/ DEPRECATED"""
     for substr, grp_name in GROUPS:
         x = x.replace(substr, grp_name)
     return x.replace("+", "_")
 
 
 def convert_hf_name_to_opus_name(hf_model_name):
-    """Relies on the assumption that there are no language codes like pt_br in models that are not in GROUP_TO_OPUS_NAME."""
+    """Relies on the assumption that there are no language codes like pt_br in models that are not in
+    GROUP_TO_OPUS_NAME."""
     hf_model_name = remove_prefix(hf_model_name, ORG_NAME)
     if hf_model_name in GROUP_TO_OPUS_NAME:
         opus_w_prefix = GROUP_TO_OPUS_NAME[hf_model_name]
@@ -247,8 +173,9 @@ def get_system_metadata(repo_root):
     )
 
 
-front_matter = """---
-language: {}
+FRONT_MATTER_TEMPLATE = """---
+language:
+{}
 tags:
 - translation
 
@@ -256,11 +183,13 @@ license: apache-2.0
 ---
 
 """
+DEFAULT_REPO = "Tatoeba-Challenge"
+DEFAULT_MODEL_DIR = os.path.join(DEFAULT_REPO, "models")
 
 
 def write_model_card(
     hf_model_name: str,
-    repo_root="OPUS-MT-train",
+    repo_root=DEFAULT_REPO,
     save_dir=Path("marian_converted"),
     dry_run=False,
     extra_metadata={},
@@ -294,7 +223,10 @@ def write_model_card(
 
     # combine with opus markdown
 
-    extra_markdown = f"### {hf_model_name}\n\n* source group: {metadata['src_name']} \n* target group: {metadata['tgt_name']} \n*  OPUS readme: [{opus_name}]({readme_url})\n"
+    extra_markdown = (
+        f"### {hf_model_name}\n\n* source group: {metadata['src_name']} \n* target group: "
+        f"{metadata['tgt_name']} \n*  OPUS readme: [{opus_name}]({readme_url})\n"
+    )
 
     content = opus_readme_path.open().read()
     content = content.split("\n# ")[-1]  # Get the lowest level 1 header in the README -- the most recent model.
@@ -302,7 +234,7 @@ def write_model_card(
     print(splat[3])
     content = "*".join(splat)
     content = (
-        front_matter.format(metadata["src_alpha2"])
+        FRONT_MATTER_TEMPLATE.format(metadata["src_alpha2"])
         + extra_markdown
         + "\n* "
         + content.replace("download", "download original weights")
@@ -323,48 +255,6 @@ def write_model_card(
     return content, metadata
 
 
-def get_clean_model_id_mapping(multiling_model_ids):
-    return {x: convert_opus_name_to_hf_name(x) for x in multiling_model_ids}
-
-
-def expand_group_to_two_letter_codes(grp_name):
-    raise NotImplementedError()
-
-
-def get_two_letter_code(three_letter_code):
-    raise NotImplementedError()
-    # return two_letter_code
-
-
-def get_tags(code, ref_name):
-    if len(code) == 2:
-        assert "languages" not in ref_name, f"{code}: {ref_name}"
-        return [code], False
-    elif "languages" in ref_name:
-        group = expand_group_to_two_letter_codes(code)
-        group.append(code)
-        return group, True
-    else:  # zho-> zh
-        raise ValueError(f"Three letter monolingual code: {code}")
-
-
-def resolve_lang_code(r):
-    """R is a row in ported"""
-    short_pair = r.short_pair
-    src, tgt = short_pair.split("-")
-    src_tags, src_multilingual = get_tags(src, r.src_name)
-    assert isinstance(src_tags, list)
-    tgt_tags, tgt_multilingual = get_tags(src, r.tgt_name)
-    assert isinstance(tgt_tags, list)
-    if src_multilingual:
-        src_tags.append("multilingual_src")
-    if tgt_multilingual:
-        tgt_tags.append("multilingual_tgt")
-    return src_tags + tgt_tags
-
-    # process target
-
-
 def make_registry(repo_path="Opus-MT-train/models"):
     if not (Path(repo_path) / "fr-en" / "README.md").exists():
         raise ValueError(
@@ -382,36 +272,25 @@ def make_registry(repo_path="Opus-MT-train/models"):
     return [(k, v["pre-processing"], v["download"], v["download"][:-4] + ".test.txt") for k, v in results.items()]
 
 
-def make_tatoeba_registry(repo_path="Tatoeba-Challenge/models"):
-    if not (Path(repo_path) / "zho-eng" / "README.md").exists():
-        raise ValueError(
-            f"repo_path:{repo_path} does not exist: "
-            "You must run: git clone git@github.com:Helsinki-NLP/Tatoeba-Challenge.git before calling."
-        )
-    results = {}
-    for p in Path(repo_path).iterdir():
-        if len(p.name) != 7:
-            continue
-        lns = list(open(p / "README.md").readlines())
-        results[p.name] = _parse_readme(lns)
-    return [(k, v["pre-processing"], v["download"], v["download"][:-4] + ".test.txt") for k, v in results.items()]
-
-
-def convert_all_sentencepiece_models(model_list=None, repo_path=None):
+def convert_all_sentencepiece_models(model_list=None, repo_path=None, dest_dir=Path("marian_converted")):
     """Requires 300GB"""
     save_dir = Path("marian_ckpt")
-    dest_dir = Path("marian_converted")
+    dest_dir = Path(dest_dir)
     dest_dir.mkdir(exist_ok=True)
+    save_paths = []
     if model_list is None:
         model_list: list = make_registry(repo_path=repo_path)
     for k, prepro, download, test_set_url in tqdm(model_list):
         if "SentencePiece" not in prepro:  # dont convert BPE models.
             continue
-        if not os.path.exists(save_dir / k / "pytorch_model.bin"):
+        if not os.path.exists(save_dir / k):
             download_and_unzip(download, save_dir / k)
         pair_name = convert_opus_name_to_hf_name(k)
         convert(save_dir / k, dest_dir / f"opus-mt-{pair_name}")
 
+        save_paths.append(dest_dir / f"opus-mt-{pair_name}")
+    return save_paths
+
 
 def lmap(f, x) -> List:
     return list(map(f, x))
@@ -493,15 +372,6 @@ def add_special_tokens_to_vocab(model_dir: Path) -> None:
     save_tokenizer_config(model_dir)
 
 
-def save_tokenizer(self, save_directory):
-    dest = Path(save_directory)
-    src_path = Path(self.init_kwargs["source_spm"])
-
-    for dest_name in {"source.spm", "target.spm", "tokenizer_config.json"}:
-        shutil.copyfile(src_path.parent / dest_name, dest / dest_name)
-    save_json(self.encoder, dest / "vocab.json")
-
-
 def check_equal(marian_cfg, k1, k2):
     v1, v2 = marian_cfg[k1], marian_cfg[k2]
     assert v1 == v2, f"hparams {k1},{k2} differ: {v1} != {v2}"
@@ -698,14 +568,14 @@ def convert(source_dir: Path, dest_dir):
 
     add_special_tokens_to_vocab(source_dir)
     tokenizer = MarianTokenizer.from_pretrained(str(source_dir))
-    save_tokenizer(tokenizer, dest_dir)
+    tokenizer.save_pretrained(dest_dir)
 
     opus_state = OpusState(source_dir)
     assert opus_state.cfg["vocab_size"] == len(
         tokenizer.encoder
     ), f"Original vocab size {opus_state.cfg['vocab_size']} and new vocab size {len(tokenizer.encoder)} mismatched"
     # save_json(opus_state.cfg, dest_dir / "marian_original_config.json")
-    # ^^ Save human readable marian config for debugging
+    # ^^ Uncomment to save human readable marian config for debugging
 
     model = opus_state.load_marian_model()
     model = model.half()
@@ -732,15 +602,11 @@ def unzip(zip_path: str, dest_dir: str) -> None:
 
 if __name__ == "__main__":
     """
-    To bulk convert, run
-    >>> from transformers.convert_marian_to_pytorch import make_tatoeba_registry, convert_all_sentencepiece_models
-    >>> reg = make_tatoeba_registry()
-    >>> convert_all_sentencepiece_models(model_list=reg)  # saves to marian_converted
-    (bash) aws s3 sync marian_converted s3://models.huggingface.co/bert/Helsinki-NLP/ --dryrun
+    Tatoeba conversion instructions in scripts/tatoeba/README.md
     """
     parser = argparse.ArgumentParser()
     # Required parameters
-    parser.add_argument("--src", type=str, help="path to marian model dir", default="en-de")
+    parser.add_argument("--src", type=str, help="path to marian model sub dir", default="en-de")
     parser.add_argument("--dest", type=str, default=None, help="Path to the output PyTorch model.")
     args = parser.parse_args()
 

src/transformers/convert_slow_tokenizer.py

@@ -1,3 +1,23 @@
+# coding=utf-8
+# Copyright 2018 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.
+""" Utilities to convert slow tokenizers in their fast tokenizers counterparts.
+
+    All the conversions are grouped here to gather SentencePiece dependencies outside of
+    the fast tokenizers files and allow to make our dependency on SentencePiece optional.
+"""
+
 from typing import Dict, List, Tuple
 
 from sentencepiece import SentencePieceProcessor
@@ -527,6 +547,7 @@ CONVERTERS = {
     "DPRReaderTokenizer": BertConverter,
     "DPRQuestionEncoderTokenizer": BertConverter,
     "DPRContextEncoderTokenizer": BertConverter,
+    "ElectraTokenizer": BertConverter,
     "FunnelTokenizer": FunnelConverter,
     "GPT2Tokenizer": GPT2Converter,
     "LxmertTokenizer": BertConverter,

src/transformers/data/data_collator.py

@@ -446,7 +446,7 @@ class DataCollatorForNextSentencePrediction:
             "input_ids": input_ids,
             "attention_mask": self._tensorize_batch(attention_masks),
             "token_type_ids": self._tensorize_batch(segment_ids),
-            "masked_lm_labels": mlm_labels if self.mlm else None,
+            "labels": mlm_labels if self.mlm else None,
             "next_sentence_label": torch.tensor(nsp_labels),
         }
         if self.mlm:

src/transformers/data/processors/squad.py

@@ -560,7 +560,7 @@ class SquadProcessor(DataProcessor):
 
         Args:
             dataset: The tfds dataset loaded from `tensorflow_datasets.load("squad")`
-            evaluate: boolean specifying if in evaluation mode or in training mode
+            evaluate: Boolean specifying if in evaluation mode or in training mode
 
         Returns:
             List of SquadExample
@@ -614,7 +614,7 @@ class SquadProcessor(DataProcessor):
         Args:
             data_dir: Directory containing the data files used for training and evaluating.
             filename: None by default, specify this if the evaluation file has a different name than the original one
-                which is `train-v1.1.json` and `train-v2.0.json` for squad versions 1.1 and 2.0 respectively.
+                which is `dev-v1.1.json` and `dev-v2.0.json` for squad versions 1.1 and 2.0 respectively.
         """
         if data_dir is None:
             data_dir = ""

src/transformers/file_utils.py

@@ -143,24 +143,6 @@ except (AttributeError, ImportError):
     _has_sklearn = False
 
 
-try:
-    import sentencepiece  # noqa: F401
-
-    _sentencepiece_available = True
-
-except ImportError:
-    _sentencepiece_available = False
-
-
-try:
-    import tokenizers  # noqa: F401
-
-    _tokenizers_available = True
-
-except ImportError:
-    _tokenizers_available = False
-
-
 default_cache_path = os.path.join(torch_cache_home, "transformers")
 
 
@@ -221,18 +203,23 @@ def is_faiss_available():
     return _faiss_available
 
 
+def torch_only_method(fn):
+    def wrapper(*args, **kwargs):
+        if not _torch_available:
+            raise ImportError(
+                "You need to install pytorch to use this method or class, "
+                "or activate it with environment variables USE_TORCH=1 and USE_TF=0."
+            )
+        else:
+            return fn(*args, **kwargs)
+
+    return wrapper
+
+
 def is_sklearn_available():
     return _has_sklearn
 
 
-def is_sentencepiece_available():
-    return _sentencepiece_available
-
-
-def is_tokenizers_available():
-    return _tokenizers_available
-
-
 DATASETS_IMPORT_ERROR = """
 {0} requires the 🤗 Datasets library but it was not found in your enviromnent. You can install it with:
 ```
@@ -1106,7 +1093,7 @@ def is_tensor(x):
 class ModelOutput(OrderedDict):
     """
     Base class for all model outputs as dataclass. Has a ``__getitem__`` that allows indexing by integer or slice (like
-    a tuple) or strings (like a dictionnary) that will ignore the ``None`` attributes. Otherwise behaves like a
+    a tuple) or strings (like a dictionary) that will ignore the ``None`` attributes. Otherwise behaves like a
     regular python dictionary.
 
     .. warning::

src/transformers/generation_tf_utils.py

@@ -225,7 +225,7 @@ class TFGenerationMixin:
         assert isinstance(do_sample, bool), "`do_sample` should be a boolean."
         assert isinstance(early_stopping, bool), "`early_stopping` should be a boolean."
         assert isinstance(use_cache, bool), "`use_cache` should be a boolean."
-        assert isinstance(num_beams, int) and num_beams > 0, "`num_beams` should be a strictely positive integer."
+        assert isinstance(num_beams, int) and num_beams > 0, "`num_beams` should be a strictly positive integer."
         assert temperature > 0, "`temperature` should be strictely positive."
         assert isinstance(top_k, int) and top_k >= 0, "`top_k` should be a positive integer."
         assert 0 <= top_p <= 1, "`top_p` should be between 0 and 1."

src/transformers/integrations.py

@@ -3,7 +3,10 @@ import math
 import os
 
 
+# Import 3rd-party integrations first:
+
 try:
+    # Comet needs to be imported before any ML frameworks
     import comet_ml  # noqa: F401
 
     _has_comet = True
@@ -36,15 +39,6 @@ try:
 except (ImportError):
     _has_ray = False
 
-
-# No ML framework or transformer imports above this point
-
-from .trainer_utils import PREFIX_CHECKPOINT_DIR, BestRun  # isort:skip
-from .utils import logging  # isort:skip
-
-logger = logging.get_logger(__name__)
-
-
 try:
     from torch.utils.tensorboard import SummaryWriter  # noqa: F401
 
@@ -57,9 +51,18 @@ except ImportError:
     except ImportError:
         _has_tensorboard = False
 
+# No transformer imports above this point
+
+from .file_utils import is_torch_tpu_available
+from .trainer_callback import TrainerCallback
+from .trainer_utils import PREFIX_CHECKPOINT_DIR, BestRun
+from .utils import logging
+
+
+logger = logging.get_logger(__name__)
+
+
 # Integration functions:
-
-
 def is_wandb_available():
     return _has_wandb
 
@@ -128,8 +131,8 @@ def run_hp_search_ray(trainer, n_trials: int, direction: str, **kwargs) -> BestR
 
     # The model and TensorBoard writer do not pickle so we have to remove them (if they exists)
     # while doing the ray hp search.
-    _tb_writer = trainer.tb_writer
-    trainer.tb_writer = None
+
+    _tb_writer = trainer.pop_callback(TensorBoardCallback)
     trainer.model = None
     # Setup default `resources_per_trial` and `reporter`.
     if "resources_per_trial" not in kwargs and trainer.args.n_gpu > 0:
@@ -182,5 +185,159 @@ def run_hp_search_ray(trainer, n_trials: int, direction: str, **kwargs) -> BestR
     analysis = ray.tune.run(_objective, config=trainer.hp_space(None), num_samples=n_trials, **kwargs)
     best_trial = analysis.get_best_trial(metric="objective", mode=direction[:3])
     best_run = BestRun(best_trial.trial_id, best_trial.last_result["objective"], best_trial.config)
-    trainer.tb_writer = _tb_writer
+    if _tb_writer is not None:
+        trainer.add_callback(_tb_writer)
     return best_run
+
+
+class TensorBoardCallback(TrainerCallback):
+    """
+    A :class:`~transformers.TrainerCallback` that sends the logs to `TensorBoard
+    <https://www.tensorflow.org/tensorboard>`__.
+
+    Args:
+        tb_writer (:obj:`SummaryWriter`, `optional`):
+            The writer to use. Will instantiate one if not set.
+    """
+
+    def __init__(self, tb_writer=None):
+        assert (
+            _has_tensorboard
+        ), "TensorBoardCallback requires tensorboard to be installed. Either update your PyTorch version or install tensorboardX."
+        self.tb_writer = tb_writer
+
+    def on_init_end(self, args, state, control, **kwargs):
+        if self.tb_writer is None and state.is_world_process_zero:
+            self.tb_writer = SummaryWriter(log_dir=args.logging_dir)
+
+    def on_train_begin(self, args, state, control, **kwargs):
+        if self.tb_writer is not None:
+            self.tb_writer.add_text("args", args.to_json_string())
+            self.tb_writer.add_hparams(args.to_sanitized_dict(), metric_dict={})
+
+    def on_log(self, args, state, control, logs=None, **kwargs):
+        if self.tb_writer:
+            for k, v in logs.items():
+                if isinstance(v, (int, float)):
+                    self.tb_writer.add_scalar(k, v, state.global_step)
+                else:
+                    logger.warning(
+                        "Trainer is attempting to log a value of "
+                        '"%s" of type %s for key "%s" as a scalar. '
+                        "This invocation of Tensorboard's writer.add_scalar() "
+                        "is incorrect so we dropped this attribute.",
+                        v,
+                        type(v),
+                        k,
+                    )
+            self.tb_writer.flush()
+
+    def on_train_end(self, args, state, control, **kwargs):
+        if self.tb_writer:
+            self.tb_writer.close()
+
+
+class WandbCallback(TrainerCallback):
+    """
+    A :class:`~transformers.TrainerCallback` that sends the logs to `Weight and Biases
+    <https://www.wandb.com/>`__.
+    """
+
+    def __init__(self):
+        assert _has_wandb, "WandbCallback requires wandb to be installed. Run `pip install wandb`."
+        self._initialized = False
+
+    def setup(self, args, state, model):
+        """
+        Setup the optional Weights & Biases (`wandb`) integration.
+
+        One can subclass and override this method to customize the setup if needed. Find more information
+        `here <https://docs.wandb.com/huggingface>`__. You can also override the following environment variables:
+
+        Environment:
+            WANDB_WATCH (:obj:`str`, `optional` defaults to :obj:`"gradients"`):
+                Can be :obj:`"gradients"`, :obj:`"all"` or :obj:`"false"`. Set to :obj:`"false"` to disable gradient
+                logging or :obj:`"all"` to log gradients and parameters.
+            WANDB_PROJECT (:obj:`str`, `optional`, defaults to :obj:`"huggingface"`):
+                Set this to a custom string to store results in a different project.
+            WANDB_DISABLED (:obj:`bool`, `optional`, defaults to :obj:`False`):
+                Whether or not to disable wandb entirely.
+        """
+        self._initialized = True
+        if state.is_world_process_zero:
+            logger.info(
+                'Automatic Weights & Biases logging enabled, to disable set os.environ["WANDB_DISABLED"] = "true"'
+            )
+            combined_dict = {**args.to_sanitized_dict()}
+            if hasattr(model, "config"):
+                combined_dict = {**model.config.to_dict(), **combined_dict}
+            wandb.init(project=os.getenv("WANDB_PROJECT", "huggingface"), config=combined_dict, name=args.run_name)
+            # keep track of model topology and gradients, unsupported on TPU
+            if not is_torch_tpu_available() and os.getenv("WANDB_WATCH") != "false":
+                wandb.watch(model, log=os.getenv("WANDB_WATCH", "gradients"), log_freq=max(100, args.logging_steps))
+
+    def on_train_begin(self, args, state, control, model=None, **kwargs):
+        if not self._initialized:
+            self.setup(args, state, model)
+
+    def on_log(self, args, state, control, model=None, logs=None, **kwargs):
+        if not self._initialized:
+            self.setup(args, state, model)
+        if state.is_world_process_zero:
+            wandb.log(logs, step=state.global_step)
+
+
+class CometCallback(TrainerCallback):
+    """
+    A :class:`~transformers.TrainerCallback` that sends the logs to `Comet ML
+    <https://www.comet.ml/site/>`__.
+    """
+
+    def __init__(self):
+        assert _has_comet, "CometCallback requires comet-ml to be installed. Run `pip install comet-ml`."
+        self._initialized = False
+
+    def setup(self, args, state, model):
+        """
+        Setup the optional Comet.ml integration.
+
+        Environment:
+            COMET_MODE (:obj:`str`, `optional`):
+                "OFFLINE", "ONLINE", or "DISABLED"
+            COMET_PROJECT_NAME (:obj:`str`, `optional`):
+                Comet.ml project name for experiments
+            COMET_OFFLINE_DIRECTORY (:obj:`str`, `optional`):
+                Folder to use for saving offline experiments when :obj:`COMET_MODE` is "OFFLINE"
+
+        For a number of configurable items in the environment,
+        see `here <https://www.comet.ml/docs/python-sdk/advanced/#comet-configuration-variables>`__.
+        """
+        self._initialized = True
+        if state.is_world_process_zero:
+            comet_mode = os.getenv("COMET_MODE", "ONLINE").upper()
+            args = {"project_name": os.getenv("COMET_PROJECT_NAME", "huggingface")}
+            experiment = None
+            if comet_mode == "ONLINE":
+                experiment = comet_ml.Experiment(**args)
+                logger.info("Automatic Comet.ml online logging enabled")
+            elif comet_mode == "OFFLINE":
+                args["offline_directory"] = os.getenv("COMET_OFFLINE_DIRECTORY", "./")
+                experiment = comet_ml.OfflineExperiment(**args)
+                logger.info("Automatic Comet.ml offline logging enabled; use `comet upload` when finished")
+            if experiment is not None:
+                experiment._set_model_graph(model, framework="transformers")
+                experiment._log_parameters(args, prefix="args/", framework="transformers")
+                if hasattr(model, "config"):
+                    experiment._log_parameters(model.config, prefix="config/", framework="transformers")
+
+    def on_train_begin(self, args, state, control, model=None, **kwargs):
+        if not self._initialized:
+            self.setup(args, state, model)
+
+    def on_log(self, args, state, control, model=None, logs=None, **kwargs):
+        if not self._initialized:
+            self.setup(args, state, model)
+        if state.is_world_process_zero:
+            experiment = comet_ml.config.get_global_experiment()
+            if experiment is not None:
+                experiment._log_metrics(logs, step=state.global_step, epoch=state.epoch, framework="transformers")

src/transformers/modeling_albert.py

@@ -539,7 +539,7 @@ ALBERT_INPUTS_DOCSTRING = r"""
             Mask values selected in ``[0, 1]``:
 
             - 1 for tokens that are **not masked**,
-            - 0 for tokens that are **maked**.
+            - 0 for tokens that are **masked**.
 
             `What are attention masks? <../glossary.html#attention-mask>`__
         token_type_ids (:obj:`torch.LongTensor` of shape :obj:`({0})`, `optional`):

src/transformers/modeling_auto.py

@@ -24,6 +24,7 @@ from .configuration_auto import (
     BartConfig,
     BertConfig,
     BertGenerationConfig,
+    BlenderbotConfig,
     CamembertConfig,
     CTRLConfig,
     DebertaConfig,
@@ -82,6 +83,7 @@ from .modeling_bert import (
     BertModel,
 )
 from .modeling_bert_generation import BertGenerationDecoder, BertGenerationEncoder
+from .modeling_blenderbot import BlenderbotForConditionalGeneration
 from .modeling_camembert import (
     CamembertForCausalLM,
     CamembertForMaskedLM,
@@ -129,7 +131,7 @@ from .modeling_funnel import (
     FunnelForTokenClassification,
     FunnelModel,
 )
-from .modeling_gpt2 import GPT2LMHeadModel, GPT2Model
+from .modeling_gpt2 import GPT2ForSequenceClassification, GPT2LMHeadModel, GPT2Model
 from .modeling_layoutlm import LayoutLMForMaskedLM, LayoutLMForTokenClassification, LayoutLMModel
 from .modeling_longformer import (
     LongformerForMaskedLM,
@@ -151,7 +153,7 @@ from .modeling_mobilebert import (
     MobileBertForTokenClassification,
     MobileBertModel,
 )
-from .modeling_openai import OpenAIGPTLMHeadModel, OpenAIGPTModel
+from .modeling_openai import OpenAIGPTForSequenceClassification, OpenAIGPTLMHeadModel, OpenAIGPTModel
 from .modeling_pegasus import PegasusForConditionalGeneration
 from .modeling_rag import (  # noqa: F401 - need to import all RagModels to be in globals() function
     RagModel,
@@ -353,6 +355,7 @@ MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING = OrderedDict(
         (PegasusConfig, PegasusForConditionalGeneration),
         (MarianConfig, MarianMTModel),
         (MBartConfig, MBartForConditionalGeneration),
+        (BlenderbotConfig, BlenderbotForConditionalGeneration),
         (BartConfig, BartForConditionalGeneration),
         (FSMTConfig, FSMTForConditionalGeneration),
         (EncoderDecoderConfig, EncoderDecoderModel),
@@ -377,6 +380,8 @@ MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING = OrderedDict(
         (ElectraConfig, ElectraForSequenceClassification),
         (FunnelConfig, FunnelForSequenceClassification),
         (DebertaConfig, DebertaForSequenceClassification),
+        (GPT2Config, GPT2ForSequenceClassification),
+        (OpenAIGPTConfig, OpenAIGPTForSequenceClassification),
     ]
 )
 
@@ -502,7 +507,7 @@ AUTO_MODEL_PRETRAINED_DOCSTRING = r"""
                 :obj:`{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. The proxies are used on each
                 request.
             output_loading_info(:obj:`bool`, `optional`, defaults to :obj:`False`):
-                Whether ot not to also return a dictionnary containing missing keys, unexpected keys and error
+                Whether ot not to also return a dictionary containing missing keys, unexpected keys and error
                 messages.
             local_files_only(:obj:`bool`, `optional`, defaults to :obj:`False`):
                 Whether or not to only look at local files (e.g., not try doanloading the model).
@@ -528,7 +533,7 @@ AUTO_MODEL_PRETRAINED_DOCSTRING = r"""
 class AutoModel:
     r"""
     This is a generic model class that will be instantiated as one of the base model classes of the library
-    when created with the when created with the :meth:`~transformers.AutoModel.from_pretrained` class method or the
+    when created with the :meth:`~transformers.AutoModel.from_pretrained` class method or the
     :meth:`~transformers.AutoModel.from_config` class methods.
 
     This class cannot be instantiated directly using ``__init__()`` (throws an error).

src/transformers/modeling_bart.py

@@ -113,7 +113,7 @@ BART_INPUTS_DOCSTRING = r"""
             Mask values selected in ``[0, 1]``:
 
             - 1 for tokens that are **not masked**,
-            - 0 for tokens that are **maked**.
+            - 0 for tokens that are **masked**.
 
             `What are attention masks? <../glossary.html#attention-mask>`__
         decoder_input_ids (:obj:`torch.LongTensor` of shape :obj:`(batch_size, target_sequence_length)`, `optional`):
@@ -499,6 +499,7 @@ class BartDecoder(nn.Module):
         super().__init__()
         self.dropout = config.dropout
         self.layerdrop = config.decoder_layerdrop
+        self.do_blenderbot_90_layernorm = config.do_blenderbot_90_layernorm  # layernorm variant
         self.padding_idx = embed_tokens.padding_idx
         self.max_target_positions = config.max_position_embeddings
         self.embed_scale = math.sqrt(config.d_model) if config.scale_embedding else 1.0
@@ -578,8 +579,13 @@ class BartDecoder(nn.Module):
             positions = positions[:, -1:]
 
         x = self.embed_tokens(input_ids) * self.embed_scale
-        x += positions
-        x = self.layernorm_embedding(x)
+        if self.do_blenderbot_90_layernorm:
+            x = self.layernorm_embedding(x)
+            x += positions
+        else:
+            x += positions
+            x = self.layernorm_embedding(x)
+
         x = F.dropout(x, p=self.dropout, training=self.training)
 
         # Convert to Bart output format: (seq_len, BS, model_dim) -> (BS, seq_len, model_dim)
@@ -1176,7 +1182,7 @@ class BartForSequenceClassification(PretrainedBartModel):
             config.d_model,
             config.d_model,
             config.num_labels,
-            config.classif_dropout,
+            config.classifier_dropout,
         )
         self.model._init_weights(self.classification_head.dense)
         self.model._init_weights(self.classification_head.out_proj)

src/transformers/modeling_bert.py

@@ -667,7 +667,7 @@ BERT_INPUTS_DOCSTRING = r"""
             Mask values selected in ``[0, 1]``:
 
             - 1 for tokens that are **not masked**,
-            - 0 for tokens that are **maked**.
+            - 0 for tokens that are **masked**.
 
             `What are attention masks? <../glossary.html#attention-mask>`__
         token_type_ids (:obj:`torch.LongTensor` of shape :obj:`({0})`, `optional`):
@@ -781,7 +781,7 @@ class BertModel(BertPreTrainedModel):
             Mask values selected in ``[0, 1]``:
 
             - 1 for tokens that are **not masked**,
-            - 0 for tokens that are **maked**.
+            - 0 for tokens that are **masked**.
         """
         output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
         output_hidden_states = (
@@ -1012,7 +1012,7 @@ class BertLMHeadModel(BertPreTrainedModel):
             Mask values selected in ``[0, 1]``:
 
             - 1 for tokens that are **not masked**,
-            - 0 for tokens that are **maked**.
+            - 0 for tokens that are **masked**.
         labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`):
             Labels for computing the left-to-right language modeling loss (next word prediction).
             Indices should be in ``[-100, 0, ..., config.vocab_size]`` (see ``input_ids`` docstring)

src/transformers/modeling_bert_generation.py

@@ -218,7 +218,7 @@ BERT_GENERATION_INPUTS_DOCSTRING = r"""
             Mask values selected in ``[0, 1]``:
 
             - 1 for tokens that are **not masked**,
-            - 0 for tokens that are **maked**.
+            - 0 for tokens that are **masked**.
 
             `What are attention masks? <../glossary.html#attention-mask>`__
         position_ids (:obj:`torch.LongTensor` of shape :obj:`({0})`, `optional`):
@@ -450,7 +450,7 @@ class BertGenerationDecoder(BertGenerationPreTrainedModel):
             Mask values selected in ``[0, 1]``:
 
             - 1 for tokens that are **not masked**,
-            - 0 for tokens that are **maked**.
+            - 0 for tokens that are **masked**.
         labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`):
             Labels for computing the left-to-right language modeling loss (next word prediction).
             Indices should be in ``[-100, 0, ..., config.vocab_size]`` (see ``input_ids`` docstring)

src/transformers/modeling_blenderbot.py

@@ -0,0 +1,56 @@
+#!/usr/bin/env python3
+# coding=utf-8
+# Copyright (c) Facebook, Inc. and its affiliates.
+#
+# This source code is licensed under the MIT license found in the;
+# 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.
+# LICENSE file in the root directory of this source tree.
+""""BlenderbotForConditionalGeneration which inherits from BART"""
+
+import torch
+
+from .configuration_blenderbot import BlenderbotConfig
+from .file_utils import add_start_docstrings
+from .modeling_bart import BartForConditionalGeneration
+
+
+BLENDER_START_DOCSTRING = r"""
+
+    This model inherits from :class:`~transformers.PreTrainedModel`. Check the superclass documentation for the generic
+    methods the library implements for all its model (such as downloading or saving, resizing the input embeddings,
+    pruning heads etc.)
+
+    This model is also a PyTorch `torch.nn.Module <https://pytorch.org/docs/stable/nn.html#torch.nn.Module>`__ subclass.
+    Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general
+    usage and behavior.
+
+"""
+
+BLENDERBOT_PRETRAINED_MODEL_ARCHIVE_LIST = ["facebook/blenderbot-3B", "facebook/blenderbot-90M"]
+
+
+@add_start_docstrings(
+    "The BART Model with a language modeling head. Can be used for summarization.", BLENDER_START_DOCSTRING
+)
+class BlenderbotForConditionalGeneration(BartForConditionalGeneration):
+    """
+    This class overrides :class:`~transformers.BartForConditionalGeneration`. Please check the
+    superclass for the appropriate documentation alongside usage examples.
+    """
+
+    config_class = BlenderbotConfig
+
+    def adjust_logits_during_generation(self, logits, cur_len, max_length):
+        logits[:, self.config.bos_token_id] = -torch.finfo(torch.float16).max  # near infinity fp16
+        if cur_len == max_length - 1 and self.config.eos_token_id is not None:
+            self._force_token_ids_generation(logits, self.config.eos_token_id)
+        return logits

src/transformers/modeling_ctrl.py

@@ -273,7 +273,7 @@ CTRL_INPUTS_DOCSTRING = r"""
             Mask values selected in ``[0, 1]``:
 
             - 1 for tokens that are **not masked**,
-            - 0 for tokens that are **maked**.
+            - 0 for tokens that are **masked**.
 
             `What are attention masks? <../glossary.html#attention-mask>`__
         token_type_ids (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`):

src/transformers/modeling_distilbert.py

@@ -401,7 +401,7 @@ DISTILBERT_INPUTS_DOCSTRING = r"""
             Mask values selected in ``[0, 1]``:
 
             - 1 for tokens that are **not masked**,
-            - 0 for tokens that are **maked**.
+            - 0 for tokens that are **masked**.
 
             `What are attention masks? <../glossary.html#attention-mask>`__
         head_mask (:obj:`torch.FloatTensor` of shape :obj:`(num_heads,)` or :obj:`(num_layers, num_heads)`, `optional`):

src/transformers/modeling_dpr.py

@@ -358,7 +358,7 @@ DPR_ENCODERS_INPUTS_DOCSTRING = r"""
             Mask values selected in ``[0, 1]``:
 
             - 1 for tokens that are **not masked**,
-            - 0 for tokens that are **maked**.
+            - 0 for tokens that are **masked**.
 
             `What are attention masks? <../glossary.html#attention-mask>`__
         token_type_ids (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`):
@@ -403,7 +403,7 @@ DPR_READER_INPUTS_DOCSTRING = r"""
             Mask values selected in ``[0, 1]``:
 
             - 1 for tokens that are **not masked**,
-            - 0 for tokens that are **maked**.
+            - 0 for tokens that are **masked**.
 
             `What are attention masks? <../glossary.html#attention-mask>`__
         inputs_embeds (:obj:`torch.FloatTensor` of shape :obj:`(n_passages, sequence_length, hidden_size)`, `optional`):

src/transformers/modeling_electra.py

@@ -611,7 +611,7 @@ ELECTRA_INPUTS_DOCSTRING = r"""
             Mask values selected in ``[0, 1]``:
 
             - 1 for tokens that are **not masked**,
-            - 0 for tokens that are **maked**.
+            - 0 for tokens that are **masked**.
 
             `What are attention masks? <../glossary.html#attention-mask>`__
         token_type_ids (:obj:`torch.LongTensor` of shape :obj:`({0})`, `optional`):

src/transformers/modeling_encoder_decoder.py

@@ -74,7 +74,7 @@ ENCODER_DECODER_INPUTS_DOCSTRING = r"""
             Mask values selected in ``[0, 1]``:
 
             - 1 for tokens that are **not masked**,
-            - 0 for tokens that are **maked**.
+            - 0 for tokens that are **masked**.
 
             `What are attention masks? <../glossary.html#attention-mask>`__
         decoder_input_ids (:obj:`torch.LongTensor` of shape :obj:`(batch_size, target_sequence_length)`, `optional`):

src/transformers/modeling_flaubert.py

@@ -81,7 +81,7 @@ FLAUBERT_INPUTS_DOCSTRING = r"""
             Mask values selected in ``[0, 1]``:
 
             - 1 for tokens that are **not masked**,
-            - 0 for tokens that are **maked**.
+            - 0 for tokens that are **masked**.
 
             `What are attention masks? <../glossary.html#attention-mask>`__
         token_type_ids (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`):

src/transformers/modeling_fsmt.py

@@ -224,7 +224,7 @@ FSMT_INPUTS_DOCSTRING = r"""
             Mask values selected in ``[0, 1]``:
 
             - 1 for tokens that are **not masked**,
-            - 0 for tokens that are **maked**.
+            - 0 for tokens that are **masked**.
 
             `What are attention masks? <../glossary.html#attention-mask>`__
         decoder_input_ids (:obj:`torch.LongTensor` of shape :obj:`(batch_size, target_sequence_length)`, `optional`):

src/transformers/modeling_funnel.py

@@ -857,7 +857,7 @@ FUNNEL_INPUTS_DOCSTRING = r"""
             Mask values selected in ``[0, 1]``:
 
             - 1 for tokens that are **not masked**,
-            - 0 for tokens that are **maked**.
+            - 0 for tokens that are **masked**.
 
             `What are attention masks? <../glossary.html#attention-mask>`__
         token_type_ids (:obj:`torch.LongTensor` of shape :obj:`({0})`, `optional`):

src/transformers/modeling_gpt2.py

@@ -22,7 +22,7 @@ from typing import List, Optional, Tuple
 
 import torch
 import torch.nn as nn
-from torch.nn import CrossEntropyLoss
+from torch.nn import CrossEntropyLoss, MSELoss
 
 from .activations import ACT2FN
 from .configuration_gpt2 import GPT2Config
@@ -33,7 +33,11 @@ from .file_utils import (
     add_start_docstrings_to_callable,
     replace_return_docstrings,
 )
-from .modeling_outputs import BaseModelOutputWithPast, CausalLMOutputWithPast
+from .modeling_outputs import (
+    BaseModelOutputWithPast,
+    CausalLMOutputWithPast,
+    SequenceClassifierOutputWithPast,
+)
 from .modeling_utils import (
     Conv1D,
     PreTrainedModel,
@@ -42,6 +46,7 @@ from .modeling_utils import (
     prune_conv1d_layer,
 )
 from .utils import logging
+from .utils.model_parallel_utils import assert_device_map, get_device_map
 
 
 logger = logging.get_logger(__name__)
@@ -124,7 +129,10 @@ class Attention(nn.Module):
         # [switch nx => n_state from Block to Attention to keep identical to TF implem]
         assert n_state % config.n_head == 0
         self.register_buffer(
-            "bias", torch.tril(torch.ones((n_ctx, n_ctx), dtype=torch.uint8)).view(1, 1, n_ctx, n_ctx)
+            "bias",
+            torch.tril(torch.ones((n_ctx, n_ctx), dtype=torch.uint8)).view(
+                1, 1, n_ctx, n_ctx
+            ),
         )
         self.register_buffer("masked_bias", torch.tensor(-1e4))
         self.n_head = config.n_head
@@ -147,7 +155,9 @@ class Attention(nn.Module):
         heads, index = find_pruneable_heads_and_indices(
             heads, self.n_head, self.split_size // self.n_head, self.pruned_heads
         )
-        index_attn = torch.cat([index, index + self.split_size, index + (2 * self.split_size)])
+        index_attn = torch.cat(
+            [index, index + self.split_size, index + (2 * self.split_size)]
+        )
 
         # Prune conv1d layers
         self.c_attn = prune_conv1d_layer(self.c_attn, index_attn, dim=1)
@@ -158,7 +168,9 @@ class Attention(nn.Module):
         self.n_head = self.n_head - len(heads)
         self.pruned_heads = self.pruned_heads.union(heads)
 
-    def _attn(self, q, k, v, attention_mask=None, head_mask=None, output_attentions=False):
+    def _attn(
+        self, q, k, v, attention_mask=None, head_mask=None, output_attentions=False
+    ):
         w = torch.matmul(q, k)
         if self.scale:
             w = w / (float(v.size(-1)) ** 0.5)
@@ -214,7 +226,9 @@ class Attention(nn.Module):
                 self, "q_attn"
             ), "If class is used as cross attention, the weights `q_attn` have to be defined. Please make sure to instantiate class with `Attention(..., is_cross_attention=True)`."
             query = self.q_attn(hidden_states)
-            key, value = self.c_attn(encoder_hidden_states).split(self.split_size, dim=2)
+            key, value = self.c_attn(encoder_hidden_states).split(
+                self.split_size, dim=2
+            )
             attention_mask = encoder_attention_mask
         else:
             query, key, value = self.c_attn(hidden_states).split(self.split_size, dim=2)
@@ -223,16 +237,23 @@ class Attention(nn.Module):
         key = self.split_heads(key, k=True)
         value = self.split_heads(value)
         if layer_past is not None:
-            past_key, past_value = layer_past[0].transpose(-2, -1), layer_past[1]  # transpose back cf below
+            past_key, past_value = (
+                layer_past[0].transpose(-2, -1),
+                layer_past[1],
+            )  # transpose back cf below
             key = torch.cat((past_key, key), dim=-1)
             value = torch.cat((past_value, value), dim=-2)
 
         if use_cache is True:
-            present = torch.stack((key.transpose(-2, -1), value))  # transpose to have same shapes for stacking
+            present = torch.stack(
+                (key.transpose(-2, -1), value)
+            )  # transpose to have same shapes for stacking
         else:
             present = (None,)
 
-        attn_outputs = self._attn(query, key, value, attention_mask, head_mask, output_attentions)
+        attn_outputs = self._attn(
+            query, key, value, attention_mask, head_mask, output_attentions
+        )
         a = attn_outputs[0]
 
         a = self.merge_heads(a)
@@ -267,8 +288,12 @@ class Block(nn.Module):
         self.attn = Attention(hidden_size, n_ctx, config, scale)
         self.ln_2 = nn.LayerNorm(hidden_size, eps=config.layer_norm_epsilon)
         if config.add_cross_attention:
-            self.crossattention = Attention(hidden_size, n_ctx, config, scale, is_cross_attention=True)
-            self.ln_cross_attn = nn.LayerNorm(hidden_size, eps=config.layer_norm_epsilon)
+            self.crossattention = Attention(
+                hidden_size, n_ctx, config, scale, is_cross_attention=True
+            )
+            self.ln_cross_attn = nn.LayerNorm(
+                hidden_size, eps=config.layer_norm_epsilon
+            )
         self.mlp = MLP(inner_dim, config)
 
     def forward(
@@ -311,7 +336,9 @@ class Block(nn.Module):
             attn_output = cross_attn_outputs[0]
             # residual connection
             hidden_states = hidden_states + attn_output
-            outputs = outputs + cross_attn_outputs[1:]  # add cross attentions if we output attention weights
+            outputs = (
+                outputs + cross_attn_outputs[1:]
+            )  # add cross attentions if we output attention weights
 
         feed_forward_hidden_states = self.mlp(self.ln_2(hidden_states))
         # residual connection
@@ -429,7 +456,7 @@ GPT2_INPUTS_DOCSTRING = r"""
             Mask values selected in ``[0, 1]``:
 
             - 1 for tokens that are **not masked**,
-            - 0 for tokens that are **maked**.
+            - 0 for tokens that are **masked**.
 
             `What are attention masks? <../glossary.html#attention-mask>`__
         token_type_ids (:obj:`torch.LongTensor` of shape :obj:`(batch_size, input_ids_length)`, `optional`):
@@ -472,6 +499,48 @@ GPT2_INPUTS_DOCSTRING = r"""
             Whether or not to return a :class:`~transformers.file_utils.ModelOutput` instead of a plain tuple.
 """
 
+PARALLELIZE_DOCSTRING = r"""
+    Uses a device map to distribute attention modules of the model across several devices. If no device map is given, it
+    will evenly distribute blocks across all devices.
+    Args:
+        device_map (:obj:`Dict[int, list]`, optional, defaults to None):
+            A dictionary that maps attention modules to devices. Note that the embedding module and LMHead are
+            always automatically mapped to the first device (for esoteric reasons). That means that the first
+            device should have fewer attention modules mapped to it than other devices.
+
+            For reference, the gpt2 models have the following number of attention modules:
+
+                - gpt2: 12
+                - gpt2-medium: 24
+                - gpt2-large: 36
+                - gpt2-xl: 48
+
+    Example::
+        Here is an example of a device map on a machine with 4 GPUs using gpt2-xl, which has a total of 48 attention modules:
+
+            model = GPT2LMHeadModel.from_pretrained('gpt2-xl')
+            device_map = {0: [0, 1, 2, 3, 4, 5, 6, 7, 8],
+                          1: [9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21],
+                          2: [22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34],
+                          3: [35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47]}
+            model.parallelize(device_map)
+"""
+
+DEPARALLELIZE_DOCSTRING = r"""
+    Moves the model to cpu from a model parallel state.
+
+    Example::
+        On a 4 GPU machine with gpt2-large:
+
+        model = GPT2LMHeadModel.from_pretrained('gpt2-large')
+        device_map = {0: [0, 1, 2, 3, 4, 5, 6, 7],
+                    1: [8, 9, 10, 11, 12, 13, 14, 15],
+                    2: [16, 17, 18, 19, 20, 21, 22, 23],
+                    3: [24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35]}
+        model.parallelize(device_map) # Splits the model across several devices
+        model.deparallelize() # Put the model back on cpu and cleans memory by calling torch.cuda.empty_cache()
+"""
+
 
 @add_start_docstrings(
     "The bare GPT2 Model transformer outputting raw hidden-states without any specific head on top.",
@@ -484,11 +553,58 @@ class GPT2Model(GPT2PreTrainedModel):
         self.wte = nn.Embedding(config.vocab_size, config.n_embd)
         self.wpe = nn.Embedding(config.n_positions, config.n_embd)
         self.drop = nn.Dropout(config.embd_pdrop)
-        self.h = nn.ModuleList([Block(config.n_ctx, config, scale=True) for _ in range(config.n_layer)])
+        self.h = nn.ModuleList(
+            [Block(config.n_ctx, config, scale=True) for _ in range(config.n_layer)]
+        )
         self.ln_f = nn.LayerNorm(config.n_embd, eps=config.layer_norm_epsilon)
 
         self.init_weights()
 
+        # Model parallel
+        self.model_parallel = False
+        self.device_map = None
+
+    @add_start_docstrings(PARALLELIZE_DOCSTRING)
+    def parallelize(self, device_map=None):
+        # Check validity of device_map
+        self.device_map = (
+            get_device_map(len(self.h), range(torch.cuda.device_count()))
+            if device_map is None
+            else device_map
+        )
+        assert_device_map(self.device_map, len(self.h))
+
+        self.model_parallel = True
+        self.first_device = (
+            "cpu"
+            if "cpu" in self.device_map.keys()
+            else "cuda:" + str(min(self.device_map.keys()))
+        )
+        self.last_device = "cuda:" + str(max(self.device_map.keys()))
+        self.wte = self.wte.to(self.first_device)
+        self.wpe = self.wpe.to(self.first_device)
+        # Load onto devices
+        for k, v in self.device_map.items():
+            for block in v:
+                cuda_device = "cuda:" + str(k)
+                self.h[block] = self.h[block].to(cuda_device)
+        # ln_f to last
+        self.ln_f = self.ln_f.to(self.last_device)
+
+    @add_start_docstrings(DEPARALLELIZE_DOCSTRING)
+    def deparallelize(self):
+        self.model_parallel = False
+        self.device_map = None
+        self.first_device = "cpu"
+        self.last_device = "cpu"
+        self.wte = self.wte.to("cpu")
+        self.wpe = self.wpe.to("cpu")
+        for index in range(len(self.h)):
+            self.h[index] = self.h[index].to("cpu")
+
+        self.ln_f = self.ln_f.to("cpu")
+        torch.cuda.empty_cache()
+
     def get_input_embeddings(self):
         return self.wte
 
@@ -534,15 +650,25 @@ class GPT2Model(GPT2PreTrainedModel):
             past_key_values = kwargs.pop("past")
         assert kwargs == {}, f"Unexpected keyword arguments: {list(kwargs.keys())}."
 
-        output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
+        output_attentions = (
+            output_attentions
+            if output_attentions is not None
+            else self.config.output_attentions
+        )
         output_hidden_states = (
-            output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
+            output_hidden_states
+            if output_hidden_states is not None
+            else self.config.output_hidden_states
         )
         use_cache = use_cache if use_cache is not None else self.config.use_cache
-        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+        return_dict = (
+            return_dict if return_dict is not None else self.config.use_return_dict
+        )
 
         if input_ids is not None and inputs_embeds is not None:
-            raise ValueError("You cannot specify both input_ids and inputs_embeds at the same time")
+            raise ValueError(
+                "You cannot specify both input_ids and inputs_embeds at the same time"
+            )
         elif input_ids is not None:
             input_shape = input_ids.size()
             input_ids = input_ids.view(-1, input_shape[-1])
@@ -565,7 +691,12 @@ class GPT2Model(GPT2PreTrainedModel):
             past_length = past_key_values[0][0].size(-2)
         if position_ids is None:
             device = input_ids.device if input_ids is not None else inputs_embeds.device
-            position_ids = torch.arange(past_length, input_shape[-1] + past_length, dtype=torch.long, device=device)
+            position_ids = torch.arange(
+                past_length,
+                input_shape[-1] + past_length,
+                dtype=torch.long,
+                device=device,
+            )
             position_ids = position_ids.unsqueeze(0).view(-1, input_shape[-1])
 
         # Attention mask.
@@ -584,13 +715,17 @@ class GPT2Model(GPT2PreTrainedModel):
             # positions we want to attend and -10000.0 for masked positions.
             # Since we are adding it to the raw scores before the softmax, this is
             # effectively the same as removing these entirely.
-            attention_mask = attention_mask.to(dtype=next(self.parameters()).dtype)  # fp16 compatibility
+            attention_mask = attention_mask.to(dtype=self.dtype)  # fp16 compatibility
             attention_mask = (1.0 - attention_mask) * -10000.0
 
         # If a 2D ou 3D attention mask is provided for the cross-attention
         # we need to make broadcastabe to [batch_size, num_heads, seq_length, seq_length]
         if self.config.add_cross_attention and encoder_hidden_states is not None:
-            encoder_batch_size, encoder_sequence_length, _ = encoder_hidden_states.size()
+            (
+                encoder_batch_size,
+                encoder_sequence_length,
+                _,
+            ) = encoder_hidden_states.size()
             encoder_hidden_shape = (encoder_batch_size, encoder_sequence_length)
             if encoder_attention_mask is None:
                 encoder_attention_mask = torch.ones(encoder_hidden_shape, device=device)
@@ -620,15 +755,33 @@ class GPT2Model(GPT2PreTrainedModel):
         all_attentions = () if output_attentions else None
         all_hidden_states = () if output_hidden_states else None
         for i, (block, layer_past) in enumerate(zip(self.h, past_key_values)):
+            # Model parallel
+            if self.model_parallel:
+                torch.cuda.set_device(hidden_states.device)
+                # Ensure layer_past is on same device as hidden_states (might not be correct)
+                if layer_past is not None:
+                    layer_past = layer_past.to(hidden_states.device)
+                # Ensure that attention_mask is always on the same device as hidden_states
+                if attention_mask is not None:
+                    attention_mask = attention_mask.to(hidden_states.device)
+
+                if isinstance(head_mask, torch.Tensor):
+                    head_mask = head_mask.to(hidden_states.device)
+
             if output_hidden_states:
-                all_hidden_states = all_hidden_states + (hidden_states.view(*output_shape),)
+                all_hidden_states = all_hidden_states + (
+                    hidden_states.view(*output_shape),
+                )
 
             if getattr(self.config, "gradient_checkpointing", False):
 
                 def create_custom_forward(module):
                     def custom_forward(*inputs):
                         # checkpointing only works with tuple returns, not with lists
-                        return tuple(output for output in module(*inputs, use_cache, output_attentions))
+                        return tuple(
+                            output
+                            for output in module(*inputs, use_cache, output_attentions)
+                        )
 
                     return custom_forward
 
@@ -660,6 +813,12 @@ class GPT2Model(GPT2PreTrainedModel):
             if output_attentions:
                 all_attentions = all_attentions + (outputs[2],)
 
+            # Model Parallel: If it's the last layer for that device, put things on the next device
+            if self.model_parallel:
+                for k, v in self.device_map.items():
+                    if i == v[-1] and "cuda:" + str(k) != self.last_device:
+                        hidden_states = hidden_states.to("cuda:" + str(k + 1))
+
         hidden_states = self.ln_f(hidden_states)
 
         hidden_states = hidden_states.view(*output_shape)
@@ -668,7 +827,11 @@ class GPT2Model(GPT2PreTrainedModel):
             all_hidden_states = all_hidden_states + (hidden_states,)
 
         if not return_dict:
-            return tuple(v for v in [hidden_states, presents, all_hidden_states, all_attentions] if v is not None)
+            return tuple(
+                v
+                for v in [hidden_states, presents, all_hidden_states, all_attentions]
+                if v is not None
+            )
 
         return BaseModelOutputWithPast(
             last_hidden_state=hidden_states,
@@ -693,6 +856,29 @@ class GPT2LMHeadModel(GPT2PreTrainedModel):
 
         self.init_weights()
 
+        self.model_parallel = False
+
+    @add_start_docstrings(PARALLELIZE_DOCSTRING)
+    def parallelize(self, device_map=None):
+        self.device_map = (
+            get_device_map(len(self.transformer.h), range(torch.cuda.device_count()))
+            if device_map is None
+            else device_map
+        )
+        assert_device_map(self.device_map, len(self.transformer.h))
+
+        self.transformer.parallelize(self.device_map)
+        self.lm_head = self.lm_head.to(self.transformer.first_device)
+        self.model_parallel = True
+
+    @add_start_docstrings(DEPARALLELIZE_DOCSTRING)
+    def deparallelize(self):
+        self.transformer.deparallelize()
+        self.transformer = self.transformer.to("cpu")
+        self.lm_head = self.lm_head.to("cpu")
+        self.model_parallel = False
+        torch.cuda.empty_cache()
+
     def get_output_embeddings(self):
         return self.lm_head
 
@@ -747,7 +933,9 @@ class GPT2LMHeadModel(GPT2PreTrainedModel):
             )
             past_key_values = kwargs.pop("past")
         assert kwargs == {}, f"Unexpected keyword arguments: {list(kwargs.keys())}."
-        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+        return_dict = (
+            return_dict if return_dict is not None else self.config.use_return_dict
+        )
 
         transformer_outputs = self.transformer(
             input_ids,
@@ -766,6 +954,11 @@ class GPT2LMHeadModel(GPT2PreTrainedModel):
         )
         hidden_states = transformer_outputs[0]
 
+        # Set device for model parallelism
+        if self.model_parallel:
+            torch.cuda.set_device(self.transformer.first_device)
+            hidden_states = hidden_states.to(self.lm_head.weight.device)
+
         lm_logits = self.lm_head(hidden_states)
 
         loss = None
@@ -775,7 +968,9 @@ class GPT2LMHeadModel(GPT2PreTrainedModel):
             shift_labels = labels[..., 1:].contiguous()
             # Flatten the tokens
             loss_fct = CrossEntropyLoss()
-            loss = loss_fct(shift_logits.view(-1, shift_logits.size(-1)), shift_labels.view(-1))
+            loss = loss_fct(
+                shift_logits.view(-1, shift_logits.size(-1)), shift_labels.view(-1)
+            )
 
         if not return_dict:
             output = (lm_logits,) + transformer_outputs[1:]
@@ -823,7 +1018,9 @@ class GPT2DoubleHeadsModel(GPT2PreTrainedModel):
         }
 
     @add_start_docstrings_to_callable(GPT2_INPUTS_DOCSTRING)
-    @replace_return_docstrings(output_type=GPT2DoubleHeadsModelOutput, config_class=_CONFIG_FOR_DOC)
+    @replace_return_docstrings(
+        output_type=GPT2DoubleHeadsModelOutput, config_class=_CONFIG_FOR_DOC
+    )
     def forward(
         self,
         input_ids=None,
@@ -899,7 +1096,9 @@ class GPT2DoubleHeadsModel(GPT2PreTrainedModel):
             )
             past_key_values = kwargs.pop("past")
         assert kwargs == {}, f"Unexpected keyword arguments: {list(kwargs.keys())}."
-        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+        return_dict = (
+            return_dict if return_dict is not None else self.config.use_return_dict
+        )
 
         transformer_outputs = self.transformer(
             input_ids,
@@ -923,13 +1122,17 @@ class GPT2DoubleHeadsModel(GPT2PreTrainedModel):
         mc_loss = None
         if mc_labels is not None:
             loss_fct = CrossEntropyLoss()
-            mc_loss = loss_fct(mc_logits.view(-1, mc_logits.size(-1)), mc_labels.view(-1))
+            mc_loss = loss_fct(
+                mc_logits.view(-1, mc_logits.size(-1)), mc_labels.view(-1)
+            )
         lm_loss = None
         if labels is not None:
             shift_logits = lm_logits[..., :-1, :].contiguous()
             shift_labels = labels[..., 1:].contiguous()
             loss_fct = CrossEntropyLoss()
-            lm_loss = loss_fct(shift_logits.view(-1, shift_logits.size(-1)), shift_labels.view(-1))
+            lm_loss = loss_fct(
+                shift_logits.view(-1, shift_logits.size(-1)), shift_labels.view(-1)
+            )
 
         if not return_dict:
             output = (lm_logits, mc_logits) + transformer_outputs[1:]
@@ -946,3 +1149,127 @@ class GPT2DoubleHeadsModel(GPT2PreTrainedModel):
             hidden_states=transformer_outputs.hidden_states,
             attentions=transformer_outputs.attentions,
         )
+
+
+@add_start_docstrings(
+    """The GPT2 Model transformer with a sequence classification head on top
+    (linear layer).
+
+    :class:`~transformers.GPT2ForSequenceClassification` uses the last token in order to do the classification, as
+    other causal models (e.g. GPT-1) do.
+
+    Since it does classification on the last token, it requires to know the position of the last token.
+    If a :obj:`pad_token_id` is defined in the configuration, it finds the last token that is not a padding token
+    in each row. If no :obj:`pad_token_id` is defined, it simply takes the last value in each row of the batch.
+    Since it cannot guess the padding tokens when :obj:`inputs_embeds` are passed instead of :obj:`input_ids`, it
+    does the same (take the last value in each row of the batch).
+    """,
+    GPT2_START_DOCSTRING,
+)
+class GPT2ForSequenceClassification(GPT2PreTrainedModel):
+    authorized_missing_keys = [r"h\.\d+\.attn\.masked_bias", r"lm_head\.weight"]
+
+    def __init__(self, config):
+        super().__init__(config)
+        self.num_labels = config.num_labels
+        self.transformer = GPT2Model(config)
+        self.score = nn.Linear(config.n_embd, self.num_labels, bias=False)
+
+        self.init_weights()
+
+    @add_start_docstrings_to_callable(GPT2_INPUTS_DOCSTRING)
+    @add_code_sample_docstrings(
+        tokenizer_class=_TOKENIZER_FOR_DOC,
+        checkpoint="microsoft/dialogrpt",
+        output_type=SequenceClassifierOutputWithPast,
+        config_class=_CONFIG_FOR_DOC,
+    )
+    def forward(
+        self,
+        input_ids=None,
+        past_key_values=None,
+        attention_mask=None,
+        token_type_ids=None,
+        position_ids=None,
+        head_mask=None,
+        inputs_embeds=None,
+        labels=None,
+        use_cache=None,
+        output_attentions=None,
+        output_hidden_states=None,
+        return_dict=None,
+    ):
+        r"""
+        labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`):
+            Labels for computing the sequence classification/regression loss.
+            Indices should be in :obj:`[0, ..., config.num_labels - 1]`.
+            If :obj:`config.num_labels == 1` a regression loss is computed (Mean-Square loss),
+            If :obj:`config.num_labels > 1` a classification loss is computed (Cross-Entropy).
+        """
+        return_dict = (
+            return_dict if return_dict is not None else self.config.use_return_dict
+        )
+
+        transformer_outputs = self.transformer(
+            input_ids,
+            past_key_values=past_key_values,
+            attention_mask=attention_mask,
+            token_type_ids=token_type_ids,
+            position_ids=position_ids,
+            head_mask=head_mask,
+            inputs_embeds=inputs_embeds,
+            use_cache=use_cache,
+            output_attentions=output_attentions,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+        )
+        hidden_states = transformer_outputs[0]
+        logits = self.score(hidden_states)
+
+        if input_ids is not None:
+            batch_size, sequence_length = input_ids.shape[:2]
+        else:
+            batch_size, sequence_length = inputs_embeds.shape[:2]
+
+        assert (
+            self.config.pad_token_id is not None or batch_size == 1
+        ), "Cannot handle batch sizes > 1 if no padding token is defined."
+        if self.config.pad_token_id is None:
+            sequence_lengths = -1
+        else:
+            if input_ids is not None:
+                sequence_lengths = (
+                    torch.ne(input_ids, self.config.pad_token_id).sum(-1) - 1
+                )
+            else:
+                sequence_lengths = -1
+                logger.warning(
+                    f"{self.__class__.__name__} will not detect padding tokens in `inputs_embeds`. Results may be "
+                    f"unexpected if using padding tokens in conjuction with `inputs_embeds.`"
+                )
+
+        pooled_logits = logits[range(batch_size), sequence_lengths]
+
+        loss = None
+        if labels is not None:
+            if self.num_labels == 1:
+                #  We are doing regression
+                loss_fct = MSELoss()
+                loss = loss_fct(pooled_logits.view(-1), labels.view(-1))
+            else:
+                loss_fct = CrossEntropyLoss()
+                loss = loss_fct(
+                    pooled_logits.view(-1, self.num_labels), labels.view(-1)
+                )
+
+        if not return_dict:
+            output = (pooled_logits,) + transformer_outputs[1:]
+            return ((loss,) + output) if loss is not None else output
+
+        return SequenceClassifierOutputWithPast(
+            loss=loss,
+            logits=pooled_logits,
+            past_key_values=transformer_outputs.past_key_values,
+            hidden_states=transformer_outputs.hidden_states,
+            attentions=transformer_outputs.attentions,
+        )

src/transformers/modeling_longformer.py

@@ -1018,7 +1018,7 @@ LONGFORMER_INPUTS_DOCSTRING = r"""
             Mask values selected in ``[0, 1]``:
 
             - 1 for tokens that are **not masked**,
-            - 0 for tokens that are **maked**.
+            - 0 for tokens that are **masked**.
 
             `What are attention masks? <../glossary.html#attention-mask>`__
         global_attention_mask (:obj:`torch.FloatTensor` of shape :obj:`({0})`, `optional`):

src/transformers/modeling_lxmert.py

@@ -848,7 +848,7 @@ LXMERT_INPUTS_DOCSTRING = r"""
             Mask values selected in ``[0, 1]``:
 
             - 1 for tokens that are **not masked**,
-            - 0 for tokens that are **maked**.
+            - 0 for tokens that are **masked**.
 
             `What are attention masks? <../glossary.html#attention-mask>`__
         visual_attention_mask (:obj:`torch.FloatTensor` of shape :obj:`({0})`, `optional`):
@@ -856,7 +856,7 @@ LXMERT_INPUTS_DOCSTRING = r"""
             Mask values selected in ``[0, 1]``:
 
             - 1 for tokens that are **not masked**,
-            - 0 for tokens that are **maked**.
+            - 0 for tokens that are **masked**.
 
             `What are attention masks? <../glossary.html#attention-mask>`__
         token_type_ids (:obj:`torch.LongTensor` of shape :obj:`({0})`, `optional`):
@@ -964,7 +964,7 @@ class LxmertModel(LxmertPreTrainedModel):
         # Process the visual attention mask
         if visual_attention_mask is not None:
             extended_visual_attention_mask = visual_attention_mask.unsqueeze(1).unsqueeze(2)
-            extended_visual_attention_mask = extended_visual_attention_mask.to(dtype=next(self.parameters()).dtype)
+            extended_visual_attention_mask = extended_visual_attention_mask.to(dtype=self.dtype)
             extended_visual_attention_mask = (1.0 - extended_visual_attention_mask) * -10000.0
         else:
             extended_visual_attention_mask = None

src/transformers/modeling_mbart.py

@@ -27,5 +27,5 @@ class MBartForConditionalGeneration(BartForConditionalGeneration):
         >>> translation = tokenizer.batch_decode(translated_tokens, skip_special_tokens=True)[0]
         >>> assert translation == "Şeful ONU declară că nu există o soluţie militară în Siria"
     """
-
+    model_type = "mbart"
     config_class = MBartConfig

src/transformers/modeling_mmbt.py

@@ -123,7 +123,7 @@ MMBT_INPUTS_DOCSTRING = r"""
             Mask values selected in ``[0, 1]``:
 
             - 1 for tokens that are **not masked**,
-            - 0 for tokens that are **maked**.
+            - 0 for tokens that are **masked**.
 
             `What are attention masks? <../glossary.html#attention-mask>`__
         token_type_ids (`optional`) ``torch.LongTensor`` of shape ``(batch_size, sequence_length)``:
@@ -167,7 +167,7 @@ MMBT_INPUTS_DOCSTRING = r"""
             Mask values selected in ``[0, 1]``:
 
             - 1 for tokens that are **not masked**,
-            - 0 for tokens that are **maked**.
+            - 0 for tokens that are **masked**.
 
         output_attentions (:obj:`bool`, `optional`):
             Whether or not to return the attentions tensors of all attention layers. See ``attentions`` under returned

src/transformers/modeling_mobilebert.py

@@ -756,7 +756,7 @@ MOBILEBERT_INPUTS_DOCSTRING = r"""
             Mask values selected in ``[0, 1]``:
 
             - 1 for tokens that are **not masked**,
-            - 0 for tokens that are **maked**.
+            - 0 for tokens that are **masked**.
 
             `What are attention masks? <../glossary.html#attention-mask>`__
         token_type_ids (:obj:`torch.LongTensor` of shape :obj:`({0})`, `optional`):
@@ -792,7 +792,7 @@ MOBILEBERT_INPUTS_DOCSTRING = r"""
             Mask values selected in ``[0, 1]``:
 
             - 1 for tokens that are **not masked**,
-            - 0 for tokens that are **maked**.
+            - 0 for tokens that are **masked**.
 
         output_attentions (:obj:`bool`, `optional`):
             Whether or not to return the attentions tensors of all attention layers. See ``attentions`` under returned

src/transformers/modeling_openai.py

@@ -25,7 +25,7 @@ from typing import Optional, Tuple
 
 import torch
 import torch.nn as nn
-from torch.nn import CrossEntropyLoss
+from torch.nn import CrossEntropyLoss, MSELoss
 
 from .activations import gelu_new, swish
 from .configuration_openai import OpenAIGPTConfig
@@ -36,7 +36,7 @@ from .file_utils import (
     add_start_docstrings_to_callable,
     replace_return_docstrings,
 )
-from .modeling_outputs import BaseModelOutput, CausalLMOutput
+from .modeling_outputs import BaseModelOutput, CausalLMOutput, SequenceClassifierOutput
 from .modeling_utils import (
     Conv1D,
     PreTrainedModel,
@@ -360,7 +360,7 @@ OPENAI_GPT_INPUTS_DOCSTRING = r"""
             Mask values selected in ``[0, 1]``:
 
             - 1 for tokens that are **not masked**,
-            - 0 for tokens that are **maked**.
+            - 0 for tokens that are **masked**.
 
             `What are attention masks? <../glossary.html#attention-mask>`__
         token_type_ids (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`):
@@ -732,3 +732,113 @@ class OpenAIGPTDoubleHeadsModel(OpenAIGPTPreTrainedModel):
             hidden_states=transformer_outputs.hidden_states,
             attentions=transformer_outputs.attentions,
         )
+
+
+@add_start_docstrings(
+    """The Original OpenAI GPT Model transformer with a sequence classification head on top
+    (linear layer).
+    :class:`~transformers.OpenAIGPTForSequenceClassification` uses the last token in order to do the classification, as
+    other causal models (e.g. GPT-2) do.
+    Since it does classification on the last token, it requires to know the position of the last token.
+    If a :obj:`pad_token_id` is defined in the configuration, it finds the last token that is not a padding token
+    in each row. If no :obj:`pad_token_id` is defined, it simply takes the last value in each row of the batch.
+    Since it cannot guess the padding tokens when :obj:`inputs_embeds` are passed instead of :obj:`input_ids`, it
+    does the same (take the last value in each row of the batch).
+    """,
+    OPENAI_GPT_START_DOCSTRING,
+)
+class OpenAIGPTForSequenceClassification(OpenAIGPTPreTrainedModel):
+    def __init__(self, config):
+        super().__init__(config)
+        self.num_labels = config.num_labels
+        self.transformer = OpenAIGPTModel(config)
+        self.score = nn.Linear(config.n_embd, self.num_labels, bias=False)
+
+        self.init_weights()
+
+    @add_start_docstrings_to_callable(OPENAI_GPT_INPUTS_DOCSTRING)
+    @add_code_sample_docstrings(
+        tokenizer_class=_TOKENIZER_FOR_DOC,
+        checkpoint="openai-gpt",
+        output_type=SequenceClassifierOutput,
+        config_class=_CONFIG_FOR_DOC,
+    )
+    def forward(
+        self,
+        input_ids=None,
+        attention_mask=None,
+        token_type_ids=None,
+        position_ids=None,
+        head_mask=None,
+        inputs_embeds=None,
+        labels=None,
+        output_attentions=None,
+        output_hidden_states=None,
+        return_dict=None,
+    ):
+        r"""
+        labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`):
+            Labels for computing the sequence classification/regression loss.
+            Indices should be in :obj:`[0, ..., config.num_labels - 1]`.
+            If :obj:`config.num_labels == 1` a regression loss is computed (Mean-Square loss),
+            If :obj:`config.num_labels > 1` a classification loss is computed (Cross-Entropy).
+        """
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        transformer_outputs = self.transformer(
+            input_ids,
+            attention_mask=attention_mask,
+            token_type_ids=token_type_ids,
+            position_ids=position_ids,
+            head_mask=head_mask,
+            inputs_embeds=inputs_embeds,
+            output_attentions=output_attentions,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+        )
+
+        hidden_states = transformer_outputs[0]
+        logits = self.score(hidden_states)
+
+        if input_ids is not None:
+            batch_size, sequence_length = input_ids.shape[:2]
+        else:
+            batch_size, sequence_length = inputs_embeds.shape[:2]
+
+        assert (
+            self.config.pad_token_id is not None or batch_size == 1
+        ), "Cannot handle batch sizes > 1 if no padding token is defined."
+        if self.config.pad_token_id is None:
+            sequence_lengths = -1
+        else:
+            if input_ids is not None:
+                sequence_lengths = torch.ne(input_ids, self.config.pad_token_id).sum(-1) - 1
+            else:
+                sequence_lengths = -1
+                logger.warning(
+                    f"{self.__class__.__name__} will not detect padding tokens in `inputs_embeds`. Results may be "
+                    f"unexpected if using padding tokens in conjuction with `inputs_embeds.`"
+                )
+
+        pooled_logits = logits[range(batch_size), sequence_lengths]
+
+        loss = None
+        if labels is not None:
+            if self.num_labels == 1:
+                #  We are doing regression
+                loss_fct = MSELoss()
+                loss = loss_fct(pooled_logits.view(-1), labels.view(-1))
+            else:
+                loss_fct = CrossEntropyLoss()
+                loss = loss_fct(pooled_logits.view(-1, self.num_labels), labels.view(-1))
+
+        if not return_dict:
+            output = (pooled_logits,) + transformer_outputs[1:]
+            return ((loss,) + output) if loss is not None else output
+
+        return SequenceClassifierOutput(
+            loss=loss,
+            logits=pooled_logits,
+            hidden_states=transformer_outputs.hidden_states,
+            attentions=transformer_outputs.attentions,
+        )