Merge remote-tracking branch 'huggingface/master'

.gitignore

@@ -127,4 +127,7 @@ proc_data
 
 # examples
 runs
-examples/runs
+examples/runs
+
+# data
+data

README.md

@@ -12,7 +12,9 @@ The library currently contains PyTorch implementations, pre-trained model weight
 4. **[Transformer-XL](https://github.com/kimiyoung/transformer-xl)** (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.
 5. **[XLNet](https://github.com/zihangdai/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.
 6. **[XLM](https://github.com/facebookresearch/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.
-7. **[RoBERTa](https://github.com/pytorch/fairseq/tree/master/examples/roberta)** (from Facebook), a [Robustly Optimized BERT Pretraining Approach](https://arxiv.org/abs/1907.11692) by Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du et al.
+7. **[RoBERTa](https://github.com/pytorch/fairseq/tree/master/examples/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.
+8. **[DistilBERT](https://github.com/huggingface/pytorch-transformers/tree/master/examples/distillation)** (from HuggingFace), released together with the blogpost [Smaller, faster, cheaper, lighter: Introducing DistilBERT, a distilled version of BERT](https://medium.com/huggingface/distilbert-8cf3380435b5
+) by Victor Sanh, Lysandre Debut and Thomas Wolf.
 
 These implementations have been tested on several datasets (see the example scripts) and should match the performances of the original implementations (e.g. ~93 F1 on SQuAD for BERT Whole-Word-Masking, ~88 F1 on RocStories for OpenAI GPT, ~18.3 perplexity on WikiText 103 for Transformer-XL, ~0.916 Peason R coefficient on STS-B for XLNet). You can find more details on the performances in the Examples section of the [documentation](https://huggingface.co/pytorch-transformers/examples.html).
 
@@ -76,7 +78,7 @@ import torch
 from pytorch_transformers import *
 
 # PyTorch-Transformers has a unified API
-# for 6 transformer architectures and 27 pretrained weights.
+# for 7 transformer architectures and 30 pretrained weights.
 #          Model          | Tokenizer          | Pretrained weights shortcut
 MODELS = [(BertModel,       BertTokenizer,      'bert-base-uncased'),
           (OpenAIGPTModel,  OpenAIGPTTokenizer, 'openai-gpt'),
@@ -328,7 +330,7 @@ Breaking change in the `from_pretrained()`method:
 
 1. Models are now set in evaluation mode by default when instantiated with the `from_pretrained()` method. To train them don't forget to set them back in training mode (`model.train()`) to activate the dropout modules.
 
-2. The additional `*input` and `**kwargs` arguments supplied to the `from_pretrained()` method used to be directly passed to the underlying model's class `__init__()` method. They are now used to update the model configuration attribute instead which can break derived model classes build based on the previous `BertForSequenceClassification` examples. We are working on a way to mitigate this breaking change in [#866](https://github.com/huggingface/pytorch-transformers/pull/866) by forwarding the the model `__init__()` method (i) the provided positional arguments and (ii) the keyword arguments which do not match any configuratoin class attributes.
+2. The additional `*input` and `**kwargs` arguments supplied to the `from_pretrained()` method used to be directly passed to the underlying model's class `__init__()` method. They are now used to update the model configuration attribute instead which can break derived model classes build based on the previous `BertForSequenceClassification` examples. We are working on a way to mitigate this breaking change in [#866](https://github.com/huggingface/pytorch-transformers/pull/866) by forwarding the the model `__init__()` method (i) the provided positional arguments and (ii) the keyword arguments which do not match any configuration class attributes.
 
 Also, while not a breaking change, the serialization methods have been standardized and you probably should switch to the new method `save_pretrained(save_directory)` if you were using any other serialization method before.
 
@@ -393,8 +395,8 @@ for batch in train_data:
     loss = model(batch)
     loss.backward()
     torch.nn.utils.clip_grad_norm_(model.parameters(), max_grad_norm)  # Gradient clipping is not in AdamW anymore (so you can use amp without issue)
-    scheduler.step()
     optimizer.step()
+    scheduler.step()
     optimizer.zero_grad()
 ```
 

docs/source/examples.rst

@@ -12,8 +12,8 @@ Examples
      - How to use gradient-accumulation, multi-gpu training, distributed training, optimize on CPU and 16-bits training to train Bert models
    * - `Fine-tuning with BERT: running the examples <#fine-tuning-bert-examples>`_
      - Running the examples in `examples <https://github.com/huggingface/pytorch-pretrained-BERT/tree/master/examples>`_\ : ``extract_classif.py``\ , ``run_bert_classifier.py``\ , ``run_bert_squad.py`` and ``run_lm_finetuning.py``
-   * - `Fine-tuning with OpenAI GPT, Transformer-XL and GPT-2 <#fine-tuning>`_
-     - Running the examples in `examples <https://github.com/huggingface/pytorch-pretrained-BERT/tree/master/examples>`_\ : ``run_openai_gpt.py``\ , ``run_transfo_xl.py`` and ``run_gpt2.py``
+   * - `Fine-tuning with OpenAI GPT, Transformer-XL, GPT-2 as well as BERT and RoBERTa <#fine-tuning>`_
+     - Running the examples in `examples <https://github.com/huggingface/pytorch-pretrained-BERT/tree/master/examples>`_\ : ``run_openai_gpt.py``\ , ``run_transfo_xl.py``, ``run_gpt2.py`` and ``run_lm_finetuning.py``
    * - `Fine-tuning BERT-large on GPUs <#fine-tuning-bert-large>`_
      - How to fine tune ``BERT large``
 
@@ -68,7 +68,9 @@ GLUE results on dev set
 ~~~~~~~~~~~~~~~~~~~~~~~
 
 We get the following results on the dev set of GLUE benchmark with an uncased BERT base
-model. All experiments were run on a P100 GPU with a batch size of 32.
+model (`bert-base-uncased`). All experiments ran on 8 V100 GPUs with a total train batch size of 24. Some of 
+these tasks have a small dataset and training can lead to high variance in the results between different runs.
+We report the median on 5 runs (with different seeds) for each of the metrics.
 
 .. list-table::
    :header-rows: 1
@@ -78,31 +80,31 @@ model. All experiments were run on a P100 GPU with a batch size of 32.
      - Result
    * - CoLA
      - Matthew's corr.
-     - 57.29
+     - 55.75
    * - SST-2
      - accuracy
-     - 93.00
+     - 92.09
    * - MRPC
      - F1/accuracy
-     - 88.85/83.82
+     - 90.48/86.27
    * - STS-B
      - Pearson/Spearman corr.
-     - 89.70/89.37
+     - 89.03/88.64
    * - QQP
      - accuracy/F1
-     - 90.72/87.41
+     - 90.92/87.72
    * - MNLI
      - matched acc./mismatched acc.
-     - 83.95/84.39
+     - 83.74/84.06
    * - QNLI
      - accuracy
-     - 89.04
+     - 91.07
    * - RTE
      - accuracy
-     - 61.01
+     - 68.59
    * - WNLI
      - accuracy
-     - 53.52
+     - 43.66
 
 
 Some of these results are significantly different from the ones reported on the test set
@@ -382,7 +384,7 @@ Training with the previous hyper-parameters on a single GPU gave us the followin
 LM Fine-tuning
 ~~~~~~~~~~~~~~
 
-The data should be a text file in the same format as `sample_text.txt <./samples/sample_text.txt>`_  (one sentence per line, docs separated by empty line).
+The data should be a text file in the same format as `sample_text.txt <./pytorch_transformers/tests/fixtures/sample_text.txt/sample_text.txt>`_  (one sentence per line, docs separated by empty line).
 You can download an `exemplary training corpus <https://ext-bert-sample.obs.eu-de.otc.t-systems.com/small_wiki_sentence_corpus.txt>`_ generated from wikipedia articles and split into ~500k sentences with spaCy.
 Training one epoch on this corpus takes about 1:20h on 4 x NVIDIA Tesla P100 with ``train_batch_size=200`` and ``max_seq_length=128``\ :
 
@@ -393,12 +395,13 @@ Thank to the work of @Rocketknight1 and @tholor there are now **several scripts*
 OpenAI GPT, Transformer-XL and GPT-2: running the examples
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 
-We provide three examples of scripts for OpenAI GPT, Transformer-XL and OpenAI GPT-2 based on (and extended from) the respective original implementations:
+We provide three examples of scripts for OpenAI GPT, Transformer-XL, OpenAI GPT-2, BERT and RoBERTa based on (and extended from) the respective original implementations:
 
 
 * fine-tuning OpenAI GPT on the ROCStories dataset
 * evaluating Transformer-XL on Wikitext 103
 * unconditional and conditional generation from a pre-trained OpenAI GPT-2 model
+* fine-tuning GPT/GPT-2 on a causal language modeling task and BERT/RoBERTa on a masked language modeling task
 
 Fine-tuning OpenAI GPT on the RocStories dataset
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
@@ -452,7 +455,47 @@ Unconditional generation:
 
    python run_gpt2.py --unconditional
 
-The same option as in the original scripts are provided, please refere to the code of the example and the original repository of OpenAI.
+The same option as in the original scripts are provided, please refer to the code of the example and the original repository of OpenAI.
+
+
+Causal LM fine-tuning on GPT/GPT-2, Masked LM fine-tuning on BERT/RoBERTa
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Before running the following examples you should download the `WikiText-2 dataset <https://blog.einstein.ai/the-wikitext-long-term-dependency-language-modeling-dataset/>`__ and unpack it to some directory `$WIKITEXT_2_DATASET`
+The following results were obtained using the `raw` WikiText-2 (no tokens were replaced before the tokenization).
+
+This example fine-tunes GPT-2 on the WikiText-2 dataset. The loss function is a causal language modeling loss (perplexity).
+
+.. code-block:: bash
+    export WIKITEXT_2_DATASET=/path/to/wikitext_dataset
+
+    python run_lm_finetuning.py
+        --output_dir=output
+        --model_type=gpt2
+        --model_name_or_path=gpt2
+        --do_train
+        --train_data_file=$WIKITEXT_2_DATASET/wiki.train.raw
+        --do_eval
+        --eval_data_file=$WIKITEXT_2_DATASET/wiki.test.raw
+
+This takes about half an hour to train on a single K80 GPU and about one minute for the evaluation to run.
+It reaches a score of about 20 perplexity once fine-tuned on the dataset.
+
+This example fine-tunes RoBERTa on the WikiText-2 dataset. The loss function is a masked language modeling loss (masked perplexity).
+The `--mlm` flag is necessary to fine-tune BERT/RoBERTa on masked language modeling.
+
+.. code-block:: bash
+    export WIKITEXT_2_DATASET=/path/to/wikitext_dataset
+
+    python run_lm_finetuning.py
+        --output_dir=output
+        --model_type=roberta
+        --model_name_or_path=roberta-base
+        --do_train
+        --train_data_file=$WIKITEXT_2_DATASET/wiki.train.raw
+        --do_eval
+        --eval_data_file=$WIKITEXT_2_DATASET/wiki.test.raw
+        --mlm
 
 .. _fine-tuning-BERT-large:
 

docs/source/index.rst

@@ -48,3 +48,4 @@ The library currently contains PyTorch implementations, pre-trained model weight
     model_doc/xlm
     model_doc/xlnet
     model_doc/roberta
+    model_doc/distilbert

docs/source/model_doc/distilbert.rst

@@ -0,0 +1,43 @@
+DistilBERT
+----------------------------------------------------
+
+``DistilBertConfig``
+~~~~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: pytorch_transformers.DistilBertConfig
+    :members:
+
+
+``DistilBertTokenizer``
+~~~~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: pytorch_transformers.DistilBertTokenizer
+    :members:
+
+
+``DistilBertModel``
+~~~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: pytorch_transformers.DistilBertModel
+    :members:
+
+
+``DistilBertForMaskedLM``
+~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: pytorch_transformers.DistilBertForMaskedLM
+    :members:
+
+
+``DistilBertForSequenceClassification``
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: pytorch_transformers.DistilBertForSequenceClassification
+    :members:
+
+
+``DistilBertForQuestionAnswering``
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: pytorch_transformers.DistilBertForQuestionAnswering
+    :members: