Truncate when overflowing

.circleci/config.yml

@@ -3,6 +3,28 @@ orbs:
     gcp-gke: circleci/gcp-gke@1.0.4
     go: circleci/go@1.3.0
 
+commands:
+  skip-job-on-doc-only-changes:
+    description: "Do not continue this job and exit with success for PRs with only doc changes"
+    steps:
+
+      - run:
+          name: docs-only changes skip check
+          command: |
+            # pipeline.git.base_revision is not always defined, so only proceed if all external vars are defined
+            if test -n "<< pipeline.git.base_revision >>" && test -n "<< pipeline.git.revision >>" && test -n "$(git diff --name-only << pipeline.git.base_revision >>...<< pipeline.git.revision >>)"
+            then
+                if git diff --name-only << pipeline.git.base_revision >>...<< pipeline.git.revision >> | egrep -qv '\.(md|rst)$'
+                then
+                    echo "Non-docs were modified in this PR, proceeding normally"
+                else
+                    echo "Only docs were modified in this PR, quitting this job"
+                    # disable skipping for now, as circleCI's base_revision is inconsistent leading to invalid ranges
+                    # circleci step halt
+                fi
+            else
+                echo "Can't perform skipping check w/o base_revision defined, continuing the job"
+            fi
 
 # TPU REFERENCES
 references:
@@ -73,6 +95,7 @@ jobs:
         parallelism: 1
         steps:
             - checkout
+            - skip-job-on-doc-only-changes
             - restore_cache:
                   keys:
                       - v0.4-torch_and_tf-{{ checksum "setup.py" }}
@@ -100,6 +123,7 @@ jobs:
         parallelism: 1
         steps:
             - checkout
+            - skip-job-on-doc-only-changes
             - restore_cache:
                   keys:
                       - v0.4-torch-{{ checksum "setup.py" }}
@@ -127,6 +151,7 @@ jobs:
         parallelism: 1
         steps:
             - checkout
+            - skip-job-on-doc-only-changes
             - restore_cache:
                   keys:
                       - v0.4-tf-{{ checksum "setup.py" }}
@@ -153,6 +178,7 @@ jobs:
         parallelism: 1
         steps:
             - checkout
+            - skip-job-on-doc-only-changes
             - restore_cache:
                 keys:
                     - v0.4-flax-{{ checksum "setup.py" }}
@@ -179,6 +205,7 @@ jobs:
         parallelism: 1
         steps:
             - checkout
+            - skip-job-on-doc-only-changes
             - restore_cache:
                   keys:
                       - v0.4-torch-{{ checksum "setup.py" }}
@@ -206,6 +233,7 @@ jobs:
         parallelism: 1
         steps:
             - checkout
+            - skip-job-on-doc-only-changes
             - restore_cache:
                   keys:
                       - v0.4-tf-{{ checksum "setup.py" }}
@@ -230,6 +258,7 @@ jobs:
             RUN_CUSTOM_TOKENIZERS: yes
         steps:
             - checkout
+            - skip-job-on-doc-only-changes
             - restore_cache:
                   keys:
                       - v0.4-custom_tokenizers-{{ checksum "setup.py" }}
@@ -257,13 +286,14 @@ jobs:
         parallelism: 1
         steps:
             - checkout
+            - skip-job-on-doc-only-changes
             - restore_cache:
                   keys:
                       - v0.4-torch_examples-{{ checksum "setup.py" }}
                       - v0.4-{{ checksum "setup.py" }}
             - run: pip install --upgrade pip
             - run: pip install .[sklearn,torch,sentencepiece,testing]
-            - run: pip install -r examples/_tests_requirements.txt
+            - run: pip install -r examples/requirements.txt
             - save_cache:
                   key: v0.4-torch_examples-{{ checksum "setup.py" }}
                   paths:

.github/workflows/self-push.yml

@@ -50,7 +50,6 @@ jobs:
           pip install --upgrade pip
           pip install .[torch,sklearn,testing,onnxruntime,sentencepiece]
           pip install git+https://github.com/huggingface/datasets
-          pip install pandas torch-scatter -f https://pytorch-geometric.com/whl/torch-$(python -c "import torch; print(''.join(torch.__version__)")+$(python -c "import torch; print(''.join(torch.version.cuda.split('.')))").html
 
       - name: Are GPUs recognized by our DL frameworks
         run: |
@@ -188,7 +187,6 @@ jobs:
           pip install --upgrade pip
           pip install .[torch,sklearn,testing,onnxruntime,sentencepiece]
           pip install git+https://github.com/huggingface/datasets
-          pip install pandas torch-scatter -f https://pytorch-geometric.com/whl/torch-$(python -c "import torch; print(''.join(torch.__version__)")+$(python -c "import torch; print(''.join(torch.version.cuda.split('.')))").html
 
       - name: Are GPUs recognized by our DL frameworks
         run: |

docs/source/internal/generation_utils.rst

@@ -32,9 +32,6 @@ generation.
 .. autoclass:: transformers.LogitsProcessorList
     :members: __call__
 
-.. autoclass:: transformers.LogitsWarper
-    :members: __call__
-
 .. autoclass:: transformers.MinLengthLogitsProcessor
     :members: __call__
 
@@ -70,10 +67,3 @@ BeamSearch
 
 .. autoclass:: transformers.BeamSearchScorer
     :members: process, finalize
-
-Utilities
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-
-.. autofunction:: transformers.top_k_top_p_filtering
-
-.. autofunction:: transformers.tf_top_k_top_p_filtering

docs/source/internal/trainer_utils.rst

@@ -22,8 +22,6 @@ Utilities
 
 .. autoclass:: transformers.EvalPrediction
 
-.. autoclass:: transformers.EvaluationStrategy
-
 .. autofunction:: transformers.set_seed
 
 .. autofunction:: transformers.torch_distributed_zero_first
@@ -34,15 +32,8 @@ Callbacks internals
 
 .. autoclass:: transformers.trainer_callback.CallbackHandler
 
-
 Distributed Evaluation
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
 .. autoclass:: transformers.trainer_pt_utils.DistributedTensorGatherer
     :members:
-
-
-Distributed Evaluation
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-
-.. autoclass:: transformers.HfArgumentParser

docs/source/main_classes/optimizer_schedules.rst

@@ -74,10 +74,6 @@ Learning Rate Schedules (Pytorch)
     :target: /imgs/warmup_linear_schedule.png
     :alt:
 
-
-.. autofunction:: transformers.get_polynomial_decay_schedule_with_warmup
-
-
 Warmup (TensorFlow)
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 

docs/source/main_classes/pipelines.rst

@@ -73,9 +73,8 @@ FillMaskPipeline
 NerPipeline
 =======================================================================================================================
 
-.. autoclass:: transformers.NerPipeline
-
-See :class:`~transformers.TokenClassificationPipeline` for all details.
+This class is an alias of the :class:`~transformers.TokenClassificationPipeline` defined below. Please refer to that
+pipeline for documentation and usage examples.
 
 QuestionAnsweringPipeline
 =======================================================================================================================
@@ -119,13 +118,6 @@ TokenClassificationPipeline
     :special-members: __call__
     :members:
 
-TranslationPipeline
-=======================================================================================================================
-
-.. autoclass:: transformers.TranslationPipeline
-    :special-members: __call__
-    :members:
-
 ZeroShotClassificationPipeline
 =======================================================================================================================
 

docs/source/model_doc/albert.rst

@@ -60,13 +60,6 @@ AlbertTokenizer
         create_token_type_ids_from_sequences, save_vocabulary
 
 
-AlbertTokenizerFast
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-
-.. autoclass:: transformers.AlbertTokenizerFast
-    :members:
-
-
 Albert specific outputs
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 

docs/source/model_doc/auto.rst

@@ -175,10 +175,3 @@ TFAutoModelForQuestionAnswering
 
 .. autoclass:: transformers.TFAutoModelForQuestionAnswering
     :members:
-
-
-FlaxAutoModel
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-
-.. autoclass:: transformers.FlaxAutoModel
-    :members:

docs/source/model_doc/bart.rst

@@ -98,12 +98,6 @@ BartTokenizer
     :members:
 
 
-BartTokenizerFast
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-
-.. autoclass:: transformers.BartTokenizerFast
-    :members:
-
 
 BartModel
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

docs/source/model_doc/barthez.rst

@@ -51,9 +51,3 @@ BarthezTokenizer
 .. autoclass:: transformers.BarthezTokenizer
     :members:
 
-
-BarthezTokenizerFast
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-
-.. autoclass:: transformers.BarthezTokenizerFast
-    :members:

docs/source/model_doc/bert.rst

@@ -207,10 +207,3 @@ FlaxBertModel
 
 .. autoclass:: transformers.FlaxBertModel
     :members: __call__
-
-
-FlaxBertForMaskedLM
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-
-.. autoclass:: transformers.FlaxBertForMaskedLM
-    :members: __call__

docs/source/model_doc/camembert.rst

@@ -54,13 +54,6 @@ CamembertTokenizer
         create_token_type_ids_from_sequences, save_vocabulary
 
 
-CamembertTokenizerFast
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-
-.. autoclass:: transformers.CamembertTokenizerFast
-    :members:
-
-
 CamembertModel
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 

docs/source/model_doc/layoutlm.rst

@@ -57,13 +57,6 @@ LayoutLMTokenizer
     :members:
 
 
-LayoutLMTokenizerFast
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-
-.. autoclass:: transformers.LayoutLMTokenizerFast
-    :members:
-
-
 LayoutLMModel
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 

docs/source/model_doc/mbart.rst

@@ -90,13 +90,6 @@ MBartTokenizer
     :members: build_inputs_with_special_tokens, prepare_seq2seq_batch
 
 
-MBartTokenizerFast
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-
-.. autoclass:: transformers.MBartTokenizerFast
-    :members:
-
-
 MBartForConditionalGeneration
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 

docs/source/model_doc/mpnet.rst

@@ -1,15 +1,3 @@
-.. 
-    Copyright 2020 The HuggingFace Team. All rights reserved.
-
-    Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
-    the License. You may obtain a copy of the License at
-
-        http://www.apache.org/licenses/LICENSE-2.0
-
-    Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
-    an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
-    specific language governing permissions and limitations under the License.
-
 MPNet
 -----------------------------------------------------------------------------------------------------------------------
 

docs/source/model_doc/mt5.rst

@@ -37,22 +37,6 @@ MT5Config
     :members:
 
 
-MT5Tokenizer
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-
-.. autoclass:: transformers.MT5Tokenizer
-
-See :class:`~transformers.T5Tokenizer` for all details.
-
-
-MT5TokenizerFast
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-
-.. autoclass:: transformers.MT5TokenizerFast
-
-See :class:`~transformers.T5TokenizerFast` for all details.
-
-
 MT5Model
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 

docs/source/model_doc/pegasus.rst

@@ -112,13 +112,6 @@ warning: ``add_tokens`` does not work at the moment.
     :members: __call__, prepare_seq2seq_batch
 
 
-PegasusTokenizerFast
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-
-.. autoclass:: transformers.PegasusTokenizerFast
-    :members:
-
-
 PegasusForConditionalGeneration
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 

docs/source/model_doc/reformer.rst

@@ -163,13 +163,6 @@ ReformerTokenizer
     :members: save_vocabulary
 
 
-ReformerTokenizerFast
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-
-.. autoclass:: transformers.ReformerTokenizerFast
-    :members:
-
-
 ReformerModel
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 

docs/source/model_doc/t5.rst

@@ -107,13 +107,6 @@ T5Tokenizer
         create_token_type_ids_from_sequences, prepare_seq2seq_batch, save_vocabulary
 
 
-T5TokenizerFast
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-
-.. autoclass:: transformers.T5TokenizerFast
-    :members:
-
-
 T5Model
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 

docs/source/model_doc/tapas.rst

@@ -11,11 +11,9 @@ answering questions about tabular data. Compared to BERT, TAPAS uses relative po
 that encode tabular structure. TAPAS is pre-trained on the masked language modeling (MLM) objective on a large dataset
 comprising millions of tables from English Wikipedia and corresponding texts. For question answering, TAPAS has 2 heads
 on top: a cell selection head and an aggregation head, for (optionally) performing aggregations (such as counting or
-summing) among selected cells. TAPAS has been fine-tuned on several datasets: `SQA
-<https://www.microsoft.com/en-us/download/details.aspx?id=54253>`_ (Sequential Question Answering by Microsoft), `WTQ
-<https://github.com/ppasupat/WikiTableQuestions>`__ (Wiki Table Questions by Stanford University) and `WikiSQL
-<https://github.com/salesforce/WikiSQL>`__ (by Salesforce). It achieves state-of-the-art on both SQA and WTQ, while
-having comparable performance to SOTA on WikiSQL, with a much simpler architecture.
+summing) among selected cells. TAPAS has been fine-tuned on several datasets: SQA (Sequential Question Answering by
+Microsoft), WTQ (Wiki Table Questions by Stanford University) and WikiSQL (by Salesforce). It achieves state-of-the-art
+on both SQA and WTQ, while having comparable performance to SOTA on WikiSQL, with a much simpler architecture.
 
 The abstract from the paper is the following:
 
@@ -32,17 +30,17 @@ improving state-of-the-art accuracy on SQA from 55.1 to 67.2 and performing on p
 and WIKITQ, but with a simpler model architecture. We additionally find that transfer learning, which is trivial in our
 setting, from WIKISQL to WIKITQ, yields 48.7 accuracy, 4.2 points above the state-of-the-art.*
 
-In addition, the authors have further pre-trained TAPAS to recognize **table entailment**, by creating a balanced
-dataset of millions of automatically created training examples which are learned in an intermediate step prior to
-fine-tuning. The authors of TAPAS call this further pre-training intermediate pre-training (since TAPAS is first
-pre-trained on MLM, and then on another dataset). They found that intermediate pre-training further improves
-performance on SQA, achieving a new state-of-the-art as well as state-of-the-art on `TabFact
-<https://github.com/wenhuchen/Table-Fact-Checking>`__, a large-scale dataset with 16k Wikipedia tables for table
-entailment (a binary classification task). For more details, see their follow-up paper: `Understanding tables with
-intermediate pre-training <https://www.aclweb.org/anthology/2020.findings-emnlp.27/>`__ by Julian Martin Eisenschlos,
-Syrine Krichene and Thomas Müller.
+In addition, the authors have further pre-trained TAPAS to recognize table entailment, by creating a balanced dataset
+of millions of automatically created training examples which are learned in an intermediate step prior to fine-tuning.
+The authors of TAPAS call this further pre-training intermediate pre-training (since TAPAS is first pre-trained on MLM,
+and then on another dataset). They found that intermediate pre-training further improves performance on SQA, achieving
+a new state-of-the-art as well as state-of-the-art on TabFact, a large-scale dataset with 16k Wikipedia tables for
+table entailment (a binary classification task). For more details, see their follow-up paper: `Understanding tables
+with intermediate pre-training <https://arxiv.org/abs/2010.00571>`__ by Julian Martin Eisenschlos, Syrine Krichene and
+Thomas Müller.
 
-The original code can be found `here <https://github.com/google-research/tapas>`__.
+The original code can be found `here <https://github.com/google-research/tapas>`__. This page also includes links to
+the datasets (SQA, WTQ, WikiSQL and TabFact).
 
 Tips:
 
@@ -52,18 +50,17 @@ Tips:
   sequences without running out of embeddings. This is reflected in the ``reset_position_index_per_cell`` parameter of
   :class:`~transformers.TapasConfig`, which is set to ``True`` by default. The latest versions of the models available
   in the `model hub <https://huggingface.co/models?search=tapas>`_ all use relative position embeddings. You can still
-  use the ones with absolute position embeddings by passing in a certain version when calling the
-  ``.from_pretrained()`` method as explained in the model cards. Note that it's usually advised to pad the inputs on
-  the right rather than the left.
+  use the ones with absolute position embeddings by passing in a certain version when calling the `.from_pretrained`
+  method as explained in the model cards. Note that it's usually advised to pad the inputs on the right rather than the
+  left.
 - TAPAS is based on BERT, so ``TAPAS-base`` for example corresponds to a ``BERT-base`` architecture. Of course,
-  TAPAS-large will result in the best performance (the results reported in the paper are from TAPAS-large). Results of
+  TAPAS-large will result in the best performance (the results reported in the paper are from TAPAS-large). Metrics of
   the various sized models are shown on the `original Github repository <https://github.com/google-research/tapas>`_.
 - TAPAS has checkpoints fine-tuned on SQA, which are capable of answering questions related to a table in a
   conversational set-up. This means that you can ask follow-up questions such as "what is his age?" related to the
   previous question. Note that the forward pass of TAPAS is a bit different in case of a conversational set-up: in that
-  case, you have to feed every table-question pair one by one to the model, such that the `prev_labels` token type ids
-  can be overwritten by the predicted `labels` of the model to the previous question. See "Usage" section for more
-  info.
+  case, you have to feed every training example one by one to the model, such that the `prev_labels` token type ids can
+  be overwritten by the predicted `label_ids` of the model to the previous question. See "Usage" section for more info.
 - TAPAS is similar to BERT and therefore relies on the masked language modeling (MLM) objective. It is therefore
   efficient at predicting masked tokens and at NLU in general, but is not optimal for text generation. Models trained
   with a causal language modeling (CLM) objective are better in that regard.
@@ -74,7 +71,9 @@ Usage: fine-tuning
 
 Here we explain how you can fine-tune :class:`~transformers.TapasForQuestionAnswering` on your own dataset.
 
-**STEP 1: Choose one of the 3 ways in which you can use TAPAS - or experiment**
+=========================================================================== STEP 1: Choose one of the 3 ways in which
+you can use TAPAS - or experiment
+=======================================================================================================================
 
 Basically, there are 3 different ways in which one can fine-tune :class:`~transformers.TapasForQuestionAnswering`,
 corresponding to the different datasets on which Tapas was fine-tuned:
@@ -94,7 +93,7 @@ corresponding to the different datasets on which Tapas was fine-tuned:
 To summarize:
 
 +------------------------------------+----------------------+-------------------------------------------------------------------------------------------------------------------+
-| **Task**                           | **Example dataset**  | **Description**                                                                                                   |
+| **Task**                           | **Example datasets** | **Description**                                                                                                   |
 +------------------------------------+----------------------+-------------------------------------------------------------------------------------------------------------------+
 | Conversational                     | SQA                  | Conversational, only cell selection questions                                                                     |
 +------------------------------------+----------------------+-------------------------------------------------------------------------------------------------------------------+
@@ -103,44 +102,15 @@ To summarize:
 | Strong supervision for aggregation | WikiSQL-supervised   | Questions might involve aggregation, and the model must learn this given the gold aggregation operator            |
 +------------------------------------+----------------------+-------------------------------------------------------------------------------------------------------------------+
 
-Initializing a model with a pre-trained base and randomly initialized classification heads from the model hub can be
-done as follows (be sure to have installed the `torch-scatter dependency <https://github.com/rusty1s/pytorch_scatter>`_
-for your environment):
+Initializing a model with a pre-trained base and randomly initialized classification heads from the model hub is as
+easy as:
 
 .. code-block::
 
-        >>> from transformers import TapasConfig, TapasForQuestionAnswering
+        >>> from transformers import TapasForQuestionAnswering
 
-        >>> # for example, the base sized model with default SQA configuration
-        >>> model = TapasForQuestionAnswering.from_pretrained('google/tapas-base')
-
-        >>> # or, the base sized model with WTQ configuration
-        >>> config = TapasConfig(
-        ...            num_aggregation_labels = 4,
-        ...            use_answer_as_supervision = True,
-        ...            answer_loss_cutoff = 0.664694,
-        ...            cell_selection_preference = 0.207951,
-        ...            huber_loss_delta = 0.121194,
-        ...            init_cell_selection_weights_to_zero = True,
-        ...            select_one_column = True,
-        ...            allow_empty_column_selection = False,
-        ...            temperature = 0.0352513,
-        ... )
-        >>> model = TapasForQuestionAnswering.from_pretrained('google/tapas-base', config=config)
-
-        >>> # or, the base sized model with WikiSQL configuration
-        >>> config = TapasConfig(
-        ...            num_aggregation_labels = 4
-        ...            use_answer_as_supervision = False
-        ...            answer_loss_cutoff = 36.4519
-        ...            cell_selection_preference = 0.903421
-        ...            huber_loss_delta = 222.088
-        ...            init_cell_selection_weights_to_zero = True
-        ...            select_one_column = True
-        ...            allow_empty_column_selection = True
-        ...            temperature = 0.763141
-        ... )
-        >>> model = TapasForQuestionAnswering.from_pretrained('google/tapas-base', config=config)
+        >>> # for example, the base sized model 
+        >>> model = TapasForQuestionAnswering.from_pretrained('google/tapas-base-uncased')
 
 
 Of course, you don't necessarily have to follow one of these three ways in which TAPAS was fine-tuned. You can also
@@ -156,7 +126,7 @@ way. Here's an example:
         >>> # you can initialize the classification heads any way you want (see docs of TapasConfig)
         >>> config = TapasConfig(num_aggregation_labels=3, average_logits_per_cell=True, select_one_column=False)
         >>> # initializing the pre-trained base sized model with our custom classification heads
-        >>> model = TapasForQuestionAnswering.from_pretrained('google/tapas-base', config=config)
+        >>> model = TapasForQuestionAnswering.from_pretrained('google/tapas-base-uncased', config=config)
 
 What you can also do is start from an already fine-tuned checkpoint. A note here is that the already fine-tuned
 checkpoint on WTQ has some issues due to the L2-loss which is somewhat brittle. See `here
@@ -165,7 +135,8 @@ checkpoint on WTQ has some issues due to the L2-loss which is somewhat brittle.
 For a list of all pre-trained and fine-tuned TAPAS checkpoints available in the HuggingFace model hub, see `here
 <https://huggingface.co/models?search=tapas>`__.
 
-**STEP 2: Prepare your data in the SQA format**
+=========================================== STEP 2: Prepare your data in the SQA format
+=======================================================================================================================
 
 Second, no matter what you picked above, you should prepare your dataset in the `SQA format
 <https://www.microsoft.com/en-us/download/details.aspx?id=54253>`__. This format is a TSV/CSV file with the following
@@ -192,7 +163,10 @@ SQA format. The author explains this `here
 are not perfect (the ``answer_coordinates`` and ``float_answer`` fields are populated based on the ``answer_text``),
 meaning that WTQ and WikiSQL results could actually be improved.
 
-**STEP 3: Convert your data into PyTorch tensors using TapasTokenizer**
+
+========================================================================================== STEP 3: Convert your data
+into PyTorch tensors using :class:`~transformers.TapasTokenizer`
+=======================================================================================================================
 
 Third, given that you've prepared your data in this TSV/CSV format (and corresponding CSV files containing the tabular
 data), you can then use :class:`~transformers.TapasTokenizer` to convert table-question pairs into :obj:`input_ids`,
@@ -202,37 +176,37 @@ data), you can then use :class:`~transformers.TapasTokenizer` to convert table-q
 +------------------------------------+----------------------------------------------------------------------------------------------+
 | **Task**                           | **Required inputs**                                                                          |
 +------------------------------------+----------------------------------------------------------------------------------------------+
-| Conversational                     | ``input_ids``, ``attention_mask``, ``token_type_ids``, ``labels``                            |
+| Conversational                     | ``input_ids``, ``attention_mask``, ``token_type_ids``, ``label_ids``                         |
 +------------------------------------+----------------------------------------------------------------------------------------------+
-| Weak supervision for aggregation   | ``input_ids``, ``attention_mask``, ``token_type_ids``, ``labels``, ``numeric_values``,       |
+| Weak supervision for aggregation   | ``input_ids``, ``attention_mask``, ``token_type_ids``, ``label_ids``, ``numeric_values``,    |
 |                                    | ``numeric_values_scale``, ``float_answer``                                                   |
 +------------------------------------+----------------------------------------------------------------------------------------------+
-| Strong supervision for aggregation | ``input ids``, ``attention mask``, ``token type ids``, ``labels``, ``aggregation_labels``    |
+| Strong supervision for aggregation | ``input ids``, ``attention mask``, ``token type ids``, ``label ids``, ``aggregation_labels`` |
 +------------------------------------+----------------------------------------------------------------------------------------------+
 
-:class:`~transformers.TapasTokenizer` creates the ``labels``, ``numeric_values`` and ``numeric_values_scale`` based on
-the ``answer_coordinates`` and ``answer_text`` columns of the TSV file. The ``float_answer`` and ``aggregation_labels``
-are already in the TSV file of step 2. Here's an example:
+:class:`~transformers.TapasTokenizer` creates the ``label_ids``, ``numeric_values`` and ``numeric_values_scale`` based
+on the ``answer_coordinates`` and ``answer_text`` columns of the TSV file. The ``float_answer`` and
+``aggregation_labels`` are already in the TSV file of step 2. Here's an example:
 
 .. code-block::
 
         >>> from transformers import TapasTokenizer
         >>> import pandas as pd
 
-        >>> model_name = 'google/tapas-base'
+        >>> model_name = 'google/tapas-base-uncased'
         >>> tokenizer = TapasTokenizer.from_pretrained(model_name)
 
         >>> data = {'Actors': ["Brad Pitt", "Leonardo Di Caprio", "George Clooney"], 'Number of movies': ["87", "53", "69"]}
         >>> queries = ["What is the name of the first actor?", "How many movies has George Clooney played in?", "What is the total number of movies?"]
-        >>> answer_coordinates = [[(0, 0)], [(2, 1)], [(0, 1), (1, 1), (2, 1)]]
+        >>> answer_coordinates = [[(0, 0)], [(1, 0)], [(0, 2), (1, 2), (2, 2)]]
         >>> answer_text = [["Brad Pitt"], ["69"], ["209"]]
-        >>> table = pd.DataFrame.from_dict(data)
+        >>> table = pd.Dataframe(data)
         >>> inputs = tokenizer(table=table, queries=queries, answer_coordinates=answer_coordinates, answer_text=answer_text, padding='max_length', return_tensors='pt')
         >>> inputs
         {'input_ids': tensor([[ ... ]]), 'attention_mask': tensor([[...]]), 'token_type_ids': tensor([[[...]]]),
-        'numeric_values': tensor([[ ... ]]), 'numeric_values_scale: tensor([[ ... ]]), labels: tensor([[ ... ]])}
+        'numeric_values': tensor([[ ... ]]), 'numeric_values_scale: tensor([[ ... ]]), label_ids: tensor([[ ... ]])}
 
-Note that :class:`~transformers.TapasTokenizer` expects the data of the table to be **text-only**. You can use
+Note that :class:`~transformers.TapasTokenizer` expects the data of the table to be text-only. You can use
 ``.astype(str)`` on a dataframe to turn it into text-only data. Of course, this only shows how to encode a single
 training example. It is advised to create a PyTorch dataset and a corresponding dataloader:
 
@@ -251,18 +225,15 @@ training example. It is advised to create a PyTorch dataset and a corresponding
         ...
         ...     def __getitem__(self, idx):
         ...         item = data.iloc[idx]
-        ...         table = pd.read_csv(table_csv_path + item.table_file).astype(str) # be sure to make your table data text only
+        ...         table = pd.read_csv(table_csv_path + item.table_file).astype(str)
         ...         encoding = self.tokenizer(table=table, 
-        ...                                   queries=item.question, 
-        ...                                   answer_coordinates=item.answer_coordinates, 
-        ...                                   answer_text=item.answer_text,
-        ...                                   truncation=True,
-        ...                                   padding="max_length",
-        ...                                   return_tensors="pt"
+        ...                                 queries=item.question, 
+        ...                                 answer_coordinates=item.answer_coordinates, 
+        ...                                 answer_text=item.answer_text,
+        ...                                 padding="max_length",
+        ...                                 return_tensors="pt"
         ...         )
-        ...         # remove the batch dimension which the tokenizer adds by default
-        ...         encoding = {key: val.squeeze(0) for key, val in encoding.items()}
-        ...         # add the float_answer which is also required (weak supervision for aggregation case)
+        ...         # we add the float_answer which is also required (weak supervision for aggregation)
         ...         encoding["float_answer"] = torch.tensor(item.float_answer) 
         ...         return encoding
         ...
@@ -279,48 +250,28 @@ together the ``queries``, ``answer_coordinates`` and ``answer_text`` per table (
 index) and batch encode each table with its questions. This will make sure that the ``prev_labels`` token types (see
 docs of :class:`~transformers.TapasTokenizer`) are set correctly.
 
-**STEP 4: Train (fine-tune) TapasForQuestionAnswering**
+=================================================== STEP 4: Train (fine-tune) TapasForQuestionAnswering
+=======================================================================================================================
 
-You can then fine-tune :class:`~transformers.TapasForQuestionAnswering` using native PyTorch as follows (shown here for
-the weak supervision for aggregation case):
+You can then fine-tune :class:`~transformers.TapasForQuestionAnswering` using native PyTorch as follows:
 
 .. code-block::
 
-        >>> from transformers import TapasConfig, TapasForQuestionAnswering, AdamW
+        >>> from transformers import TapasForQuestionAnswering
 
-        >>> # this is the default WTQ configuration
-        >>> config = TapasConfig(
-        ...            num_aggregation_labels = 4,
-        ...            use_answer_as_supervision = True,
-        ...            answer_loss_cutoff = 0.664694,
-        ...            cell_selection_preference = 0.207951,
-        ...            huber_loss_delta = 0.121194,
-        ...            init_cell_selection_weights_to_zero = True,
-        ...            select_one_column = True,
-        ...            allow_empty_column_selection = False,
-        ...            temperature = 0.0352513,
-        ... )
-        >>> model = TapasForQuestionAnswering.from_pretrained("google/tapas-base", config=config)
-
-        >>> optimizer = AdamW(model.parameters(), lr=5e-5)
+        >>> model = TapasForQuestionAnswering.from_pretrained("google/tapas-base-uncased")
 
         >>> for epoch in range(2):  # loop over the dataset multiple times
         ...    for idx, batch in enumerate(train_dataloader):
         ...         # get the inputs; 
-        ...         input_ids = batch["input_ids"]
-        ...         attention_mask = batch["attention_mask"]
-        ...         token_type_ids = batch["token_type_ids"]
-        ...         labels = batch["labels"]
-        ...         numeric_values = batch["numeric_values"]
-        ...         numeric_values_scale = batch["numeric_values_scale"]
-        ...         float_answer = batch["float_answer"]
+        ...         input_ids, attention_mask, token_type_ids, label_ids, numeric_values, numeric_values_scale, float_answer = batch
 
         ...         # zero the parameter gradients
         ...         optimizer.zero_grad()
 
         ...         # forward + backward + optimize
         ...         outputs = model(input_ids=input_ids, attention_mask=attention_mask, token_type_ids=token_type_ids, 
-        ...                        labels=labels, numeric_values=numeric_values, numeric_values_scale=numeric_values_scale, 
+        ...                        label_ids=label_ids, numeric_values=numeric_values, numeric_values_scale=numeric_values_scale, 
         ...                        float_answer=float_answer)
         ...         loss = outputs.loss
         ...         loss.backward()
@@ -344,19 +295,19 @@ of that:
         >>> from transformers import TapasTokenizer, TapasForQuestionAnswering
         >>> import pandas as pd 
 
-        >>> model_name = 'google/tapas-base-finetuned-wtq'
+        >>> model_name = 'google/tapas-base-uncased-finetuned-wtq'
         >>> model = TapasForQuestionAnswering.from_pretrained(model_name)
         >>> tokenizer = TapasTokenizer.from_pretrained(model_name)
 
         >>> data = {'Actors': ["Brad Pitt", "Leonardo Di Caprio", "George Clooney"], 'Number of movies': ["87", "53", "69"]}
         >>> queries = ["What is the name of the first actor?", "How many movies has George Clooney played in?", "What is the total number of movies?"]
-        >>> table = pd.DataFrame.from_dict(data)
+        >>> table = pd.Dataframe(data)
         >>> inputs = tokenizer(table=table, queries=queries, padding='max_length', return_tensors="pt") 
         >>> outputs = model(**inputs)
         >>> predicted_answer_coordinates, predicted_aggregation_indices = tokenizer.convert_logits_to_predictions(
         ...         inputs, 
-        ...         outputs.logits.detach(), 
-        ...         outputs.logits_aggregation.detach()
+        ...         output.logits, 
+        ...         outputs.logits_aggregation
         ...)
 
         >>> # let's print out the results:
@@ -367,15 +318,15 @@ of that:
         >>> for coordinates in predicted_answer_coordinates:
         ...   if len(coordinates) == 1:
         ...     # only a single cell:
-        ...     answers.append(table.iat[coordinates[0]])
+        ...     answers.append(df.iat[coordinates[0]])
         ...   else:
         ...     # multiple cells
         ...     cell_values = []
         ...     for coordinate in coordinates:
-        ...        cell_values.append(table.iat[coordinate])
+        ...        cell_values.append(df.iat[coordinate])
         ...     answers.append(", ".join(cell_values))
 
-        >>> display(table)
+        >>> display(df)
         >>> print("")
         >>> for query, answer, predicted_agg in zip(queries, answers, aggregation_predictions_string):
         ...   print(query)
@@ -383,15 +334,15 @@ of that:
         ...     print("Predicted answer: " + answer)
         ...   else:
         ...     print("Predicted answer: " + predicted_agg + " > " + answer)    
-        What is the name of the first actor?
-        Predicted answer: Brad Pitt
-        How many movies has George Clooney played in?
-        Predicted answer: COUNT > 69
-        What is the total number of movies?
-        Predicted answer: SUM > 87, 53, 69
+        When was Brad Pitt born?
+        Predicted answer: 18 december 1963
+        Which actor appeared in the least number of movies?
+        Predicted answer: Leonardo Di Caprio
+        What is the average number of movies?
+        Predicted answer: AVERAGE > 87, 53, 69
 
 In case of a conversational set-up, then each table-question pair must be provided **sequentially** to the model, such
-that the ``prev_labels`` token types can be overwritten by the predicted ``labels`` of the previous table-question
+that the ``prev_labels`` token types can be overwritten by the predicted ``label_ids`` of the previous table-question
 pair.
 
 

docs/source/model_doc/transformerxl.rst

@@ -105,11 +105,3 @@ TFTransfoXLLMHeadModel
 
 .. autoclass:: transformers.TFTransfoXLLMHeadModel
     :members: call
-
-
-Internal Layers
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-
-.. autoclass:: transformers.AdaptiveEmbedding
-
-.. autoclass:: transformers.TFAdaptiveEmbedding

docs/source/model_doc/xlmroberta.rst

@@ -62,13 +62,6 @@ XLMRobertaTokenizer
         create_token_type_ids_from_sequences, save_vocabulary
 
 
-XLMRobertaTokenizerFast
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-
-.. autoclass:: transformers.XLMRobertaTokenizerFast
-    :members:
-
-
 XLMRobertaModel
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 

docs/source/model_doc/xlnet.rst

@@ -62,13 +62,6 @@ XLNetTokenizer
         create_token_type_ids_from_sequences, save_vocabulary
 
 
-XLNetTokenizerFast
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-
-.. autoclass:: transformers.XLNetTokenizerFast
-    :members:
-
-
 XLNet specific outputs
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 

examples/README.md

@@ -16,58 +16,59 @@ limitations under the License.
 
 # Examples
 
-This folder contains actively maintained examples of use of 🤗 Transformers organized along NLP tasks. If you are looking for an example that used to
-be in this folder, it may have moved to our [research projects](https://github.com/huggingface/transformers/tree/master/examples/research_projects) subfolder (which contains frozen snapshots of research projects).
+Version 2.9 of 🤗 Transformers introduced a new [`Trainer`](https://github.com/huggingface/transformers/blob/master/src/transformers/trainer.py) class for PyTorch, and its equivalent [`TFTrainer`](https://github.com/huggingface/transformers/blob/master/src/transformers/trainer_tf.py) for TF 2.
+Running the examples requires PyTorch 1.3.1+ or TensorFlow 2.2+.
+
+Here is the list of all our examples:
+- **grouped by task** (all official examples work for multiple models)
+- with information on whether they are **built on top of `Trainer`/`TFTrainer`** (if not, they still work, they might
+  just lack some features),
+- whether or not they leverage the [🤗 Datasets](https://github.com/huggingface/datasets) library.
+- links to **Colab notebooks** to walk through the scripts and run them easily,
+- links to **Cloud deployments** to be able to deploy large-scale trainings in the Cloud with little to no setup.
+
 
 ## Important note
 
 **Important**
 
-To make sure you can successfully run the latest versions of the example scripts, you have to **install the library from source** and install some example-specific requirements. To do this, execute the following steps in a new virtual environment:
+To make sure you can successfully run the latest versions of the example scripts, you have to **install the library from source** and install some example-specific requirements.
+Execute the following steps in a new virtual environment:
+
 ```bash
 git clone https://github.com/huggingface/transformers
 cd transformers
 pip install .
-```
-Then cd in the example folder of your choice and run
-```bash
-pip install -r requirements.txt
+pip install -r ./examples/requirements.txt
 ```
 
-Alternatively, you can run the version of the examples as they were for your current version of Transformers via (for instance with v3.5.1):
+Alternatively, you can run the version of the examples as they were for your current version of Transformers via (for instance with v3.4.0):
 ```bash
-git checkout tags/v3.5.1
+git checkout tags/v3.4.0
 ```
 
 ## The Big Table of Tasks
 
-Here is the list of all our examples:
-- with information on whether they are **built on top of `Trainer`/`TFTrainer`** (if not, they still work, they might
-  just lack some features),
-- whether or not they leverage the [🤗 Datasets](https://github.com/huggingface/datasets) library.
-- links to **Colab notebooks** to walk through the scripts and run them easily,
-<!--
-Coming soon!
-- links to **Cloud deployments** to be able to deploy large-scale trainings in the Cloud with little to no setup.
--->
-
 | Task | Example datasets | Trainer support | TFTrainer support | 🤗 Datasets | Colab
 |---|---|:---:|:---:|:---:|:---:|
 | [**`language-modeling`**](https://github.com/huggingface/transformers/tree/master/examples/language-modeling)       | Raw text        | ✅ | -  | ✅ | [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/huggingface/blog/blob/master/notebooks/01_how_to_train.ipynb)
+| [**`text-classification`**](https://github.com/huggingface/transformers/tree/master/examples/text-classification)   | GLUE, XNLI      | ✅ | ✅ | ✅ | [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://github.com/huggingface/notebooks/blob/master/examples/text_classification.ipynb)
+| [**`token-classification`**](https://github.com/huggingface/transformers/tree/master/examples/token-classification) | CoNLL NER       | ✅ | ✅ | ✅ | -
 | [**`multiple-choice`**](https://github.com/huggingface/transformers/tree/master/examples/multiple-choice)           | SWAG, RACE, ARC | ✅ | ✅ | - | [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/ViktorAlm/notebooks/blob/master/MPC_GPU_Demo_for_TF_and_PT.ipynb)
 | [**`question-answering`**](https://github.com/huggingface/transformers/tree/master/examples/question-answering)     | SQuAD           | ✅ | ✅ | ✅ | -
-| [**`summarization`**](https://github.com/huggingface/transformers/tree/master/examples/seq2seq)                     | CNN/Daily Mail  | ✅  | - | - | -
-| [**`text-classification`**](https://github.com/huggingface/transformers/tree/master/examples/text-classification)   | GLUE, XNLI      | ✅ | ✅ | ✅ | [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://github.com/huggingface/notebooks/blob/master/examples/text_classification.ipynb)
 | [**`text-generation`**](https://github.com/huggingface/transformers/tree/master/examples/text-generation)           | -               | n/a | n/a | - | [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/huggingface/blog/blob/master/notebooks/02_how_to_generate.ipynb)
-| [**`token-classification`**](https://github.com/huggingface/transformers/tree/master/examples/token-classification) | CoNLL NER       | ✅ | ✅ | ✅ | -
+| [**`distillation`**](https://github.com/huggingface/transformers/tree/master/examples/distillation)                 | All             | - | -  | - | -
+| [**`summarization`**](https://github.com/huggingface/transformers/tree/master/examples/seq2seq)                     | CNN/Daily Mail  | ✅  | - | - | -
 | [**`translation`**](https://github.com/huggingface/transformers/tree/master/examples/seq2seq)                       | WMT             | ✅  | - | - | -
+| [**`bertology`**](https://github.com/huggingface/transformers/tree/master/examples/bertology)                       | -               | - | - | - | -
+| [**`adversarial`**](https://github.com/huggingface/transformers/tree/master/examples/adversarial)                   | HANS            | ✅ | - | - | -
 
 
-<!--
+<br>
+
 ## One-click Deploy to Cloud (wip)
 
 **Coming soon!**
--->
 
 ## Running on TPUs
 

examples/_tests_requirements.txt

@@ -1,20 +0,0 @@
-tensorboard
-scikit-learn
-seqeval
-psutil
-sacrebleu
-rouge-score
-tensorflow_datasets
-matplotlib
-git-python==1.0.3
-faiss-cpu
-streamlit
-elasticsearch
-nltk
-pandas
-datasets >= 1.1.3
-fire
-pytest
-conllu
-sentencepiece != 0.1.92
-protobuf

examples/benchmarking/README.md

@@ -1,19 +1,3 @@
-<!---
-Copyright 2020 The HuggingFace Team. All rights reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
--->
-
 # 🤗 Benchmark results
 
 Here, you can find a list of the different benchmark results created by the community.

examples/benchmarking/plot_csv_file.py

@@ -1,17 +1,3 @@
-# Copyright 2020 The HuggingFace Team. All rights reserved.
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
 import csv
 from collections import defaultdict
 from dataclasses import dataclass, field

examples/conftest.py

@@ -1,17 +1,3 @@
-# Copyright 2020 The HuggingFace Team. All rights reserved.
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
 # tests directory-specific settings - this file is run automatically
 # by pytest before any tests are run
 

examples/contrib/README.md

@@ -0,0 +1,5 @@
+# Community contributed examples
+
+This folder contains examples which are not actively maintained (mostly contributed by the community).
+
+Using these examples together with a recent version of the library usually requires to make small (sometimes big) adaptations to get the scripts working.

examples/distillation/README.md

@@ -1,7 +1,5 @@
 # Distil*
 
-Author: @VictorSanh
-
 This folder contains the original code used to train Distil* as well as examples showcasing how to use DistilBERT, DistilRoBERTa and DistilGPT2.
 
 **January 20, 2020 - Bug fixing** We have recently discovered and fixed [a bug](https://github.com/huggingface/transformers/commit/48cbf267c988b56c71a2380f748a3e6092ccaed3) in the evaluation of our `run_*.py` scripts that caused the reported metrics to be over-estimated on average. We have updated all the metrics with the latest runs.

examples/language-modeling/README.md

@@ -1,19 +1,3 @@
-<!---
-Copyright 2020 The HuggingFace Team. All rights reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
--->
-
 ## Language model training
 
 Fine-tuning (or training from scratch) the library models for language modeling on a text dataset for GPT, GPT-2,

examples/language-modeling/requirements.txt

@@ -1,3 +0,0 @@
-datasets >= 1.1.3
-sentencepiece != 0.1.92
-protobuf

examples/language-modeling/run_mlm_wwm.py

@@ -281,7 +281,7 @@ def main():
     # Add the chinese references if provided
     if data_args.train_ref_file is not None:
         tokenized_datasets["train"] = add_chinese_references(tokenized_datasets["train"], data_args.train_ref_file)
-    if data_args.validation_ref_file is not None:
+    if data_args.valid_ref_file is not None:
         tokenized_datasets["validation"] = add_chinese_references(
             tokenized_datasets["validation"], data_args.validation_ref_file
         )

examples/legacy/README.md

@@ -1,21 +0,0 @@
-<!---
-Copyright 2020 The HuggingFace Team. All rights reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
--->
-
-# Legacy examples
-
-This folder contains examples which are not actively maintained (mostly contributed by the community).
-
-Using these examples together with a recent version of the library usually requires to make small (sometimes big) adaptations to get the scripts working.

examples/legacy/token-classification/README.md

@@ -1,229 +0,0 @@
-## Token classification
-
-Based on the scripts [`run_ner.py`](https://github.com/huggingface/transformers/blob/master/examples/contrib/legacy/token-classification/run_ner.py).
-
-The following examples are covered in this section:
-
-* NER on the GermEval 2014 (German NER) dataset
-* Emerging and Rare Entities task: WNUT’17 (English NER) dataset
-
-Details and results for the fine-tuning provided by @stefan-it.
-
-### GermEval 2014 (German NER) dataset
-
-#### Data (Download and pre-processing steps)
-
-Data can be obtained from the [GermEval 2014](https://sites.google.com/site/germeval2014ner/data) shared task page.
-
-Here are the commands for downloading and pre-processing train, dev and test datasets. The original data format has four (tab-separated) columns, in a pre-processing step only the two relevant columns (token and outer span NER annotation) are extracted:
-
-```bash
-curl -L 'https://drive.google.com/uc?export=download&id=1Jjhbal535VVz2ap4v4r_rN1UEHTdLK5P' \
-| grep -v "^#" | cut -f 2,3 | tr '\t' ' ' > train.txt.tmp
-curl -L 'https://drive.google.com/uc?export=download&id=1ZfRcQThdtAR5PPRjIDtrVP7BtXSCUBbm' \
-| grep -v "^#" | cut -f 2,3 | tr '\t' ' ' > dev.txt.tmp
-curl -L 'https://drive.google.com/uc?export=download&id=1u9mb7kNJHWQCWyweMDRMuTFoOHOfeBTH' \
-| grep -v "^#" | cut -f 2,3 | tr '\t' ' ' > test.txt.tmp
-```
-
-The GermEval 2014 dataset contains some strange "control character" tokens like `'\x96', '\u200e', '\x95', '\xad' or '\x80'`.
-One problem with these tokens is, that `BertTokenizer` returns an empty token for them, resulting in misaligned `InputExample`s.
-The `preprocess.py` script located in the `scripts` folder a) filters these tokens and b) splits longer sentences into smaller ones (once the max. subtoken length is reached).
-
-Let's define some variables that we need for further pre-processing steps and training the model:
-
-```bash
-export MAX_LENGTH=128
-export BERT_MODEL=bert-base-multilingual-cased
-```
-
-Run the pre-processing script on training, dev and test datasets:
-
-```bash
-python3 scripts/preprocess.py train.txt.tmp $BERT_MODEL $MAX_LENGTH > train.txt
-python3 scripts/preprocess.py dev.txt.tmp $BERT_MODEL $MAX_LENGTH > dev.txt
-python3 scripts/preprocess.py test.txt.tmp $BERT_MODEL $MAX_LENGTH > test.txt
-```
-
-The GermEval 2014 dataset has much more labels than CoNLL-2002/2003 datasets, so an own set of labels must be used:
-
-```bash
-cat train.txt dev.txt test.txt | cut -d " " -f 2 | grep -v "^$"| sort | uniq > labels.txt
-```
-
-#### Prepare the run
-
-Additional environment variables must be set:
-
-```bash
-export OUTPUT_DIR=germeval-model
-export BATCH_SIZE=32
-export NUM_EPOCHS=3
-export SAVE_STEPS=750
-export SEED=1
-```
-
-#### Run the Pytorch version
-
-To start training, just run:
-
-```bash
-python3 run_ner.py --data_dir ./ \
---labels ./labels.txt \
---model_name_or_path $BERT_MODEL \
---output_dir $OUTPUT_DIR \
---max_seq_length  $MAX_LENGTH \
---num_train_epochs $NUM_EPOCHS \
---per_device_train_batch_size $BATCH_SIZE \
---save_steps $SAVE_STEPS \
---seed $SEED \
---do_train \
---do_eval \
---do_predict
-```
-
-If your GPU supports half-precision training, just add the `--fp16` flag. After training, the model will be both evaluated on development and test datasets.
-
-#### JSON-based configuration file
-
-Instead of passing all parameters via commandline arguments, the `run_ner.py` script also supports reading parameters from a json-based configuration file:
-
-```json
-{
-    "data_dir": ".",
-    "labels": "./labels.txt",
-    "model_name_or_path": "bert-base-multilingual-cased",
-    "output_dir": "germeval-model",
-    "max_seq_length": 128,
-    "num_train_epochs": 3,
-    "per_device_train_batch_size": 32,
-    "save_steps": 750,
-    "seed": 1,
-    "do_train": true,
-    "do_eval": true,
-    "do_predict": true
-}
-```
-
-It must be saved with a `.json` extension and can be used by running `python3 run_ner.py config.json`.
-
-#### Evaluation
-
-Evaluation on development dataset outputs the following for our example:
-
-```bash
-10/04/2019 00:42:06 - INFO - __main__ -   ***** Eval results  *****
-10/04/2019 00:42:06 - INFO - __main__ -     f1 = 0.8623348017621146
-10/04/2019 00:42:06 - INFO - __main__ -     loss = 0.07183869666975543
-10/04/2019 00:42:06 - INFO - __main__ -     precision = 0.8467916366258111
-10/04/2019 00:42:06 - INFO - __main__ -     recall = 0.8784592370979806
-```
-
-On the test dataset the following results could be achieved:
-
-```bash
-10/04/2019 00:42:42 - INFO - __main__ -   ***** Eval results  *****
-10/04/2019 00:42:42 - INFO - __main__ -     f1 = 0.8614389652384803
-10/04/2019 00:42:42 - INFO - __main__ -     loss = 0.07064602487454782
-10/04/2019 00:42:42 - INFO - __main__ -     precision = 0.8604651162790697
-10/04/2019 00:42:42 - INFO - __main__ -     recall = 0.8624150210424085
-```
-
-### Emerging and Rare Entities task: WNUT’17 (English NER) dataset
-
-Description of the WNUT’17 task from the [shared task website](http://noisy-text.github.io/2017/index.html):
-
-> The WNUT’17 shared task focuses on identifying unusual, previously-unseen entities in the context of emerging discussions.
-> Named entities form the basis of many modern approaches to other tasks (like event clustering and summarization), but recall on
-> them is a real problem in noisy text - even among annotators. This drop tends to be due to novel entities and surface forms.
-
-Six labels are available in the dataset. An overview can be found on this [page](http://noisy-text.github.io/2017/files/).
-
-#### Data (Download and pre-processing steps)
-
-The dataset can be downloaded from the [official GitHub](https://github.com/leondz/emerging_entities_17) repository.
-
-The following commands show how to prepare the dataset for fine-tuning:
-
-```bash
-mkdir -p data_wnut_17
-
-curl -L 'https://github.com/leondz/emerging_entities_17/raw/master/wnut17train.conll'  | tr '\t' ' ' > data_wnut_17/train.txt.tmp
-curl -L 'https://github.com/leondz/emerging_entities_17/raw/master/emerging.dev.conll' | tr '\t' ' ' > data_wnut_17/dev.txt.tmp
-curl -L 'https://raw.githubusercontent.com/leondz/emerging_entities_17/master/emerging.test.annotated' | tr '\t' ' ' > data_wnut_17/test.txt.tmp
-```
-
-Let's define some variables that we need for further pre-processing steps:
-
-```bash
-export MAX_LENGTH=128
-export BERT_MODEL=bert-large-cased
-```
-
-Here we use the English BERT large model for fine-tuning.
-The `preprocess.py` scripts splits longer sentences into smaller ones (once the max. subtoken length is reached):
-
-```bash
-python3 scripts/preprocess.py data_wnut_17/train.txt.tmp $BERT_MODEL $MAX_LENGTH > data_wnut_17/train.txt
-python3 scripts/preprocess.py data_wnut_17/dev.txt.tmp $BERT_MODEL $MAX_LENGTH > data_wnut_17/dev.txt
-python3 scripts/preprocess.py data_wnut_17/test.txt.tmp $BERT_MODEL $MAX_LENGTH > data_wnut_17/test.txt
-```
-
-In the last pre-processing step, the `labels.txt` file needs to be generated. This file contains all available labels:
-
-```bash
-cat data_wnut_17/train.txt data_wnut_17/dev.txt data_wnut_17/test.txt | cut -d " " -f 2 | grep -v "^$"| sort | uniq > data_wnut_17/labels.txt
-```
-
-#### Run the Pytorch version
-
-Fine-tuning with the PyTorch version can be started using the `run_ner.py` script. In this example we use a JSON-based configuration file.
-
-This configuration file looks like:
-
-```json
-{
-    "data_dir": "./data_wnut_17",
-    "labels": "./data_wnut_17/labels.txt",
-    "model_name_or_path": "bert-large-cased",
-    "output_dir": "wnut-17-model-1",
-    "max_seq_length": 128,
-    "num_train_epochs": 3,
-    "per_device_train_batch_size": 32,
-    "save_steps": 425,
-    "seed": 1,
-    "do_train": true,
-    "do_eval": true,
-    "do_predict": true,
-    "fp16": false
-}
-```
-
-If your GPU supports half-precision training, please set `fp16` to `true`.
-
-Save this JSON-based configuration under `wnut_17.json`. The fine-tuning can be started with `python3 run_ner_old.py wnut_17.json`.
-
-#### Evaluation
-
-Evaluation on development dataset outputs the following:
-
-```bash
-05/29/2020 23:33:44 - INFO - __main__ -   ***** Eval results *****
-05/29/2020 23:33:44 - INFO - __main__ -     eval_loss = 0.26505235286212275
-05/29/2020 23:33:44 - INFO - __main__ -     eval_precision = 0.7008264462809918
-05/29/2020 23:33:44 - INFO - __main__ -     eval_recall = 0.507177033492823
-05/29/2020 23:33:44 - INFO - __main__ -     eval_f1 = 0.5884802220680084
-05/29/2020 23:33:44 - INFO - __main__ -     epoch = 3.0
-```
-
-On the test dataset the following results could be achieved:
-
-```bash
-05/29/2020 23:33:44 - INFO - transformers.trainer -   ***** Running Prediction *****
-05/29/2020 23:34:02 - INFO - __main__ -     eval_loss = 0.30948806500973547
-05/29/2020 23:34:02 - INFO - __main__ -     eval_precision = 0.5840108401084011
-05/29/2020 23:34:02 - INFO - __main__ -     eval_recall = 0.3994439295644115
-05/29/2020 23:34:02 - INFO - __main__ -     eval_f1 = 0.47440836543753434
-```
-
-WNUT’17 is a very difficult task. Current state-of-the-art results on this dataset can be found [here](http://nlpprogress.com/english/named_entity_recognition.html).

examples/longform-qa/README.md

@@ -1,7 +1,5 @@
 # Long Form Question Answering
 
-Author: @yjernite
-
 This folder contains the code for the Long Form Question answering [demo](http://35.226.96.115:8080/) as well as methods to train and use a fully end-to-end Long Form Question Answering system using the [🤗transformers](https://github.com/huggingface/transformers) and [🤗datasets](https://github.com/huggingface/datasets) libraries.
 
 You can use these methods to train your own system by following along the associate [notebook](https://github.com/huggingface/notebooks/blob/master/longform-qa/Long_Form_Question_Answering_with_ELI5_and_Wikipedia.ipynb) or [blog post](https://yjernite.github.io/lfqa.html).

examples/lxmert/requirements.txt

@@ -47,7 +47,7 @@ nbclient==0.5.0
 nbconvert==6.0.1
 nbformat==5.0.7
 nest-asyncio==1.4.0
-notebook==6.1.5
+notebook==6.1.4
 numpy==1.19.2
 opencv-python==4.4.0.42
 packaging==20.3
@@ -90,7 +90,7 @@ torchvision==0.7.0
 tornado==6.0.4
 tqdm==4.48.2
 traitlets
-transformers==3.5.1
+git+https://github.com/huggingface/transformers.git
 urllib3==1.25.8
 wcwidth==0.2.5
 webencodings==0.5.1

examples/movement-pruning/README.md

@@ -1,7 +1,5 @@
 # Movement Pruning: Adaptive Sparsity by Fine-Tuning
 
-Author: @VictorSanh
-
 *Magnitude pruning is a widely used strategy for reducing model size in pure supervised learning; however, it is less effective in the transfer learning regime that has become standard for state-of-the-art natural language processing applications. We propose the use of *movement pruning*, a simple, deterministic first-order weight pruning method that is more adaptive to pretrained model fine-tuning. Experiments show that when pruning large pretrained language models, movement pruning shows significant improvements in high-sparsity regimes. When combined with distillation, the approach achieves minimal accuracy loss with down to only 3% of the model parameters:*
 
 | Fine-pruning+Distillation<br>(Teacher=BERT-base fine-tuned) | BERT base<br>fine-tuned | Remaining<br>Weights (%) | Magnitude Pruning      | L0 Regularization      | Movement Pruning       | Soft Movement Pruning          |