SUMIN/HuggingFace_transformer
1
0
Fork 0
Code Issues Pull Requests Actions 7 Packages Projects Releases Wiki Activity

Models doc (#7345)

Browse Source 
* Clean up model documentation

* Formatting

* Preparation work

* Long lines

* Main work on rst files

* Cleanup all config files

* Syntax fix

* Clean all tokenizers

* Work on first models

* Models beginning

* FaluBERT

* All PyTorch models

* All models

* Long lines again

* Fixes

* More fixes

* Update docs/source/model_doc/bert.rst

Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

* Update docs/source/model_doc/electra.rst

Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

* Last fixes

Co-authored-by: Lysandre Debut <lysandre@huggingface.co>
This commit is contained in:
Sylvain Gugger
2020-09-23 13:20:45 -04:00
committed by GitHub
parent 58405a527b
commit 3323146e90
165 changed files with 6907 additions and 5803 deletions
Download Patch File Download Diff File Expand all files Collapse all files

52
docs/source/model_doc/bertgeneration.rst
View File

@@ -1,24 +1,36 @@
BertGeneration
----------------------------------------------------
-----------------------------------------------------------------------------------------------------------------------
Overview
~~~~~~~~~~~~~~~~~~~~~
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
The BertGeneration model is a BERT model that can be leveraged for sequence-to-sequence tasks using :class:`~transformers.EncoderDecoderModel` as proposed in `Leveraging Pre-trained Checkpoints for Sequence Generation Tasks <https://arxiv.org/abs/1907.12461>`__ by Sascha Rothe, Shashi Narayan, Aliaksei Severyn.
The BertGeneration model is a BERT model that can be leveraged for sequence-to-sequence tasks using
:class:`~transformers.EncoderDecoderModel` as proposed in `Leveraging Pre-trained Checkpoints for Sequence Generation
Tasks <https://arxiv.org/abs/1907.12461>`__ by Sascha Rothe, Shashi Narayan, Aliaksei Severyn.
The abstract from the paper is the following:
*Unsupervised pre-training of large neural models has recently revolutionized Natural Language Processing. By warm-starting from the publicly released checkpoints, NLP practitioners have pushed the state-of-the-art on multiple benchmarks while saving significant amounts of compute time. So far the focus has been mainly on the Natural Language Understanding tasks. In this paper, we demonstrate the efficacy of pre-trained checkpoints for Sequence Generation. We developed a Transformer-based sequence-to-sequence model that is compatible with publicly available pre-trained BERT, GPT-2 and RoBERTa checkpoints and conducted an extensive empirical study on the utility of initializing our model, both encoder and decoder, with these checkpoints. Our models result in new state-of-the-art results on Machine Translation, Text Summarization, Sentence Splitting, and Sentence Fusion.*
*Unsupervised pre-training of large neural models has recently revolutionized Natural Language Processing. By
warm-starting from the publicly released checkpoints, NLP practitioners have pushed the state-of-the-art on multiple
benchmarks while saving significant amounts of compute time. So far the focus has been mainly on the Natural Language
Understanding tasks. In this paper, we demonstrate the efficacy of pre-trained checkpoints for Sequence Generation. We
developed a Transformer-based sequence-to-sequence model that is compatible with publicly available pre-trained BERT,
GPT-2 and RoBERTa checkpoints and conducted an extensive empirical study on the utility of initializing our model, both
encoder and decoder, with these checkpoints. Our models result in new state-of-the-art results on Machine Translation,
Text Summarization, Sentence Splitting, and Sentence Fusion.*
Usage:
- The model can be used in combination with the :class:`~transformers.EncoderDecoderModel` to leverage two bert pretrained bert checkpoints for subsequent fine-tuning.
- The model can be used in combination with the :class:`~transformers.EncoderDecoderModel` to leverage two pretrained
BERT checkpoints for subsequent fine-tuning.
::
:: code-block
# leverage checkpoints for Bert2Bert model...
encoder = BertGenerationEncoder.from_pretrained("bert-large-uncased", bos_token_id=101, eos_token_id=102) # use BERT's cls token as BOS token and sep token as EOS token
decoder = BertGenerationDecoder.from_pretrained("bert-large-uncased", add_cross_attention=True, is_decoder=True, bos_token_id=101, eos_token_id=102) # add cross attention layers and use BERT's cls token as BOS token and sep token as EOS token
# use BERT's cls token as BOS token and sep token as EOS token
encoder = BertGenerationEncoder.from_pretrained("bert-large-uncased", bos_token_id=101, eos_token_id=102)
# add cross attention layers and use BERT's cls token as BOS token and sep token as EOS token
decoder = BertGenerationDecoder.from_pretrained("bert-large-uncased", add_cross_attention=True, is_decoder=True, bos_token_id=101, eos_token_id=102)
bert2bert = EncoderDecoderModel(encoder=encoder, decoder=decoder)
# create tokenizer...
@@ -32,10 +44,10 @@ Usage:
loss.backward()
- Pretrained :class:`~transformers.EncoderDecoderModel` are also directly available in the model hub, *e.g.*:
- Pretrained :class:`~transformers.EncoderDecoderModel` are also directly available in the model hub, e.g.,
::
:: code-block
# instantiate sentence fusion model
sentence_fuser = EncoderDecoderModel.from_pretrained("google/roberta2roberta_L-24_discofuse")
@@ -50,33 +62,35 @@ Usage:
Tips:
- :class:`~transformers.BertGenerationEncoder` and :class:`~transformers.BertGenerationDecoder` should be used in combination with :class:`~transformers.EncoderDecoder`.
- For summarization, sentence splitting, sentence fusion and translation, no special tokens are required for the input. Therefore, no EOS token should be added to the end of the input.
- :class:`~transformers.BertGenerationEncoder` and :class:`~transformers.BertGenerationDecoder` should be used in
combination with :class:`~transformers.EncoderDecoder`.
- For summarization, sentence splitting, sentence fusion and translation, no special tokens are required for the input.
Therefore, no EOS token should be added to the end of the input.
The original code can be found `here <https://tfhub.dev/s?module-type=text-generation&subtype=module,placeholder>`__.
BertGenerationConfig
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
.. autoclass:: transformers.BertGenerationConfig
:members:
BertGenerationTokenizer
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
.. autoclass:: transformers.BertGenerationTokenizer
:members:
:members: save_vocabulary
BertGenerationEncoder
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
.. autoclass:: transformers.BertGenerationEncoder
:members:
:members: forward
BertGenerationDecoder
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
.. autoclass:: transformers.BertGenerationDecoder
:members:
:members: forward