riaz.somc/transformers
1
0
Fork 0
mirror of https://github.com/huggingface/transformers.git synced 2025-07-31 02:02:21 +06:00
Code Issues Actions 33 Packages Projects Releases Wiki Activity

[docs] fix xref to PreTrainedModel.generate (#11049)

Browse Source 
* fix xref to generate

* do the same for search methods

* style

* style
This commit is contained in:
Stas Bekman 2021-06-02 09:21:05 -07:00 committed by GitHub
parent 123b597f5d
commit d406a2729a
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
6 changed files with 37 additions and 29 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

14
docs/source/internal/generation_utils.rst
View File

@ -13,19 +13,21 @@
Utilities for Generation
-----------------------------------------------------------------------------------------------------------------------
This page lists all the utility functions used by :meth:`~transformers.PreTrainedModel.generate`,
:meth:`~transformers.PreTrainedModel.greedy_search`, :meth:`~transformers.PreTrainedModel.sample`,
:meth:`~transformers.PreTrainedModel.beam_search`, :meth:`~transformers.PreTrainedModel.beam_sample`, and
:meth:`~transformers.PreTrainedModel.group_beam_search`.
This page lists all the utility functions used by :meth:`~transformers.generation_utils.GenerationMixin.generate`,
:meth:`~transformers.generation_utils.GenerationMixin.greedy_search`,
:meth:`~transformers.generation_utils.GenerationMixin.sample`,
:meth:`~transformers.generation_utils.GenerationMixin.beam_search`,
:meth:`~transformers.generation_utils.GenerationMixin.beam_sample`, and
:meth:`~transformers.generation_utils.GenerationMixin.group_beam_search`.
Most of those are only useful if you are studying the code of the generate methods in the library.
Generate Outputs
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
The output of :meth:`~transformers.PreTrainedModel.generate` is an instance of a subclass of
The output of :meth:`~transformers.generation_utils.GenerationMixin.generate` is an instance of a subclass of
:class:`~transformers.file_utils.ModelOutput`. This output is a data structure containing all the information returned
by :meth:`~transformers.PreTrainedModel.generate`, but that can also be used as tuple or dictionary.
by :meth:`~transformers.generation_utils.GenerationMixin.generate`, but that can also be used as tuple or dictionary.
Here's an example:

2
docs/source/model_doc/bart.rst
View File

@ -61,7 +61,7 @@ Implementation Notes
- Model predictions are intended to be identical to the original implementation when
:obj:`force_bos_token_to_be_generated=True`. This only works, however, if the string you pass to
:func:`fairseq.encode` starts with a space.
- :meth:`~transformers.BartForConditionalGeneration.generate` should be used for conditional generation tasks like
- :meth:`~transformers.generation_utils.GenerationMixin.generate` should be used for conditional generation tasks like
summarization, see the example in that docstrings.
- Models that load the `facebook/bart-large-cnn` weights will not have a :obj:`mask_token_id`, or be able to perform
mask-filling tasks.

8
docs/source/model_doc/t5.rst
View File

@ -1,4 +1,4 @@
..
..
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
@ -44,9 +44,9 @@ Tips:
For more information about which prefix to use, it is easiest to look into Appendix D of the `paper
<https://arxiv.org/pdf/1910.10683.pdf>`__. - For sequence-to-sequence generation, it is recommended to use
:obj:`T5ForConditionalGeneration.generate()`. This method takes care of feeding the encoded input via cross-attention
layers to the decoder and auto-regressively generates the decoder output. - T5 uses relative scalar embeddings.
Encoder input padding can be done on the left and on the right.
:meth:`~transformers.generation_utils.GenerationMixin.generate`. This method takes care of feeding the encoded input
via cross-attention layers to the decoder and auto-regressively generates the decoder output. - T5 uses relative
scalar embeddings. Encoder input padding can be done on the left and on the right.
This model was contributed by `thomwolf <https://huggingface.co/thomwolf>`__. The original code can be found `here
<https://github.com/google-research/text-to-text-transfer-transformer>`__.

8
docs/source/task_summary.rst
View File

@ -1,4 +1,4 @@
..
..
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
@ -505,8 +505,8 @@ This outputs a (hopefully) coherent next token following the original sequence,
>>> print(resulting_string)
Hugging Face is based in DUMBO, New York City, and has
In the next section, we show how :func:`~transformers.PreTrainedModel.generate` can be used to generate multiple tokens
up to a specified length instead of one token at a time.
In the next section, we show how :func:`~transformers.generation_utils.GenerationMixin.generate` can be used to
generate multiple tokens up to a specified length instead of one token at a time.
Text Generation
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
@ -629,7 +629,7 @@ It leverages a fine-tuned model on CoNLL-2003, fine-tuned by `@stefan-it <https:
>>> ner_pipe = pipeline("ner")
>>> sequence = """Hugging Face Inc. is a company based in New York City. Its headquarters are in DUMBO,
>>> sequence = """Hugging Face Inc. is a company based in New York City. Its headquarters are in DUMBO,
... therefore very close to the Manhattan Bridge which is visible from the window."""

17
src/transformers/models/rag/modeling_rag.py
View File

@ -906,8 +906,9 @@ class RagSequenceForGeneration(RagPreTrainedModel):
**model_kwargs
):
"""
Implements RAG sequence "thorough" decoding. Read the :meth:`~transformers.PreTrainedModel.generate``
documentation for more information on how to set other generate input parameters.
Implements RAG sequence "thorough" decoding. Read the
:meth:`~transformers.generation_utils.GenerationMixin.generate`` documentation for more information on how to
set other generate input parameters.
Args:
input_ids (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`):
@ -942,14 +943,15 @@ class RagSequenceForGeneration(RagPreTrainedModel):
to be set to :obj:`False` if used while training with distributed backend.
num_return_sequences(:obj:`int`, `optional`, defaults to 1):
The number of independently computed returned sequences for each element in the batch. Note that this
is not the value we pass to the ``generator``'s `:func:`~transformers.PreTrainedModel.generate``
function, where we set ``num_return_sequences`` to :obj:`num_beams`.
is not the value we pass to the ``generator``'s
`:func:`~transformers.generation_utils.GenerationMixin.generate`` function, where we set
``num_return_sequences`` to :obj:`num_beams`.
num_beams (:obj:`int`, `optional`, defaults to 1):
Number of beams for beam search. 1 means no beam search.
n_docs (:obj:`int`, `optional`, defaults to :obj:`config.n_docs`)
Number of documents to retrieve and/or number of documents for which to generate an answer.
kwargs:
Additional kwargs will be passed to :meth:`~transformers.PreTrainedModel.generate`.
Additional kwargs will be passed to :meth:`~transformers.generation_utils.GenerationMixin.generate`.
Return:
:obj:`torch.LongTensor` of shape :obj:`(batch_size * num_return_sequences, sequence_length)`: The generated
@ -1452,8 +1454,9 @@ class RagTokenForGeneration(RagPreTrainedModel):
enabled.
num_return_sequences(:obj:`int`, `optional`, defaults to 1):
The number of independently computed returned sequences for each element in the batch. Note that this
is not the value we pass to the ``generator``'s `:func:`~transformers.PreTrainedModel.generate`
function, where we set ``num_return_sequences`` to :obj:`num_beams`.
is not the value we pass to the ``generator``'s
`:func:`~transformers.generation_utils.GenerationMixin.generate` function, where we set
``num_return_sequences`` to :obj:`num_beams`.
decoder_start_token_id (:obj:`int`, `optional`):
If an encoder-decoder model starts decoding with a different token than `bos`, the id of that token.
n_docs (:obj:`int`, `optional`, defaults to :obj:`config.n_docs`)

17
src/transformers/models/rag/modeling_tf_rag.py
View File

@ -1130,8 +1130,9 @@ class TFRagTokenForGeneration(TFRagPreTrainedModel, TFCausalLanguageModelingLoss
Number of beams for beam search. 1 means no beam search.
num_return_sequences(:obj:`int`, `optional`, defaults to 1):
The number of independently computed returned sequences for each element in the batch. Note that this
is not the value we pass to the ``generator``'s `:func:`~transformers.PreTrainedModel.generate`
function, where we set ``num_return_sequences`` to :obj:`num_beams`.
is not the value we pass to the ``generator``'s
`:func:`~transformers.generation_utils.GenerationMixin.generate` function, where we set
``num_return_sequences`` to :obj:`num_beams`.
decoder_start_token_id (:obj:`int`, `optional`):
If an encoder-decoder model starts decoding with a different token than `bos`, the id of that token.
n_docs (:obj:`int`, `optional`, defaults to :obj:`config.n_docs`)
@ -1682,8 +1683,9 @@ class TFRagSequenceForGeneration(TFRagPreTrainedModel, TFCausalLanguageModelingL
**model_kwargs
):
"""
Implements RAG sequence "thorough" decoding. Read the :meth:`~transformers.PreTrainedModel.generate``
documentation for more information on how to set other generate input parameters
Implements RAG sequence "thorough" decoding. Read the
:meth:`~transformers.generation_utils.GenerationMixin.generate`` documentation for more information on how to
set other generate input parameters
Args:
input_ids (:obj:`tf.Tensor` of shape :obj:`(batch_size, sequence_length)`, `optional`):
@ -1711,14 +1713,15 @@ class TFRagSequenceForGeneration(TFRagPreTrainedModel, TFCausalLanguageModelingL
to be set to :obj:`False` if used while training with distributed backend.
num_return_sequences(:obj:`int`, `optional`, defaults to 1):
The number of independently computed returned sequences for each element in the batch. Note that this
is not the value we pass to the ``generator``'s `:func:`~transformers.PreTrainedModel.generate``
function, where we set ``num_return_sequences`` to :obj:`num_beams`.
is not the value we pass to the ``generator``'s
`:func:`~transformers.generation_utils.GenerationMixin.generate`` function, where we set
``num_return_sequences`` to :obj:`num_beams`.
num_beams (:obj:`int`, `optional`, defaults to 1):
Number of beams for beam search. 1 means no beam search.
n_docs (:obj:`int`, `optional`, defaults to :obj:`config.n_docs`)
Number of documents to retrieve and/or number of documents for which to generate an answer.
kwargs:
Additional kwargs will be passed to :meth:`~transformers.PreTrainedModel.generate`
Additional kwargs will be passed to :meth:`~transformers.generation_utils.GenerationMixin.generate`
Return:
:obj:`tf.Tensor` of shape :obj:`(batch_size * num_return_sequences, sequence_length)`: The generated