Skip to content

GitLab

Unverified Commit 27b3031d authored by Sylvain Gugger's avatar Sylvain Gugger Committed by GitHub
Browse files

Mass conversion of documentation from rst to Markdown (#14866) 

* Convert docstrings of all configurations and tokenizers

* Processors and fixes

* Last modeling files and fixes to models

* Pipeline modules

* Utils files

* Data submodule

* All the other files

* Style

* Missing examples

* Style again

* Fix copies

* Say bye bye to rst docstrings forever
parent 18587639
Show whitespace changes
Inline Side-by-side
Showing with 890 additions and 859 deletions
src/transformers/models/vision_text_dual_encoder/modeling_vision_text_dual_encoder.py
View file @ 27b3031d
......@@ -310,8 +310,9 @@ class VisionTextDualEncoderModel(PreTrainedModel):
r"""
Returns:
Examples::
Examples:
```python
>>> from PIL import Image
>>> import requests
>>> from transformers import VisionTextDualEncoderModel, VisionTextDualEncoderProcessor, ViTFeatureExtractor, BertTokenizer
......@@ -336,8 +337,7 @@ class VisionTextDualEncoderModel(PreTrainedModel):
>>> outputs = model(**inputs)
>>> logits_per_image = outputs.logits_per_image # this is the image-text similarity score
>>> probs = logits_per_image.softmax(dim=1) # we can take the softmax to get the label probabilities
"""
```"""
return_dict = return_dict if return_dict is not None else self.config.return_dict
vision_outputs = self.vision_model(
......
src/transformers/models/vision_text_dual_encoder/processing_vision_text_dual_encoder.py
View file @ 27b3031d
......@@ -30,15 +30,15 @@ class VisionTextDualEncoderProcessor:
Constructs a VisionTextDualEncoder processor which wraps a vision feature extractor and a tokenizer into a single
processor.
:class:`~transformers.VisionTextDualEncoderProcessor` offers all the functionalities of
:class:`~transformers.AutoFeatureExtractor` and :class:`~transformers.AutoTokenizer`. See the
:meth:`~transformers.VisionTextDualEncoderProcessor.__call__` and
:meth:`~transformers.VisionTextDualEncoderProcessor.decode` for more information.
[`VisionTextDualEncoderProcessor`] offers all the functionalities of
[`AutoFeatureExtractor`] and [`AutoTokenizer`]. See the
[`~VisionTextDualEncoderProcessor.__call__`] and
[`~VisionTextDualEncoderProcessor.decode`] for more information.
Args:
feature_extractor (:class:`~transformers.AutoFeatureExtractor`):
feature_extractor ([`AutoFeatureExtractor`]):
The feature extractor is a required input.
tokenizer (:class:`~transformers.PreTrainedTokenizer`):
tokenizer ([`PreTrainedTokenizer`]):
The tokenizer is a required input.
"""
......@@ -61,17 +61,19 @@ class VisionTextDualEncoderProcessor:
def save_pretrained(self, save_directory):
"""
Save a VisionTextDualEncoder feature extractor object and VisionTextDualEncoder tokenizer object to the
directory ``save_directory``, so that it can be re-loaded using the
:func:`~transformers.VisionTextDualEncoderProcessor.from_pretrained` class method.
directory `save_directory`, so that it can be re-loaded using the
[`~VisionTextDualEncoderProcessor.from_pretrained`] class method.
.. note::
<Tip>
This class method is simply calling :meth:`~transformers.PreTrainedFeatureExtractor.save_pretrained` and
:meth:`~transformers.tokenization_utils_base.PreTrainedTokenizer.save_pretrained`. Please refer to the
This class method is simply calling [`~PreTrainedFeatureExtractor.save_pretrained`] and
[`~tokenization_utils_base.PreTrainedTokenizer.save_pretrained`]. Please refer to the
docstrings of the methods above for more information.
</Tip>
Args:
save_directory (:obj:`str` or :obj:`os.PathLike`):
save_directory (`str` or `os.PathLike`):
Directory where the feature extractor JSON file and the tokenizer files will be saved (directory will
be created if it does not exist).
"""
......@@ -82,32 +84,34 @@ class VisionTextDualEncoderProcessor:
@classmethod
def from_pretrained(cls, pretrained_model_name_or_path, **kwargs):
r"""
Instantiate a :class:`~transformers.VisionTextDualEncoderProcessor` from a pretrained VisionTextDualEncoder
Instantiate a [`VisionTextDualEncoderProcessor`] from a pretrained VisionTextDualEncoder
processor.
.. note::
<Tip>
This class method is simply calling AutoFeatureExtractor's
:meth:`~transformers.PreTrainedFeatureExtractor.from_pretrained` and AutoTokenizer's
:meth:`~transformers.tokenization_utils_base.PreTrainedTokenizer.from_pretrained`. Please refer to the
[`~PreTrainedFeatureExtractor.from_pretrained`] and AutoTokenizer's
[`~tokenization_utils_base.PreTrainedTokenizer.from_pretrained`]. Please refer to the
docstrings of the methods above for more information.
</Tip>
Args:
pretrained_model_name_or_path (:obj:`str` or :obj:`os.PathLike`):
pretrained_model_name_or_path (`str` or `os.PathLike`):
This can be either:
- a string, the `model id` of a pretrained feature_extractor hosted inside a model repo on
huggingface.co. Valid model ids can be located at the root-level, like ``bert-base-uncased``, or
namespaced under a user or organization name, like ``dbmdz/bert-base-german-cased``.
- a path to a `directory` containing a feature extractor file saved using the
:meth:`~transformers.PreTrainedFeatureExtractor.save_pretrained` method, e.g.,
``./my_model_directory/``.
- a path or url to a saved feature extractor JSON `file`, e.g.,
``./my_model_directory/preprocessor_config.json``.
- a string, the *model id* of a pretrained feature_extractor hosted inside a model repo on
huggingface.co. Valid model ids can be located at the root-level, like `bert-base-uncased`, or
namespaced under a user or organization name, like `dbmdz/bert-base-german-cased`.
- a path to a *directory* containing a feature extractor file saved using the
[`~PreTrainedFeatureExtractor.save_pretrained`] method, e.g.,
`./my_model_directory/`.
- a path or url to a saved feature extractor JSON *file*, e.g.,
`./my_model_directory/preprocessor_config.json`.
**kwargs
Additional keyword arguments passed along to both :class:`~transformers.PreTrainedFeatureExtractor` and
:class:`~transformers.PreTrainedTokenizer`
Additional keyword arguments passed along to both [`PreTrainedFeatureExtractor`] and
[`PreTrainedTokenizer`]
"""
feature_extractor = AutoFeatureExtractor.from_pretrained(pretrained_model_name_or_path, **kwargs)
tokenizer = AutoTokenizer.from_pretrained(pretrained_model_name_or_path, **kwargs)
......@@ -117,38 +121,38 @@ class VisionTextDualEncoderProcessor:
def __call__(self, text=None, images=None, return_tensors=None, **kwargs):
"""
Main method to prepare for the model one or several sequences(s) and image(s). This method forwards the
:obj:`text` and :obj:`kwargs` arguments to VisionTextDualEncoderTokenizer's
:meth:`~transformers.PreTrainedTokenizer.__call__` if :obj:`text` is not :obj:`None` to encode the text. To
prepare the image(s), this method forwards the :obj:`images` and :obj:`kwrags` arguments to
AutoFeatureExtractor's :meth:`~transformers.AutoFeatureExtractor.__call__` if :obj:`images` is not :obj:`None`.
`text` and `kwargs` arguments to VisionTextDualEncoderTokenizer's
[`~PreTrainedTokenizer.__call__`] if `text` is not `None` to encode the text. To
prepare the image(s), this method forwards the `images` and `kwrags` arguments to
AutoFeatureExtractor's [`~AutoFeatureExtractor.__call__`] if `images` is not `None`.
Please refer to the doctsring of the above two methods for more information.
Args:
text (:obj:`str`, :obj:`List[str]`, :obj:`List[List[str]]`):
text (`str`, `List[str]`, `List[List[str]]`):
The sequence or batch of sequences to be encoded. Each sequence can be a string or a list of strings
(pretokenized string). If the sequences are provided as list of strings (pretokenized), you must set
:obj:`is_split_into_words=True` (to lift the ambiguity with a batch of sequences).
images (:obj:`PIL.Image.Image`, :obj:`np.ndarray`, :obj:`torch.Tensor`, :obj:`List[PIL.Image.Image]`, :obj:`List[np.ndarray]`, :obj:`List[torch.Tensor]`):
`is_split_into_words=True` (to lift the ambiguity with a batch of sequences).
images (`PIL.Image.Image`, `np.ndarray`, `torch.Tensor`, `List[PIL.Image.Image]`, `List[np.ndarray]`, `List[torch.Tensor]`):
The image or batch of images to be prepared. Each image can be a PIL image, NumPy array or PyTorch
tensor. In case of a NumPy array/PyTorch tensor, each image should be of shape (C, H, W), where C is a
number of channels, H and W are image height and width.
return_tensors (:obj:`str` or :class:`~transformers.file_utils.TensorType`, `optional`):
return_tensors (`str` or [`~file_utils.TensorType`], *optional*):
If set, will return tensors of a particular framework. Acceptable values are:
* :obj:`'tf'`: Return TensorFlow :obj:`tf.constant` objects.
* :obj:`'pt'`: Return PyTorch :obj:`torch.Tensor` objects.
* :obj:`'np'`: Return NumPy :obj:`np.ndarray` objects.
* :obj:`'jax'`: Return JAX :obj:`jnp.ndarray` objects.
- `'tf'`: Return TensorFlow `tf.constant` objects.
- `'pt'`: Return PyTorch `torch.Tensor` objects.
- `'np'`: Return NumPy `np.ndarray` objects.
- `'jax'`: Return JAX `jnp.ndarray` objects.
Returns:
:class:`~transformers.BatchEncoding`: A :class:`~transformers.BatchEncoding` with the following fields:
[`BatchEncoding`]: A [`BatchEncoding`] with the following fields:
- **input_ids** -- List of token ids to be fed to a model. Returned when :obj:`text` is not :obj:`None`.
- **input_ids** -- List of token ids to be fed to a model. Returned when `text` is not `None`.
- **attention_mask** -- List of indices specifying which tokens should be attended to by the model (when
:obj:`return_attention_mask=True` or if `"attention_mask"` is in :obj:`self.model_input_names` and if
:obj:`text` is not :obj:`None`).
- **pixel_values** -- Pixel values to be fed to a model. Returned when :obj:`images` is not :obj:`None`.
`return_attention_mask=True` or if *"attention_mask"* is in `self.model_input_names` and if
`text` is not `None`).
- **pixel_values** -- Pixel values to be fed to a model. Returned when `images` is not `None`.
"""
if text is None and images is None:
......@@ -171,7 +175,7 @@ class VisionTextDualEncoderProcessor:
def batch_decode(self, *args, **kwargs):
"""
This method forwards all its arguments to VisionTextDualEncoderTokenizer's
:meth:`~transformers.PreTrainedTokenizer.batch_decode`. Please refer to the docstring of this method for more
[`~PreTrainedTokenizer.batch_decode`]. Please refer to the docstring of this method for more
information.
"""
return self.tokenizer.batch_decode(*args, **kwargs)
......@@ -179,7 +183,7 @@ class VisionTextDualEncoderProcessor:
def decode(self, *args, **kwargs):
"""
This method forwards all its arguments to VisionTextDualEncoderTokenizer's
:meth:`~transformers.PreTrainedTokenizer.decode`. Please refer to the docstring of this method for more
[`~PreTrainedTokenizer.decode`]. Please refer to the docstring of this method for more
information.
"""
return self.tokenizer.decode(*args, **kwargs)
src/transformers/models/visual_bert/configuration_visual_bert.py
View file @ 27b3031d
......@@ -36,60 +36,61 @@ VISUAL_BERT_PRETRAINED_CONFIG_ARCHIVE_MAP = {
class VisualBertConfig(PretrainedConfig):
r"""
This is the configuration class to store the configuration of a :class:`~transformers.VisualBertModel`. It is used
This is the configuration class to store the configuration of a [`VisualBertModel`]. It is used
to instantiate an VisualBERT model according to the specified arguments, defining the model architecture.
Instantiating a configuration with the defaults will yield a similar configuration to that of the VisualBERT
`visualbert-vqa-coco-pre <https://huggingface.co/uclanlp/visualbert-vqa-coco-pre>`__ architecture.
[visualbert-vqa-coco-pre](https://huggingface.co/uclanlp/visualbert-vqa-coco-pre) architecture.
Configuration objects inherit from :class:`~transformers.PretrainedConfig` and can be used to control the model
outputs. Read the documentation from :class:`~transformers.PretrainedConfig` for more information.
Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model
outputs. Read the documentation from [`PretrainedConfig`] for more information.
Args:
vocab_size (:obj:`int`, `optional`, defaults to 30522):
vocab_size (`int`, *optional*, defaults to 30522):
Vocabulary size of the VisualBERT model. Defines the number of different tokens that can be represented by
the :obj:`inputs_ids` passed when calling :class:`~transformers.VisualBertModel`. Vocabulary size of the
model. Defines the different tokens that can be represented by the ``inputs_ids`` passed to the forward
method of :class:`~transformers.VisualBertModel`.
hidden_size (:obj:`int`, `optional`, defaults to 768):
the `inputs_ids` passed when calling [`VisualBertModel`]. Vocabulary size of the
model. Defines the different tokens that can be represented by the `inputs_ids` passed to the forward
method of [`VisualBertModel`].
hidden_size (`int`, *optional*, defaults to 768):
Dimensionality of the encoder layers and the pooler layer.
visual_embedding_dim (:obj:`int`, `optional`, defaults to 512):
visual_embedding_dim (`int`, *optional*, defaults to 512):
Dimensionality of the visual embeddings to be passed to the model.
num_hidden_layers (:obj:`int`, `optional`, defaults to 12):
num_hidden_layers (`int`, *optional*, defaults to 12):
Number of hidden layers in the Transformer encoder.
num_attention_heads (:obj:`int`, `optional`, defaults to 12):
num_attention_heads (`int`, *optional*, defaults to 12):
Number of attention heads for each attention layer in the Transformer encoder.
intermediate_size (:obj:`int`, `optional`, defaults to 3072):
intermediate_size (`int`, *optional*, defaults to 3072):
Dimensionality of the "intermediate" (i.e., feed-forward) layer in the Transformer encoder.
hidden_act (:obj:`str` or :obj:`function`, `optional`, defaults to :obj:`"gelu"`):
hidden_act (`str` or `function`, *optional*, defaults to `"gelu"`):
The non-linear activation function (function or string) in the encoder and pooler. If string,
:obj:`"gelu"`, :obj:`"relu"`, :obj:`"selu"` and :obj:`"gelu_new"` are supported.
hidden_dropout_prob (:obj:`float`, `optional`, defaults to 0.1):
`"gelu"`, `"relu"`, `"selu"` and `"gelu_new"` are supported.
hidden_dropout_prob (`float`, *optional*, defaults to 0.1):
The dropout probabilitiy for all fully connected layers in the embeddings, encoder, and pooler.
attention_probs_dropout_prob (:obj:`float`, `optional`, defaults to 0.1):
attention_probs_dropout_prob (`float`, *optional*, defaults to 0.1):
The dropout ratio for the attention probabilities.
max_position_embeddings (:obj:`int`, `optional`, defaults to 512):
max_position_embeddings (`int`, *optional*, defaults to 512):
The maximum sequence length that this model might ever be used with. Typically set this to something large
just in case (e.g., 512 or 1024 or 2048).
type_vocab_size (:obj:`int`, `optional`, defaults to 2):
The vocabulary size of the :obj:`token_type_ids` passed when calling
:class:`~transformers.VisualBertModel`.
initializer_range (:obj:`float`, `optional`, defaults to 0.02):
type_vocab_size (`int`, *optional*, defaults to 2):
The vocabulary size of the `token_type_ids` passed when calling
[`VisualBertModel`].
initializer_range (`float`, *optional*, defaults to 0.02):
The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
layer_norm_eps (:obj:`float`, `optional`, defaults to 1e-12):
layer_norm_eps (`float`, *optional*, defaults to 1e-12):
The epsilon used by the layer normalization layers.
bypass_transformer (:obj:`bool`, `optional`, defaults to :obj:`False`):
Whether or not the model should bypass the transformer for the visual embeddings. If set to :obj:`True`,
the model directly concatenates the visual embeddings from :class:`~transformers.VisualBertEmbeddings` with
bypass_transformer (`bool`, *optional*, defaults to `False`):
Whether or not the model should bypass the transformer for the visual embeddings. If set to `True`,
the model directly concatenates the visual embeddings from [`VisualBertEmbeddings`] with
text output from transformers, and then pass it to a self-attention layer.
special_visual_initialize (:obj:`bool`, `optional`, defaults to :obj:`True`):
special_visual_initialize (`bool`, *optional*, defaults to `True`):
Whether or not the visual token type and position type embedding weights should be initialized the same as
the textual token type and positive type embeddings. When set to :obj:`True`, the weights of the textual
the textual token type and positive type embeddings. When set to `True`, the weights of the textual
token type and position type embeddings are copied to the respective visual embedding layers.
Example::
Example:
```python
>>> from transformers import VisualBertModel, VisualBertConfig
>>> # Initializing a VisualBERT visualbert-vqa-coco-pre style configuration
......@@ -100,7 +101,7 @@ class VisualBertConfig(PretrainedConfig):
>>> # Accessing the model configuration
>>> configuration = model.config
"""
```"""
model_type = "visual_bert"
......
src/transformers/models/visual_bert/modeling_visual_bert.py
View file @ 27b3031d
......@@ -736,9 +736,10 @@ class VisualBertModel(VisualBertPreTrainedModel):
Returns:
Example::
Example:
# Assumption: `get_visual_embeddings(image)` gets the visual embeddings of the image.
```python
# Assumption: *get_visual_embeddings(image)* gets the visual embeddings of the image.
from transformers import BertTokenizer, VisualBertModel
import torch
......@@ -759,7 +760,7 @@ class VisualBertModel(VisualBertPreTrainedModel):
outputs = model(**inputs)
last_hidden_states = outputs.last_hidden_state
"""
```"""
output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
output_hidden_states = (
......
src/transformers/models/vit/configuration_vit.py
View file @ 27b3031d
......@@ -28,47 +28,47 @@ VIT_PRETRAINED_CONFIG_ARCHIVE_MAP = {
class ViTConfig(PretrainedConfig):
r"""
This is the configuration class to store the configuration of a :class:`~transformers.ViTModel`. It is used to
This is the configuration class to store the configuration of a [`ViTModel`]. It is used to
instantiate an ViT model according to the specified arguments, defining the model architecture. Instantiating a
configuration with the defaults will yield a similar configuration to that of the ViT `google/vit-base-patch16-224
<https://huggingface.co/google/vit-base-patch16-224>`__ architecture.
configuration with the defaults will yield a similar configuration to that of the ViT [google/vit-base-patch16-224](https://huggingface.co/google/vit-base-patch16-224) architecture.
Configuration objects inherit from :class:`~transformers.PretrainedConfig` and can be used to control the model
outputs. Read the documentation from :class:`~transformers.PretrainedConfig` for more information.
Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model
outputs. Read the documentation from [`PretrainedConfig`] for more information.
Args:
hidden_size (:obj:`int`, `optional`, defaults to 768):
hidden_size (`int`, *optional*, defaults to 768):
Dimensionality of the encoder layers and the pooler layer.
num_hidden_layers (:obj:`int`, `optional`, defaults to 12):
num_hidden_layers (`int`, *optional*, defaults to 12):
Number of hidden layers in the Transformer encoder.
num_attention_heads (:obj:`int`, `optional`, defaults to 12):
num_attention_heads (`int`, *optional*, defaults to 12):
Number of attention heads for each attention layer in the Transformer encoder.
intermediate_size (:obj:`int`, `optional`, defaults to 3072):
intermediate_size (`int`, *optional*, defaults to 3072):
Dimensionality of the "intermediate" (i.e., feed-forward) layer in the Transformer encoder.
hidden_act (:obj:`str` or :obj:`function`, `optional`, defaults to :obj:`"gelu"`):
hidden_act (`str` or `function`, *optional*, defaults to `"gelu"`):
The non-linear activation function (function or string) in the encoder and pooler. If string,
:obj:`"gelu"`, :obj:`"relu"`, :obj:`"selu"` and :obj:`"gelu_new"` are supported.
hidden_dropout_prob (:obj:`float`, `optional`, defaults to 0.1):
`"gelu"`, `"relu"`, `"selu"` and `"gelu_new"` are supported.
hidden_dropout_prob (`float`, *optional*, defaults to 0.1):
The dropout probabilitiy for all fully connected layers in the embeddings, encoder, and pooler.
attention_probs_dropout_prob (:obj:`float`, `optional`, defaults to 0.1):
attention_probs_dropout_prob (`float`, *optional*, defaults to 0.1):
The dropout ratio for the attention probabilities.
initializer_range (:obj:`float`, `optional`, defaults to 0.02):
initializer_range (`float`, *optional*, defaults to 0.02):
The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
layer_norm_eps (:obj:`float`, `optional`, defaults to 1e-12):
layer_norm_eps (`float`, *optional*, defaults to 1e-12):
The epsilon used by the layer normalization layers.
image_size (:obj:`int`, `optional`, defaults to :obj:`224`):
image_size (`int`, *optional*, defaults to `224`):
The size (resolution) of each image.
patch_size (:obj:`int`, `optional`, defaults to :obj:`16`):
patch_size (`int`, *optional*, defaults to `16`):
The size (resolution) of each patch.
num_channels (:obj:`int`, `optional`, defaults to :obj:`3`):
num_channels (`int`, *optional*, defaults to `3`):
The number of input channels.
qkv_bias (:obj:`bool`, `optional`, defaults to :obj:`True`):
qkv_bias (`bool`, *optional*, defaults to `True`):
Whether to add a bias to the queries, keys and values.
Example::
Example:
```python
>>> from transformers import ViTModel, ViTConfig
>>> # Initializing a ViT vit-base-patch16-224 style configuration
......@@ -79,7 +79,7 @@ class ViTConfig(PretrainedConfig):
>>> # Accessing the model configuration
>>> configuration = model.config
"""
```"""
model_type = "vit"
def __init__(
......
src/transformers/models/vit/feature_extraction_vit.py
View file @ 27b3031d
......@@ -38,25 +38,25 @@ class ViTFeatureExtractor(FeatureExtractionMixin, ImageFeatureExtractionMixin):
r"""
Constructs a ViT feature extractor.
This feature extractor inherits from :class:`~transformers.FeatureExtractionMixin` which contains most of the main
This feature extractor inherits from [`FeatureExtractionMixin`] which contains most of the main
methods. Users should refer to this superclass for more information regarding those methods.
Args:
do_resize (:obj:`bool`, `optional`, defaults to :obj:`True`):
Whether to resize the input to a certain :obj:`size`.
size (:obj:`int` or :obj:`Tuple(int)`, `optional`, defaults to 224):
do_resize (`bool`, *optional*, defaults to `True`):
Whether to resize the input to a certain `size`.
size (`int` or `Tuple(int)`, *optional*, defaults to 224):
Resize the input to the given size. If a tuple is provided, it should be (width, height). If only an
integer is provided, then the input will be resized to (size, size). Only has an effect if :obj:`do_resize`
is set to :obj:`True`.
resample (:obj:`int`, `optional`, defaults to :obj:`PIL.Image.BILINEAR`):
An optional resampling filter. This can be one of :obj:`PIL.Image.NEAREST`, :obj:`PIL.Image.BOX`,
:obj:`PIL.Image.BILINEAR`, :obj:`PIL.Image.HAMMING`, :obj:`PIL.Image.BICUBIC` or :obj:`PIL.Image.LANCZOS`.
Only has an effect if :obj:`do_resize` is set to :obj:`True`.
do_normalize (:obj:`bool`, `optional`, defaults to :obj:`True`):
integer is provided, then the input will be resized to (size, size). Only has an effect if `do_resize`
is set to `True`.
resample (`int`, *optional*, defaults to `PIL.Image.BILINEAR`):
An optional resampling filter. This can be one of `PIL.Image.NEAREST`, `PIL.Image.BOX`,
`PIL.Image.BILINEAR`, `PIL.Image.HAMMING`, `PIL.Image.BICUBIC` or `PIL.Image.LANCZOS`.
Only has an effect if `do_resize` is set to `True`.
do_normalize (`bool`, *optional*, defaults to `True`):
Whether or not to normalize the input with mean and standard deviation.
image_mean (:obj:`List[int]`, defaults to :obj:`[0.5, 0.5, 0.5]`):
image_mean (`List[int]`, defaults to `[0.5, 0.5, 0.5]`):
The sequence of means for each channel, to be used when normalizing images.
image_std (:obj:`List[int]`, defaults to :obj:`[0.5, 0.5, 0.5]`):
image_std (`List[int]`, defaults to `[0.5, 0.5, 0.5]`):
The sequence of standard deviations for each channel, to be used when normalizing images.
"""
......@@ -86,27 +86,29 @@ class ViTFeatureExtractor(FeatureExtractionMixin, ImageFeatureExtractionMixin):
"""
Main method to prepare for the model one or several image(s).
.. warning::
<Tip warning={true}>
NumPy arrays and PyTorch tensors are converted to PIL images when resizing, so the most efficient is to pass
PIL images.
</Tip>
Args:
images (:obj:`PIL.Image.Image`, :obj:`np.ndarray`, :obj:`torch.Tensor`, :obj:`List[PIL.Image.Image]`, :obj:`List[np.ndarray]`, :obj:`List[torch.Tensor]`):
images (`PIL.Image.Image`, `np.ndarray`, `torch.Tensor`, `List[PIL.Image.Image]`, `List[np.ndarray]`, `List[torch.Tensor]`):
The image or batch of images to be prepared. Each image can be a PIL image, NumPy array or PyTorch
tensor. In case of a NumPy array/PyTorch tensor, each image should be of shape (C, H, W), where C is a
number of channels, H and W are image height and width.
return_tensors (:obj:`str` or :class:`~transformers.file_utils.TensorType`, `optional`, defaults to :obj:`'np'`):
return_tensors (`str` or [`~file_utils.TensorType`], *optional*, defaults to `'np'`):
If set, will return tensors of a particular framework. Acceptable values are:
* :obj:`'tf'`: Return TensorFlow :obj:`tf.constant` objects.
* :obj:`'pt'`: Return PyTorch :obj:`torch.Tensor` objects.
* :obj:`'np'`: Return NumPy :obj:`np.ndarray` objects.
* :obj:`'jax'`: Return JAX :obj:`jnp.ndarray` objects.
- `'tf'`: Return TensorFlow `tf.constant` objects.
- `'pt'`: Return PyTorch `torch.Tensor` objects.
- `'np'`: Return NumPy `np.ndarray` objects.
- `'jax'`: Return JAX `jnp.ndarray` objects.
Returns:
:class:`~transformers.BatchFeature`: A :class:`~transformers.BatchFeature` with the following fields:
[`BatchFeature`]: A [`BatchFeature`] with the following fields:
- **pixel_values** -- Pixel values to be fed to a model, of shape (batch_size, num_channels, height,
width).
......
src/transformers/models/vit/modeling_flax_vit.py
View file @ 27b3031d
......@@ -515,8 +515,9 @@ class FlaxViTModel(FlaxViTPreTrainedModel):
FLAX_VISION_MODEL_DOCSTRING = """
Returns:
Examples::
Examples:
```python
>>> from transformers import ViTFeatureExtractor, FlaxViTModel
>>> from PIL import Image
>>> import requests
......@@ -530,6 +531,7 @@ FLAX_VISION_MODEL_DOCSTRING = """
>>> inputs = feature_extractor(images=image, return_tensors="np")
>>> outputs = model(**inputs)
>>> last_hidden_states = outputs.last_hidden_state
```
"""
overwrite_call_docstring(FlaxViTModel, FLAX_VISION_MODEL_DOCSTRING)
......@@ -594,8 +596,9 @@ class FlaxViTForImageClassification(FlaxViTPreTrainedModel):
FLAX_VISION_CLASSIF_DOCSTRING = """
Returns:
Example::
Example:
```python
>>> from transformers import ViTFeatureExtractor, FlaxViTForImageClassification
>>> from PIL import Image
>>> import jax
......@@ -614,6 +617,7 @@ FLAX_VISION_CLASSIF_DOCSTRING = """
>>> # model predicts one of the 1000 ImageNet classes
>>> predicted_class_idx = jax.numpy.argmax(logits, axis=-1)
>>> print("Predicted class:", model.config.id2label[predicted_class_idx.item()])
```
"""
overwrite_call_docstring(FlaxViTForImageClassification, FLAX_VISION_CLASSIF_DOCSTRING)
......
src/transformers/models/vit/modeling_tf_vit.py
View file @ 27b3031d
......@@ -674,8 +674,9 @@ class TFViTModel(TFViTPreTrainedModel):
r"""
Returns:
Examples::
Examples:
```python
>>> from transformers import ViTFeatureExtractor, TFViTModel
>>> from PIL import Image
>>> import requests
......@@ -689,7 +690,7 @@ class TFViTModel(TFViTPreTrainedModel):
>>> inputs = feature_extractor(images=image, return_tensors="tf")
>>> outputs = model(**inputs)
>>> last_hidden_states = outputs.last_hidden_state
"""
```"""
inputs = input_processing(
func=self.call,
config=self.config,
......
src/transformers/models/vit/modeling_vit.py
View file @ 27b3031d
......@@ -517,8 +517,9 @@ class ViTModel(ViTPreTrainedModel):
r"""
Returns:
Examples::
Examples:
```python
>>> from transformers import ViTFeatureExtractor, ViTModel
>>> from PIL import Image
>>> import requests
......@@ -532,7 +533,7 @@ class ViTModel(ViTPreTrainedModel):
>>> inputs = feature_extractor(images=image, return_tensors="pt")
>>> outputs = model(**inputs)
>>> last_hidden_states = outputs.last_hidden_state
"""
```"""
output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
output_hidden_states = (
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
......
src/transformers/models/wav2vec2/configuration_wav2vec2.py
View file @ 27b3031d
......@@ -28,158 +28,154 @@ WAV_2_VEC_2_PRETRAINED_CONFIG_ARCHIVE_MAP = {
class Wav2Vec2Config(PretrainedConfig):
r"""
This is the configuration class to store the configuration of a :class:`~transformers.Wav2Vec2Model`. It is used to
This is the configuration class to store the configuration of a [`Wav2Vec2Model`]. It is used to
instantiate an Wav2Vec2 model according to the specified arguments, defining the model architecture. Instantiating
a configuration with the defaults will yield a similar configuration to that of the Wav2Vec2
`facebook/wav2vec2-base-960h <https://huggingface.co/facebook/wav2vec2-base-960h>`__ architecture.
[facebook/wav2vec2-base-960h](https://huggingface.co/facebook/wav2vec2-base-960h) architecture.
Configuration objects inherit from :class:`~transformers.PretrainedConfig` and can be used to control the model
outputs. Read the documentation from :class:`~transformers.PretrainedConfig` for more information.
Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model
outputs. Read the documentation from [`PretrainedConfig`] for more information.
Args:
vocab_size (:obj:`int`, `optional`, defaults to 32):
vocab_size (`int`, *optional*, defaults to 32):
Vocabulary size of the Wav2Vec2 model. Defines the number of different tokens that can be represented by
the :obj:`inputs_ids` passed when calling :class:`~transformers.Wav2Vec2Model` or
:class:`~transformers.TFWav2Vec2Model`. Vocabulary size of the model. Defines the different tokens that can
be represented by the `inputs_ids` passed to the forward method of :class:`~transformers.Wav2Vec2Model`.
hidden_size (:obj:`int`, `optional`, defaults to 768):
the `inputs_ids` passed when calling [`Wav2Vec2Model`] or
[`TFWav2Vec2Model`]. Vocabulary size of the model. Defines the different tokens that can
be represented by the *inputs_ids* passed to the forward method of [`Wav2Vec2Model`].
hidden_size (`int`, *optional*, defaults to 768):
Dimensionality of the encoder layers and the pooler layer.
num_hidden_layers (:obj:`int`, `optional`, defaults to 12):
num_hidden_layers (`int`, *optional*, defaults to 12):
Number of hidden layers in the Transformer encoder.
num_attention_heads (:obj:`int`, `optional`, defaults to 12):
num_attention_heads (`int`, *optional*, defaults to 12):
Number of attention heads for each attention layer in the Transformer encoder.
intermediate_size (:obj:`int`, `optional`, defaults to 3072):
intermediate_size (`int`, *optional*, defaults to 3072):
Dimensionality of the "intermediate" (i.e., feed-forward) layer in the Transformer encoder.
hidden_act (:obj:`str` or :obj:`function`, `optional`, defaults to :obj:`"gelu"`):
hidden_act (`str` or `function`, *optional*, defaults to `"gelu"`):
The non-linear activation function (function or string) in the encoder and pooler. If string,
:obj:`"gelu"`, :obj:`"relu"`, :obj:`"selu"` and :obj:`"gelu_new"` are supported.
hidden_dropout (:obj:`float`, `optional`, defaults to 0.1):
`"gelu"`, `"relu"`, `"selu"` and `"gelu_new"` are supported.
hidden_dropout (`float`, *optional*, defaults to 0.1):
The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
attention_dropout (:obj:`float`, `optional`, defaults to 0.1):
attention_dropout (`float`, *optional*, defaults to 0.1):
The dropout ratio for the attention probabilities.
final_dropout (:obj:`float`, `optional`, defaults to 0.1):
The dropout probability for the final projection layer of :class:`Wav2Vec2ForCTC`.
initializer_range (:obj:`float`, `optional`, defaults to 0.02):
final_dropout (`float`, *optional*, defaults to 0.1):
The dropout probability for the final projection layer of [`Wav2Vec2ForCTC`].
initializer_range (`float`, *optional*, defaults to 0.02):
The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
layer_norm_eps (:obj:`float`, `optional`, defaults to 1e-12):
layer_norm_eps (`float`, *optional*, defaults to 1e-12):
The epsilon used by the layer normalization layers.
feat_extract_norm (:obj:`str`, `optional`, defaults to :obj:`"group"`):
The norm to be applied to 1D convolutional layers in feature extractor. One of :obj:`"group"` for group
normalization of only the first 1D convolutional layer or :obj:`"layer"` for layer normalization of all 1D
feat_extract_norm (`str`, *optional*, defaults to `"group"`):
The norm to be applied to 1D convolutional layers in feature extractor. One of `"group"` for group
normalization of only the first 1D convolutional layer or `"layer"` for layer normalization of all 1D
convolutional layers.
feat_proj_dropout (:obj:`float`, `optional`, defaults to 0.0):
feat_proj_dropout (`float`, *optional*, defaults to 0.0):
The dropout probability for output of the feature extractor.
feat_extract_activation (:obj:`str, `optional`, defaults to :obj:`"gelu"`):
feat_extract_activation (`str, `optional`, defaults to `"gelu"`):
The non-linear activation function (function or string) in the 1D convolutional layers of the feature
extractor. If string, :obj:`"gelu"`, :obj:`"relu"`, :obj:`"selu"` and :obj:`"gelu_new"` are supported.
feat_quantizer_dropout (obj:`float`, `optional`, defaults to 0.0):
extractor. If string, `"gelu"`, `"relu"`, `"selu"` and `"gelu_new"` are supported.
feat_quantizer_dropout (obj:*float*, *optional*, defaults to 0.0):
The dropout probabilitiy for quantized feature extractor states.
conv_dim (:obj:`Tuple[int]`, `optional`, defaults to :obj:`(512, 512, 512, 512, 512, 512, 512)`):
conv_dim (`Tuple[int]`, *optional*, defaults to `(512, 512, 512, 512, 512, 512, 512)`):
A tuple of integers defining the number of input and output channels of each 1D convolutional layer in the
feature extractor. The length of `conv_dim` defines the number of 1D convolutional layers.
conv_stride (:obj:`Tuple[int]`, `optional`, defaults to :obj:`(5, 2, 2, 2, 2, 2, 2)`):
feature extractor. The length of *conv_dim* defines the number of 1D convolutional layers.
conv_stride (`Tuple[int]`, *optional*, defaults to `(5, 2, 2, 2, 2, 2, 2)`):
A tuple of integers defining the stride of each 1D convolutional layer in the feature extractor. The length
of `conv_stride` defines the number of convolutional layers and has to match the length of `conv_dim`.
conv_kernel (:obj:`Tuple[int]`, `optional`, defaults to :obj:`(10, 3, 3, 3, 3, 3, 3)`):
of *conv_stride* defines the number of convolutional layers and has to match the length of *conv_dim*.
conv_kernel (`Tuple[int]`, *optional*, defaults to `(10, 3, 3, 3, 3, 3, 3)`):
A tuple of integers defining the kernel size of each 1D convolutional layer in the feature extractor. The
length of `conv_kernel` defines the number of convolutional layers and has to match the length of
`conv_dim`.
conv_bias (:obj:`bool`, `optional`, defaults to :obj:`False`):
length of *conv_kernel* defines the number of convolutional layers and has to match the length of
*conv_dim*.
conv_bias (`bool`, *optional*, defaults to `False`):
Whether the 1D convolutional layers have a bias.
num_conv_pos_embeddings (:obj:`int`, `optional`, defaults to 128):
num_conv_pos_embeddings (`int`, *optional*, defaults to 128):
Number of convolutional positional embeddings. Defines the kernel size of 1D convolutional positional
embeddings layer.
num_conv_pos_embedding_groups (:obj:`int`, `optional`, defaults to 16):
num_conv_pos_embedding_groups (`int`, *optional*, defaults to 16):
Number of groups of 1D convolutional positional embeddings layer.
do_stable_layer_norm (:obj:`bool`, `optional`, defaults to :obj:`False`):
Whether to apply `stable` layer norm architecture of the Transformer encoder. ``do_stable_layer_norm is
True`` corresponds to applying layer norm before the attention layer, whereas ``do_stable_layer_norm is
False`` corresponds to applying layer norm after the attention layer.
apply_spec_augment (:obj:`bool`, `optional`, defaults to :obj:`True`):
do_stable_layer_norm (`bool`, *optional*, defaults to `False`):
Whether to apply *stable* layer norm architecture of the Transformer encoder. `do_stable_layer_norm is True` corresponds to applying layer norm before the attention layer, whereas `do_stable_layer_norm is False` corresponds to applying layer norm after the attention layer.
apply_spec_augment (`bool`, *optional*, defaults to `True`):
Whether to apply *SpecAugment* data augmentation to the outputs of the feature extractor. For reference see
`SpecAugment: A Simple Data Augmentation Method for Automatic Speech Recognition
<https://arxiv.org/abs/1904.08779>`__.
mask_time_prob (:obj:`float`, `optional`, defaults to 0.05):
[SpecAugment: A Simple Data Augmentation Method for Automatic Speech Recognition](https://arxiv.org/abs/1904.08779).
mask_time_prob (`float`, *optional*, defaults to 0.05):
Percentage (between 0 and 1) of all feature vectors along the time axis which will be masked. The masking
procecure generates ''mask_time_prob*len(time_axis)/mask_time_length'' independent masks over the axis. If
reasoning from the propability of each feature vector to be chosen as the start of the vector span to be
masked, `mask_time_prob` should be ``prob_vector_start*mask_time_length``. Note that overlap may decrease
the actual percentage of masked vectors. This is only relevant if ``apply_spec_augment is True``.
mask_time_length (:obj:`int`, `optional`, defaults to 10):
masked, *mask_time_prob* should be `prob_vector_start*mask_time_length`. Note that overlap may decrease
the actual percentage of masked vectors. This is only relevant if `apply_spec_augment is True`.
mask_time_length (`int`, *optional*, defaults to 10):
Length of vector span along the time axis.
mask_time_min_masks (:obj:`int`, `optional`, defaults to 2),:
The minimum number of masks of length ``mask_feature_length`` generated along the time axis, each time
step, irrespectively of ``mask_feature_prob``. Only relevant if
mask_time_min_masks (`int`, *optional*, defaults to 2),:
The minimum number of masks of length `mask_feature_length` generated along the time axis, each time
step, irrespectively of `mask_feature_prob`. Only relevant if
''mask_time_prob*len(time_axis)/mask_time_length < mask_time_min_masks''
mask_feature_prob (:obj:`float`, `optional`, defaults to 0.0):
mask_feature_prob (`float`, *optional*, defaults to 0.0):
Percentage (between 0 and 1) of all feature vectors along the feature axis which will be masked. The
masking procecure generates ''mask_feature_prob*len(feature_axis)/mask_time_length'' independent masks over
the axis. If reasoning from the propability of each feature vector to be chosen as the start of the vector
span to be masked, `mask_feature_prob` should be ``prob_vector_start*mask_feature_length``. Note that
overlap may decrease the actual percentage of masked vectors. This is only relevant if ``apply_spec_augment
is True``.
mask_feature_length (:obj:`int`, `optional`, defaults to 10):
span to be masked, *mask_feature_prob* should be `prob_vector_start*mask_feature_length`. Note that
overlap may decrease the actual percentage of masked vectors. This is only relevant if `apply_spec_augment is True`.
mask_feature_length (`int`, *optional*, defaults to 10):
Length of vector span along the feature axis.
mask_feature_min_masks (:obj:`int`, `optional`, defaults to 0),:
The minimum number of masks of length ``mask_feature_length`` generated along the feature axis, each time
step, irrespectively of ``mask_feature_prob``. Only relevant if
mask_feature_min_masks (`int`, *optional*, defaults to 0),:
The minimum number of masks of length `mask_feature_length` generated along the feature axis, each time
step, irrespectively of `mask_feature_prob`. Only relevant if
''mask_feature_prob*len(feature_axis)/mask_feature_length < mask_feature_min_masks''
num_codevectors_per_group (:obj:`int`, `optional`, defaults to 320):
num_codevectors_per_group (`int`, *optional*, defaults to 320):
Number of entries in each quantization codebook (group).
num_codevector_groups (:obj:`int`, `optional`, defaults to 2):
num_codevector_groups (`int`, *optional*, defaults to 2):
Number of codevector groups for product codevector quantization.
contrastive_logits_temperature (:obj:`float`, `optional`, defaults to 0.1):
The temperature `kappa` in the contrastive loss.
feat_quantizer_dropout (:obj:`float`, `optional`, defaults to 0.0):
contrastive_logits_temperature (`float`, *optional*, defaults to 0.1):
The temperature *kappa* in the contrastive loss.
feat_quantizer_dropout (`float`, *optional*, defaults to 0.0):
The dropout probabilitiy for the output of the feature extractor that's used by the quantizer.
num_negatives (:obj:`int`, `optional`, defaults to 100):
num_negatives (`int`, *optional*, defaults to 100):
Number of negative samples for the contrastive loss.
codevector_dim (:obj:`int`, `optional`, defaults to 256):
codevector_dim (`int`, *optional*, defaults to 256):
Dimensionality of the quantized feature vectors.
proj_codevector_dim (:obj:`int`, `optional`, defaults to 256):
proj_codevector_dim (`int`, *optional*, defaults to 256):
Dimensionality of the final projection of both the quantized and the transformer features.
diversity_loss_weight (:obj:`int`, `optional`, defaults to 0.1):
diversity_loss_weight (`int`, *optional*, defaults to 0.1):
The weight of the codebook diversity loss component.
ctc_loss_reduction (:obj:`str`, `optional`, defaults to :obj:`"sum"`):
Specifies the reduction to apply to the output of ``torch.nn.CTCLoss``. Only relevant when training an
instance of :class:`~transformers.Wav2Vec2ForCTC`.
ctc_zero_infinity (:obj:`bool`, `optional`, defaults to :obj:`False`):
Whether to zero infinite losses and the associated gradients of ``torch.nn.CTCLoss``. Infinite losses
ctc_loss_reduction (`str`, *optional*, defaults to `"sum"`):
Specifies the reduction to apply to the output of `torch.nn.CTCLoss`. Only relevant when training an
instance of [`Wav2Vec2ForCTC`].
ctc_zero_infinity (`bool`, *optional*, defaults to `False`):
Whether to zero infinite losses and the associated gradients of `torch.nn.CTCLoss`. Infinite losses
mainly occur when the inputs are too short to be aligned to the targets. Only relevant when training an
instance of :class:`~transformers.Wav2Vec2ForCTC`.
use_weighted_layer_sum (:obj:`bool`, `optional`, defaults to :obj:`False`):
instance of [`Wav2Vec2ForCTC`].
use_weighted_layer_sum (`bool`, *optional*, defaults to `False`):
Whether to use a weighted average of layer outputs with learned weights. Only relevant when using an
instance of :class:`~transformers.Wav2Vec2ForSequenceClassification`.
classifier_proj_size (:obj:`int`, `optional`, defaults to 256):
instance of [`Wav2Vec2ForSequenceClassification`].
classifier_proj_size (`int`, *optional*, defaults to 256):
Dimensionality of the projection before token mean-pooling for classification.
tdnn_dim (:obj:`Tuple[int]`, `optional`, defaults to :obj:`(512, 512, 512, 512, 1500)`):
A tuple of integers defining the number of output channels of each 1D convolutional layer in the `TDNN`
module of the `XVector` model. The length of `tdnn_dim` defines the number of `TDNN` layers.
tdnn_kernel (:obj:`Tuple[int]`, `optional`, defaults to :obj:`(5, 3, 3, 1, 1)`):
A tuple of integers defining the kernel size of each 1D convolutional layer in the `TDNN` module of the
`XVector` model. The length of `tdnn_kernel` has to match the length of `tdnn_dim`.
tdnn_dilation (:obj:`Tuple[int]`, `optional`, defaults to :obj:`(1, 2, 3, 1, 1)`):
A tuple of integers defining the dilation factor of each 1D convolutional layer in `TDNN` module of the
`XVector` model. The length of `tdnn_dilation` has to match the length of `tdnn_dim`.
xvector_output_dim (:obj:`int`, `optional`, defaults to 512):
Dimensionality of the `XVector` embedding vectors.
add_adapter (:obj:`bool`, `optional`, defaults to :obj:`False`):
tdnn_dim (`Tuple[int]`, *optional*, defaults to `(512, 512, 512, 512, 1500)`):
A tuple of integers defining the number of output channels of each 1D convolutional layer in the *TDNN*
module of the *XVector* model. The length of *tdnn_dim* defines the number of *TDNN* layers.
tdnn_kernel (`Tuple[int]`, *optional*, defaults to `(5, 3, 3, 1, 1)`):
A tuple of integers defining the kernel size of each 1D convolutional layer in the *TDNN* module of the
*XVector* model. The length of *tdnn_kernel* has to match the length of *tdnn_dim*.
tdnn_dilation (`Tuple[int]`, *optional*, defaults to `(1, 2, 3, 1, 1)`):
A tuple of integers defining the dilation factor of each 1D convolutional layer in *TDNN* module of the
*XVector* model. The length of *tdnn_dilation* has to match the length of *tdnn_dim*.
xvector_output_dim (`int`, *optional*, defaults to 512):
Dimensionality of the *XVector* embedding vectors.
add_adapter (`bool`, *optional*, defaults to `False`):
Whether a convolutional network should be stacked on top of the Wav2Vec2 Encoder. Can be very useful for
warm-starting Wav2Vec2 for SpeechEncoderDecoder models.
adapter_kernel_size (:obj:`int`, `optional`, defaults to 3):
Kernel size of the convolutional layers in the adapter network. Only relevant if ``add_adapter is True``.
adapter_stride (:obj:`int`, `optional`, defaults to 2):
Stride of the convolutional layers in the adapter network. Only relevant if ``add_adapter is True``.
num_adapter_layers (:obj:`int`, `optional`, defaults to 3):
Number of convolutional layers that should be used in the adapter network. Only relevant if ``add_adapter
is True``.
output_hidden_size (:obj:`int`, `optional`):
Dimensionality of the encoder output layer. If not defined, this defaults to `hidden-size`. Only relevant
if ``add_adapter is True``.
adapter_kernel_size (`int`, *optional*, defaults to 3):
Kernel size of the convolutional layers in the adapter network. Only relevant if `add_adapter is True`.
adapter_stride (`int`, *optional*, defaults to 2):
Stride of the convolutional layers in the adapter network. Only relevant if `add_adapter is True`.
num_adapter_layers (`int`, *optional*, defaults to 3):
Number of convolutional layers that should be used in the adapter network. Only relevant if `add_adapter is True`.
output_hidden_size (`int`, *optional*):
Dimensionality of the encoder output layer. If not defined, this defaults to *hidden-size*. Only relevant
if `add_adapter is True`.
Example::
Example:
```python
>>> from transformers import Wav2Vec2Model, Wav2Vec2Config
>>> # Initializing a Wav2Vec2 facebook/wav2vec2-base-960h style configuration
......@@ -190,7 +186,7 @@ class Wav2Vec2Config(PretrainedConfig):
>>> # Accessing the model configuration
>>> configuration = model.config
"""
```"""
model_type = "wav2vec2"
def __init__(
......
src/transformers/models/wav2vec2/feature_extraction_wav2vec2.py
View file @ 27b3031d
......@@ -34,34 +34,32 @@ class Wav2Vec2FeatureExtractor(SequenceFeatureExtractor):
Constructs a Wav2Vec2 feature extractor.
This feature extractor inherits from
:class:`~transformers.feature_extraction_sequence_utils.SequenceFeatureExtractor` which contains most of the main
[`~feature_extraction_sequence_utils.SequenceFeatureExtractor`] which contains most of the main
methods. Users should refer to this superclass for more information regarding those methods.
Args:
feature_size (:obj:`int`, defaults to 1):
feature_size (`int`, defaults to 1):
The feature dimension of the extracted features.
sampling_rate (:obj:`int`, defaults to 16000):
sampling_rate (`int`, defaults to 16000):
The sampling rate at which the audio files should be digitalized expressed in Hertz per second (Hz).
padding_value (:obj:`float`, defaults to 0.0):
padding_value (`float`, defaults to 0.0):
The value that is used to fill the padding values.
do_normalize (:obj:`bool`, `optional`, defaults to :obj:`False`):
do_normalize (`bool`, *optional*, defaults to `False`):
Whether or not to zero-mean unit-variance normalize the input. Normalizing can help to significantly
improve the performance for some models, *e.g.*, `wav2vec2-lv60
<https://huggingface.co/models?search=lv60>`__.
return_attention_mask (:obj:`bool`, `optional`, defaults to :obj:`False`):
Whether or not :meth:`~transformers.Wav2Vec2FeatureExtractor.__call__` should return :obj:`attention_mask`.
improve the performance for some models, *e.g.*, [wav2vec2-lv60](https://huggingface.co/models?search=lv60).
return_attention_mask (`bool`, *optional*, defaults to `False`):
Whether or not [`~Wav2Vec2FeatureExtractor.__call__`] should return `attention_mask`.
.. note::
<Tip>
Wav2Vec2 models that have set ``config.feat_extract_norm == "group"``, such as `wav2vec2-base
<https://huggingface.co/facebook/wav2vec2-base-960h>`__, have **not** been trained using
:obj:`attention_mask`. For such models, :obj:`input_values` should simply be padded with 0 and no
:obj:`attention_mask` should be passed.
Wav2Vec2 models that have set `config.feat_extract_norm == "group"`, such as [wav2vec2-base](https://huggingface.co/facebook/wav2vec2-base-960h), have **not** been trained using
`attention_mask`. For such models, `input_values` should simply be padded with 0 and no
`attention_mask` should be passed.
For Wav2Vec2 models that have set ``config.feat_extract_norm == "layer"``, such as `wav2vec2-lv60
<https://huggingface.co/facebook/wav2vec2-large-960h-lv60-self>`__, :obj:`attention_mask` should be
For Wav2Vec2 models that have set `config.feat_extract_norm == "layer"`, such as [wav2vec2-lv60](https://huggingface.co/facebook/wav2vec2-large-960h-lv60-self), `attention_mask` should be
passed for batched inference.
"""
</Tip>"""
model_input_names = ["input_values", "attention_mask"]
......@@ -116,55 +114,55 @@ class Wav2Vec2FeatureExtractor(SequenceFeatureExtractor):
Main method to featurize and prepare for the model one or several sequence(s). sequences.
Args:
raw_speech (:obj:`np.ndarray`, :obj:`List[float]`, :obj:`List[np.ndarray]`, :obj:`List[List[float]]`):
raw_speech (`np.ndarray`, `List[float]`, `List[np.ndarray]`, `List[List[float]]`):
The sequence or batch of sequences to be padded. Each sequence can be a numpy array, a list of float
values, a list of numpy arrays or a list of list of float values.
padding (:obj:`bool`, :obj:`str` or :class:`~transformers.file_utils.PaddingStrategy`, `optional`, defaults to :obj:`False`):
padding (`bool`, `str` or [`~file_utils.PaddingStrategy`], *optional*, defaults to `False`):
Select a strategy to pad the returned sequences (according to the model's padding side and padding
index) among:
* :obj:`True` or :obj:`'longest'`: Pad to the longest sequence in the batch (or no padding if only a
- `True` or `'longest'`: Pad to the longest sequence in the batch (or no padding if only a
single sequence if provided).
* :obj:`'max_length'`: Pad to a maximum length specified with the argument :obj:`max_length` or to the
- `'max_length'`: Pad to a maximum length specified with the argument `max_length` or to the
maximum acceptable input length for the model if that argument is not provided.
* :obj:`False` or :obj:`'do_not_pad'` (default): No padding (i.e., can output a batch with sequences of
- `False` or `'do_not_pad'` (default): No padding (i.e., can output a batch with sequences of
different lengths).
max_length (:obj:`int`, `optional`):
max_length (`int`, *optional*):
Maximum length of the returned list and optionally padding length (see above).
truncation (:obj:`bool`):
Activates truncation to cut input sequences longer than `max_length` to `max_length`.
pad_to_multiple_of (:obj:`int`, `optional`):
truncation (`bool`):
Activates truncation to cut input sequences longer than *max_length* to *max_length*.
pad_to_multiple_of (`int`, *optional*):
If set will pad the sequence to a multiple of the provided value.
This is especially useful to enable the use of Tensor Cores on NVIDIA hardware with compute capability
>= 7.5 (Volta), or on TPUs which benefit from having sequence lengths be a multiple of 128.
return_attention_mask (:obj:`bool`, `optional`):
return_attention_mask (`bool`, *optional*):
Whether to return the attention mask. If left to the default, will return the attention mask according
to the specific feature_extractor's default.
`What are attention masks? <../glossary.html#attention-mask>`__
[What are attention masks?](../glossary#attention-mask)
.. note::
<Tip>
Wav2Vec2 models that have set ``config.feat_extract_norm == "group"``, such as `wav2vec2-base
<https://huggingface.co/facebook/wav2vec2-base-960h>`__, have **not** been trained using
:obj:`attention_mask`. For such models, :obj:`input_values` should simply be padded with 0 and no
:obj:`attention_mask` should be passed.
Wav2Vec2 models that have set `config.feat_extract_norm == "group"`, such as [wav2vec2-base](https://huggingface.co/facebook/wav2vec2-base-960h), have **not** been trained using
`attention_mask`. For such models, `input_values` should simply be padded with 0 and no
`attention_mask` should be passed.
For Wav2Vec2 models that have set ``config.feat_extract_norm == "layer"``, such as `wav2vec2-lv60
<https://huggingface.co/facebook/wav2vec2-large-960h-lv60-self>`__, :obj:`attention_mask` should be
For Wav2Vec2 models that have set `config.feat_extract_norm == "layer"`, such as [wav2vec2-lv60](https://huggingface.co/facebook/wav2vec2-large-960h-lv60-self), `attention_mask` should be
passed for batched inference.
return_tensors (:obj:`str` or :class:`~transformers.file_utils.TensorType`, `optional`):
</Tip>
return_tensors (`str` or [`~file_utils.TensorType`], *optional*):
If set, will return tensors instead of list of python integers. Acceptable values are:
* :obj:`'tf'`: Return TensorFlow :obj:`tf.constant` objects.
* :obj:`'pt'`: Return PyTorch :obj:`torch.Tensor` objects.
* :obj:`'np'`: Return Numpy :obj:`np.ndarray` objects.
sampling_rate (:obj:`int`, `optional`):
The sampling rate at which the ``raw_speech`` input was sampled. It is strongly recommended to pass
``sampling_rate`` at the forward call to prevent silent errors.
padding_value (:obj:`float`, defaults to 0.0):
- `'tf'`: Return TensorFlow `tf.constant` objects.
- `'pt'`: Return PyTorch `torch.Tensor` objects.
- `'np'`: Return Numpy `np.ndarray` objects.
sampling_rate (`int`, *optional*):
The sampling rate at which the `raw_speech` input was sampled. It is strongly recommended to pass
`sampling_rate` at the forward call to prevent silent errors.
padding_value (`float`, defaults to 0.0):
"""
if sampling_rate is not None:
......
src/transformers/models/wav2vec2/modeling_flax_wav2vec2.py
View file @ 27b3031d
......@@ -937,8 +937,9 @@ class FlaxWav2Vec2Model(FlaxWav2Vec2PreTrainedModel):
FLAX_WAV2VEC2_MODEL_DOCSTRING = """
Returns:
Example::
Example:
```python
>>> from transformers import Wav2Vec2Processor, FlaxWav2Vec2Model
>>> from datasets import load_dataset
>>> import soundfile as sf
......@@ -956,6 +957,7 @@ FLAX_WAV2VEC2_MODEL_DOCSTRING = """
>>> input_values = processor(ds["speech"][0], sampling_rate=16_000, return_tensors="np").input_values # Batch size 1
>>> hidden_states = model(input_values).last_hidden_state
```
"""
overwrite_call_docstring(
......@@ -1037,8 +1039,9 @@ class FlaxWav2Vec2ForCTC(FlaxWav2Vec2PreTrainedModel):
FLAX_WAV2VEC2_FOR_CTC_DOCSTRING = """
Returns:
Example::
Example:
```python
>>> import jax.numpy as jnp
>>> from transformers import Wav2Vec2Processor, FlaxWav2Vec2ForCTC
>>> from datasets import load_dataset
......@@ -1061,6 +1064,7 @@ FLAX_WAV2VEC2_FOR_CTC_DOCSTRING = """
>>> transcription = processor.decode(predicted_ids[0])
>>> # should give: "A MAN SAID TO THE UNIVERSE SIR I EXIST"
```
"""
overwrite_call_docstring(
......@@ -1108,10 +1112,11 @@ class FlaxWav2Vec2ForPreTrainingModule(nn.Module):
r"""
Returns:
Example::
Example:
```python
"""
```"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
......@@ -1220,8 +1225,9 @@ class FlaxWav2Vec2ForPreTraining(FlaxWav2Vec2PreTrainedModel):
FLAX_WAV2VEC2_FOR_PRETRAINING_DOCSTRING = """
Returns:
Example::
Example:
```python
>>> import optax
>>> import numpy as np
>>> import jax.numpy as jnp
......@@ -1259,6 +1265,7 @@ FLAX_WAV2VEC2_FOR_PRETRAINING_DOCSTRING = """
>>> # show that cosine similarity is much higher than random
>>> assert np.asarray(cosine_sim)[mask_time_indices].mean() > 0.5
```
"""
overwrite_call_docstring(
......
src/transformers/models/wav2vec2/modeling_tf_wav2vec2.py
View file @ 27b3031d
......@@ -1396,8 +1396,9 @@ class TFWav2Vec2Model(TFWav2Vec2PreTrainedModel):
Returns:
Example::
Example:
```python
>>> from transformers import Wav2Vec2Processor, TFWav2Vec2Model
>>> from datasets import load_dataset
>>> import soundfile as sf
......@@ -1415,7 +1416,7 @@ class TFWav2Vec2Model(TFWav2Vec2PreTrainedModel):
>>> input_values = processor(ds["speech"][0], return_tensors="tf").input_values # Batch size 1
>>> hidden_states = model(input_values).last_hidden_state
"""
```"""
inputs = input_values_processing(
func=self.call,
......
src/transformers/models/wav2vec2/processing_wav2vec2.py
View file @ 27b3031d
......@@ -29,16 +29,16 @@ class Wav2Vec2Processor:
Constructs a Wav2Vec2 processor which wraps a Wav2Vec2 feature extractor and a Wav2Vec2 CTC tokenizer into a single
processor.
:class:`~transformers.Wav2Vec2Processor` offers all the functionalities of
:class:`~transformers.Wav2Vec2FeatureExtractor` and :class:`~transformers.PreTrainedTokenizer`. See the docstring
of :meth:`~transformers.Wav2Vec2Processor.__call__` and :meth:`~transformers.Wav2Vec2Processor.decode` for more
[`Wav2Vec2Processor`] offers all the functionalities of
[`Wav2Vec2FeatureExtractor`] and [`PreTrainedTokenizer`]. See the docstring
of [`~Wav2Vec2Processor.__call__`] and [`~Wav2Vec2Processor.decode`] for more
information.
Args:
feature_extractor (:obj:`Wav2Vec2FeatureExtractor`):
An instance of :class:`~transformers.Wav2Vec2FeatureExtractor`. The feature extractor is a required input.
tokenizer (:class:`~transformers.PreTrainedTokenizer`):
An instance of :class:`~transformers.PreTrainedTokenizer`. The tokenizer is a required input.
feature_extractor (`Wav2Vec2FeatureExtractor`):
An instance of [`Wav2Vec2FeatureExtractor`]. The feature extractor is a required input.
tokenizer ([`PreTrainedTokenizer`]):
An instance of [`PreTrainedTokenizer`]. The tokenizer is a required input.
"""
def __init__(self, feature_extractor, tokenizer):
......@@ -57,18 +57,20 @@ class Wav2Vec2Processor:
def save_pretrained(self, save_directory):
"""
Save a Wav2Vec2 feature_extractor object and Wav2Vec2 tokenizer object to the directory ``save_directory``, so
that it can be re-loaded using the :func:`~transformers.Wav2Vec2Processor.from_pretrained` class method.
Save a Wav2Vec2 feature_extractor object and Wav2Vec2 tokenizer object to the directory `save_directory`, so
that it can be re-loaded using the [`~Wav2Vec2Processor.from_pretrained`] class method.
.. note::
<Tip>
This class method is simply calling
:meth:`~transformers.feature_extraction_utils.FeatureExtractionMixin.save_pretrained` and
:meth:`~transformers.tokenization_utils_base.PreTrainedTokenizer.save_pretrained`. Please refer to the
[`~feature_extraction_utils.FeatureExtractionMixin.save_pretrained`] and
[`~tokenization_utils_base.PreTrainedTokenizer.save_pretrained`]. Please refer to the
docstrings of the methods above for more information.
</Tip>
Args:
save_directory (:obj:`str` or :obj:`os.PathLike`):
save_directory (`str` or `os.PathLike`):
Directory where the feature extractor JSON file and the tokenizer files will be saved (directory will
be created if it does not exist).
"""
......@@ -79,30 +81,32 @@ class Wav2Vec2Processor:
@classmethod
def from_pretrained(cls, pretrained_model_name_or_path, **kwargs):
r"""
Instantiate a :class:`~transformers.Wav2Vec2Processor` from a pretrained Wav2Vec2 processor.
Instantiate a [`Wav2Vec2Processor`] from a pretrained Wav2Vec2 processor.
.. note::
<Tip>
This class method is simply calling Wav2Vec2FeatureExtractor's
:meth:`~transformers.feature_extraction_utils.FeatureExtractionMixin.from_pretrained` and
PreTrainedTokenizer's :meth:`~transformers.tokenization_utils_base.PreTrainedTokenizer.from_pretrained`.
[`~feature_extraction_utils.FeatureExtractionMixin.from_pretrained`] and
PreTrainedTokenizer's [`~tokenization_utils_base.PreTrainedTokenizer.from_pretrained`].
Please refer to the docstrings of the methods above for more information.
</Tip>
Args:
pretrained_model_name_or_path (:obj:`str` or :obj:`os.PathLike`):
pretrained_model_name_or_path (`str` or `os.PathLike`):
This can be either:
- a string, the `model id` of a pretrained feature_extractor hosted inside a model repo on
huggingface.co. Valid model ids can be located at the root-level, like ``bert-base-uncased``, or
namespaced under a user or organization name, like ``dbmdz/bert-base-german-cased``.
- a path to a `directory` containing a feature extractor file saved using the
:meth:`~transformers.SequenceFeatureExtractor.save_pretrained` method, e.g.,
``./my_model_directory/``.
- a path or url to a saved feature extractor JSON `file`, e.g.,
``./my_model_directory/preprocessor_config.json``.
- a string, the *model id* of a pretrained feature_extractor hosted inside a model repo on
huggingface.co. Valid model ids can be located at the root-level, like `bert-base-uncased`, or
namespaced under a user or organization name, like `dbmdz/bert-base-german-cased`.
- a path to a *directory* containing a feature extractor file saved using the
[`~SequenceFeatureExtractor.save_pretrained`] method, e.g.,
`./my_model_directory/`.
- a path or url to a saved feature extractor JSON *file*, e.g.,
`./my_model_directory/preprocessor_config.json`.
**kwargs
Additional keyword arguments passed along to both :class:`~transformers.SequenceFeatureExtractor` and
:class:`~transformers.PreTrainedTokenizer`
Additional keyword arguments passed along to both [`SequenceFeatureExtractor`] and
[`PreTrainedTokenizer`]
"""
feature_extractor = Wav2Vec2FeatureExtractor.from_pretrained(pretrained_model_name_or_path, **kwargs)
......@@ -128,9 +132,9 @@ class Wav2Vec2Processor:
def __call__(self, *args, **kwargs):
"""
When used in normal mode, this method forwards all its arguments to Wav2Vec2FeatureExtractor's
:meth:`~transformers.Wav2Vec2FeatureExtractor.__call__` and returns its output. If used in the context
:meth:`~transformers.Wav2Vec2Processor.as_target_processor` this method forwards all its arguments to
PreTrainedTokenizer's :meth:`~transformers.PreTrainedTokenizer.__call__`. Please refer to the docstring of the
[`~Wav2Vec2FeatureExtractor.__call__`] and returns its output. If used in the context
[`~Wav2Vec2Processor.as_target_processor`] this method forwards all its arguments to
PreTrainedTokenizer's [`~PreTrainedTokenizer.__call__`]. Please refer to the docstring of the
above two methods for more information.
"""
return self.current_processor(*args, **kwargs)
......@@ -138,9 +142,9 @@ class Wav2Vec2Processor:
def pad(self, *args, **kwargs):
"""
When used in normal mode, this method forwards all its arguments to Wav2Vec2FeatureExtractor's
:meth:`~transformers.Wav2Vec2FeatureExtractor.pad` and returns its output. If used in the context
:meth:`~transformers.Wav2Vec2Processor.as_target_processor` this method forwards all its arguments to
PreTrainedTokenizer's :meth:`~transformers.PreTrainedTokenizer.pad`. Please refer to the docstring of the above
[`~Wav2Vec2FeatureExtractor.pad`] and returns its output. If used in the context
[`~Wav2Vec2Processor.as_target_processor`] this method forwards all its arguments to
PreTrainedTokenizer's [`~PreTrainedTokenizer.pad`]. Please refer to the docstring of the above
two methods for more information.
"""
return self.current_processor.pad(*args, **kwargs)
......@@ -148,7 +152,7 @@ class Wav2Vec2Processor:
def batch_decode(self, *args, **kwargs):
"""
This method forwards all its arguments to PreTrainedTokenizer's
:meth:`~transformers.PreTrainedTokenizer.batch_decode`. Please refer to the docstring of this method for more
[`~PreTrainedTokenizer.batch_decode`]. Please refer to the docstring of this method for more
information.
"""
return self.tokenizer.batch_decode(*args, **kwargs)
......@@ -156,7 +160,7 @@ class Wav2Vec2Processor:
def decode(self, *args, **kwargs):
"""
This method forwards all its arguments to PreTrainedTokenizer's
:meth:`~transformers.PreTrainedTokenizer.decode`. Please refer to the docstring of this method for more
[`~PreTrainedTokenizer.decode`]. Please refer to the docstring of this method for more
information.
"""
return self.tokenizer.decode(*args, **kwargs)
......
src/transformers/models/wav2vec2/tokenization_wav2vec2.py
View file @ 27b3031d
......@@ -50,31 +50,31 @@ PRETRAINED_VOCAB_FILES_MAP = {
PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES = {"facebook/wav2vec2-base-960h": sys.maxsize}
WAV2VEC2_KWARGS_DOCSTRING = r"""
padding (:obj:`bool`, :obj:`str` or :class:`~transformers.file_utils.PaddingStrategy`, `optional`, defaults to :obj:`False`):
padding (`bool`, `str` or [`~file_utils.PaddingStrategy`], *optional*, defaults to `False`):
Activates and controls padding. Accepts the following values:
* :obj:`True` or :obj:`'longest'`: Pad to the longest sequence in the batch (or no padding if only a
- `True` or `'longest'`: Pad to the longest sequence in the batch (or no padding if only a
single sequence if provided).
* :obj:`'max_length'`: Pad to a maximum length specified with the argument :obj:`max_length` or to the
- `'max_length'`: Pad to a maximum length specified with the argument `max_length` or to the
maximum acceptable input length for the model if that argument is not provided.
* :obj:`False` or :obj:`'do_not_pad'` (default): No padding (i.e., can output a batch with sequences of
- `False` or `'do_not_pad'` (default): No padding (i.e., can output a batch with sequences of
different lengths).
max_length (:obj:`int`, `optional`):
max_length (`int`, *optional*):
Controls the maximum length to use by one of the truncation/padding parameters.
If left unset or set to :obj:`None`, this will use the predefined model maximum length if a maximum
If left unset or set to `None`, this will use the predefined model maximum length if a maximum
length is required by one of the truncation/padding parameters. If the model has no specific maximum
input length (like XLNet) truncation/padding to a maximum length will be deactivated.
pad_to_multiple_of (:obj:`int`, `optional`):
pad_to_multiple_of (`int`, *optional*):
If set will pad the sequence to a multiple of the provided value. This is especially useful to enable
the use of Tensor Cores on NVIDIA hardware with compute capability >= 7.5 (Volta).
return_tensors (:obj:`str` or :class:`~transformers.file_utils.TensorType`, `optional`):
return_tensors (`str` or [`~file_utils.TensorType`], *optional*):
If set, will return tensors instead of list of python integers. Acceptable values are:
* :obj:`'tf'`: Return TensorFlow :obj:`tf.constant` objects.
* :obj:`'pt'`: Return PyTorch :obj:`torch.Tensor` objects.
* :obj:`'np'`: Return Numpy :obj:`np.ndarray` objects.
verbose (:obj:`bool`, `optional`, defaults to :obj:`True`):
- `'tf'`: Return TensorFlow `tf.constant` objects.
- `'pt'`: Return PyTorch `torch.Tensor` objects.
- `'np'`: Return Numpy `np.ndarray` objects.
verbose (`bool`, *optional*, defaults to `True`):
Whether or not to print more information and warnings.
"""
......@@ -84,28 +84,28 @@ class Wav2Vec2CTCTokenizer(PreTrainedTokenizer):
"""
Constructs a Wav2Vec2CTC tokenizer.
This tokenizer inherits from :class:`~transformers.PreTrainedTokenizer` which contains some of the main methods.
This tokenizer inherits from [`PreTrainedTokenizer`] which contains some of the main methods.
Users should refer to the superclass for more information regarding such methods.
Args:
vocab_file (:obj:`str`):
vocab_file (`str`):
File containing the vocabulary.
bos_token (:obj:`str`, `optional`, defaults to :obj:`"<s>"`):
bos_token (`str`, *optional*, defaults to `"<s>"`):
The beginning of sentence token.
eos_token (:obj:`str`, `optional`, defaults to :obj:`"</s>"`):
eos_token (`str`, *optional*, defaults to `"</s>"`):
The end of sentence token.
unk_token (:obj:`str`, `optional`, defaults to :obj:`"<unk>"`):
unk_token (`str`, *optional*, defaults to `"<unk>"`):
The unknown token. A token that is not in the vocabulary cannot be converted to an ID and is set to be this
token instead.
pad_token (:obj:`str`, `optional`, defaults to :obj:`"<pad>"`):
pad_token (`str`, *optional*, defaults to `"<pad>"`):
The token used for padding, for example when batching sequences of different lengths.
word_delimiter_token (:obj:`str`, `optional`, defaults to :obj:`"|"`):
word_delimiter_token (`str`, *optional*, defaults to `"|"`):
The token used for defining the end of a word.
do_lower_case (:obj:`bool`, `optional`, defaults to :obj:`False`):
do_lower_case (`bool`, *optional*, defaults to `False`):
Whether or not to accept lowercase input and lowercase the output when decoding.
**kwargs
Additional keyword arguments passed along to :class:`~transformers.PreTrainedTokenizer`
Additional keyword arguments passed along to [`PreTrainedTokenizer`]
"""
vocab_files_names = VOCAB_FILES_NAMES
......@@ -153,7 +153,7 @@ class Wav2Vec2CTCTokenizer(PreTrainedTokenizer):
@property
def word_delimiter_token(self) -> str:
"""
:obj:`str`: Word delimiter token. Log an error if used while not having been set.
`str`: Word delimiter token. Log an error if used while not having been set.
"""
if self._word_delimiter_token is None and self.verbose:
logger.error("Using word_delimiter_token, but it is not set yet.")
......@@ -163,7 +163,7 @@ class Wav2Vec2CTCTokenizer(PreTrainedTokenizer):
@property
def word_delimiter_token_id(self) -> Optional[int]:
"""
:obj:`Optional[int]`: Id of the word_delimiter_token in the vocabulary. Returns :obj:`None` if the token has
`Optional[int]`: Id of the word_delimiter_token in the vocabulary. Returns `None` if the token has
not been set.
"""
if self._word_delimiter_token is None:
......@@ -285,17 +285,18 @@ class Wav2Vec2CTCTokenizer(PreTrainedTokenizer):
it with indices starting from length of the current vocabulary.
Args:
new_tokens (:obj:`List[str]`or :obj:`List[tokenizers.AddedToken]`):
new_tokens (`List[str]`or `List[tokenizers.AddedToken]`):
Token(s) to add in vocabulary. A token is only added if it's not already in the vocabulary (tested by
checking if the tokenizer assign the index of the ``unk_token`` to them).
special_tokens (:obj:`bool`, `optional`, defaults to :obj:`False`):
checking if the tokenizer assign the index of the `unk_token` to them).
special_tokens (`bool`, *optional*, defaults to `False`):
Whether or not the tokens should be added as special tokens.
Returns:
:obj:`int`: The number of tokens actually added to the vocabulary.
`int`: The number of tokens actually added to the vocabulary.
Examples::
Examples:
```python
# Let's see how to increase the vocabulary of Bert model and tokenizer
tokenizer = Wav2Vec2CTCTokenizer.from_pretrained('facebook/wav2vec2-base-960h')
model = Wav2Vec2ForCTC.from_pretrained('facebook/wav2vec2-base-960h')
......@@ -304,7 +305,7 @@ class Wav2Vec2CTCTokenizer(PreTrainedTokenizer):
print('We have added', num_added_toks, 'tokens')
# Note: resize_token_embeddings expects to receive the full size of the new vocabulary, i.e. the length of the tokenizer.
model.resize_token_embeddings(len(tokenizer))
"""
```"""
new_tokens = [str(tok) for tok in new_tokens]
tokens_to_add = []
......@@ -341,45 +342,44 @@ class Wav2Vec2Tokenizer(PreTrainedTokenizer):
"""
Constructs a Wav2Vec2 tokenizer.
This tokenizer inherits from :class:`~transformers.PreTrainedTokenizer` which contains some of the main methods.
This tokenizer inherits from [`PreTrainedTokenizer`] which contains some of the main methods.
Users should refer to the superclass for more information regarding such methods.
Args:
vocab_file (:obj:`str`):
vocab_file (`str`):
File containing the vocabulary.
bos_token (:obj:`str`, `optional`, defaults to :obj:`"<s>"`):
bos_token (`str`, *optional*, defaults to `"<s>"`):
The beginning of sentence token.
eos_token (:obj:`str`, `optional`, defaults to :obj:`"</s>"`):
eos_token (`str`, *optional*, defaults to `"</s>"`):
The end of sentence token.
unk_token (:obj:`str`, `optional`, defaults to :obj:`"<unk>"`):
unk_token (`str`, *optional*, defaults to `"<unk>"`):
The unknown token. A token that is not in the vocabulary cannot be converted to an ID and is set to be this
token instead.
pad_token (:obj:`str`, `optional`, defaults to :obj:`"<pad>"`):
pad_token (`str`, *optional*, defaults to `"<pad>"`):
The token used for padding, for example when batching sequences of different lengths.
word_delimiter_token (:obj:`str`, `optional`, defaults to :obj:`"|"`):
word_delimiter_token (`str`, *optional*, defaults to `"|"`):
The token used for defining the end of a word.
do_lower_case (:obj:`bool`, `optional`, defaults to :obj:`False`):
do_lower_case (`bool`, *optional*, defaults to `False`):
Whether or not to lowercase the output when decoding.
do_normalize (:obj:`bool`, `optional`, defaults to :obj:`False`):
do_normalize (`bool`, *optional*, defaults to `False`):
Whether or not to zero-mean unit-variance normalize the input. Normalizing can help to significantly
improve the performance for some models, *e.g.*, `wav2vec2-lv60
<https://huggingface.co/models?search=lv60>`__.
return_attention_mask (:obj:`bool`, `optional`, defaults to :obj:`False`):
Whether or not :meth:`~transformers.Wav2Vec2Tokenizer.__call__` should return :obj:`attention_mask`.
improve the performance for some models, *e.g.*, [wav2vec2-lv60](https://huggingface.co/models?search=lv60).
return_attention_mask (`bool`, *optional*, defaults to `False`):
Whether or not [`~Wav2Vec2Tokenizer.__call__`] should return `attention_mask`.
.. note::
<Tip>
Wav2Vec2 models that have set ``config.feat_extract_norm == "group"``, such as `wav2vec2-base
<https://huggingface.co/facebook/wav2vec2-base-960h>`__, have **not** been trained using
:obj:`attention_mask`. For such models, :obj:`input_values` should simply be padded with 0 and no
:obj:`attention_mask` should be passed.
Wav2Vec2 models that have set `config.feat_extract_norm == "group"`, such as [wav2vec2-base](https://huggingface.co/facebook/wav2vec2-base-960h), have **not** been trained using
`attention_mask`. For such models, `input_values` should simply be padded with 0 and no
`attention_mask` should be passed.
For Wav2Vec2 models that have set ``config.feat_extract_norm == "layer"``, such as `wav2vec2-lv60
<https://huggingface.co/facebook/wav2vec2-large-960h-lv60-self>`__, :obj:`attention_mask` should be
For Wav2Vec2 models that have set `config.feat_extract_norm == "layer"`, such as [wav2vec2-lv60](https://huggingface.co/facebook/wav2vec2-large-960h-lv60-self), `attention_mask` should be
passed for batched inference.
</Tip>
**kwargs
Additional keyword arguments passed along to :class:`~transformers.PreTrainedTokenizer`
Additional keyword arguments passed along to [`PreTrainedTokenizer`]
"""
vocab_files_names = VOCAB_FILES_NAMES
......@@ -437,7 +437,7 @@ class Wav2Vec2Tokenizer(PreTrainedTokenizer):
@property
def word_delimiter_token(self) -> str:
"""
:obj:`str`: Padding token. Log an error if used while not having been set.
`str`: Padding token. Log an error if used while not having been set.
"""
if self._word_delimiter_token is None and self.verbose:
logger.error("Using word_delimiter_token, but it is not set yet.")
......@@ -447,7 +447,7 @@ class Wav2Vec2Tokenizer(PreTrainedTokenizer):
@property
def word_delimiter_token_id(self) -> Optional[int]:
"""
:obj:`Optional[int]`: Id of the word_delimiter_token in the vocabulary. Returns :obj:`None` if the token has
`Optional[int]`: Id of the word_delimiter_token in the vocabulary. Returns `None` if the token has
not been set.
"""
if self._word_delimiter_token is None:
......@@ -478,7 +478,7 @@ class Wav2Vec2Tokenizer(PreTrainedTokenizer):
sequences.
Args:
raw_speech (:obj:`np.ndarray`, :obj:`List[float]`, :obj:`List[np.ndarray]`, :obj:`List[List[float]]`):
raw_speech (`np.ndarray`, `List[float]`, `List[np.ndarray]`, `List[List[float]]`):
The sequence or batch of sequences to be padded. Each sequence can be a numpy array, a list of float
values, a list of numpy arrayr or a list of list of float values.
"""
......
src/transformers/models/wav2vec2_phoneme/tokenization_wav2vec2_phoneme.py
View file @ 27b3031d
......@@ -52,32 +52,32 @@ class Wav2Vec2PhonemeCTCTokenizer(PreTrainedTokenizer):
"""
Constructs a Wav2Vec2PhonemeCTC tokenizer.
This tokenizer inherits from :class:`~transformers.PreTrainedTokenizer` which contains some of the main methods.
This tokenizer inherits from [`PreTrainedTokenizer`] which contains some of the main methods.
Users should refer to the superclass for more information regarding such methods.
Args:
vocab_file (:obj:`str`):
vocab_file (`str`):
File containing the vocabulary.
bos_token (:obj:`str`, `optional`, defaults to :obj:`"<s>"`):
bos_token (`str`, *optional*, defaults to `"<s>"`):
The beginning of sentence token.
eos_token (:obj:`str`, `optional`, defaults to :obj:`"</s>"`):
eos_token (`str`, *optional*, defaults to `"</s>"`):
The end of sentence token.
unk_token (:obj:`str`, `optional`, defaults to :obj:`"<unk>"`):
unk_token (`str`, *optional*, defaults to `"<unk>"`):
The unknown token. A token that is not in the vocabulary cannot be converted to an ID and is set to be this
token instead.
pad_token (:obj:`str`, `optional`, defaults to :obj:`"<pad>"`):
pad_token (`str`, *optional*, defaults to `"<pad>"`):
The token used for padding, for example when batching sequences of different lengths.
do_phonemize (:obj:`bool`, `optional`, defaults to :obj:`True`):
do_phonemize (`bool`, *optional*, defaults to `True`):
Whether the tokenizer should phonetize the input or not. Only if a sequence of phonemes is passed to the
tokenizer, :obj:`do_phonemize` should be set to ``False``.
phonemizer_lang (:obj:`str`, `optional`, defaults to :obj:`"en-us"`):
tokenizer, `do_phonemize` should be set to `False`.
phonemizer_lang (`str`, *optional*, defaults to `"en-us"`):
The language of the phoneme set to which the tokenizer should phonetize the input text to.
phonemizer_backend (:obj:`str`, `optional`. defaults to :obj:`"espeak"`):
The backend phonetization library that shall be used by the phonemizer library. Defaults to ``espeak-ng``.
See the `phonemizer package <https://github.com/bootphon/phonemizer#readme>`_. for more information.
phonemizer_backend (`str`, *optional*. defaults to `"espeak"`):
The backend phonetization library that shall be used by the phonemizer library. Defaults to `espeak-ng`.
See the [phonemizer package](https://github.com/bootphon/phonemizer#readme). for more information.
**kwargs
Additional keyword arguments passed along to :class:`~transformers.PreTrainedTokenizer`
Additional keyword arguments passed along to [`PreTrainedTokenizer`]
"""
vocab_files_names = VOCAB_FILES_NAMES
......@@ -139,25 +139,25 @@ class Wav2Vec2PhonemeCTCTokenizer(PreTrainedTokenizer):
"""
Performs any necessary transformations before tokenization.
This method should pop the arguments from kwargs and return the remaining :obj:`kwargs` as well. We test the
:obj:`kwargs` at the end of the encoding process to be sure all the arguments have been used.
This method should pop the arguments from kwargs and return the remaining `kwargs` as well. We test the
`kwargs` at the end of the encoding process to be sure all the arguments have been used.
Args:
text (:obj:`str`):
text (`str`):
The text to prepare.
is_split_into_words (:obj:`bool`, `optional`, defaults to :obj:`False`):
Whether or not the input is already pre-tokenized (e.g., split into words). If set to :obj:`True`, the
is_split_into_words (`bool`, *optional*, defaults to `False`):
Whether or not the input is already pre-tokenized (e.g., split into words). If set to `True`, the
tokenizer assumes the input is already split into words (for instance, by splitting it on whitespace)
which it will tokenize. This is useful for NER or token classification.
phonemizer_lang (:obj:`str`, `optional`):
phonemizer_lang (`str`, *optional*):
The language of the phoneme set to which the tokenizer should phonetize the input text to.
do_phonemize (:obj:`bool`, `optional`):
do_phonemize (`bool`, *optional*):
Whether the tokenizer should phonetize the input text or not. Only if a sequence of phonemes is passed
to the tokenizer, :obj:`do_phonemize` should be set to ``False``.
to the tokenizer, `do_phonemize` should be set to `False`.
Returns:
:obj:`Tuple[str, Dict[str, Any]]`: The prepared text and the unused kwargs.
`Tuple[str, Dict[str, Any]]`: The prepared text and the unused kwargs.
"""
if is_split_into_words:
text = " " + text
......@@ -217,7 +217,7 @@ class Wav2Vec2PhonemeCTCTokenizer(PreTrainedTokenizer):
@property
def word_delimiter_token(self) -> str:
"""
:obj:`str`: Word delimiter token. Log an error if used while not having been set.
`str`: Word delimiter token. Log an error if used while not having been set.
"""
if self._word_delimiter_token is None and self.verbose:
return None
......@@ -226,7 +226,7 @@ class Wav2Vec2PhonemeCTCTokenizer(PreTrainedTokenizer):
@property
def word_delimiter_token_id(self) -> Optional[int]:
"""
:obj:`Optional[int]`: Id of the word_delimiter_token in the vocabulary. Returns :obj:`None` if the token has
`Optional[int]`: Id of the word_delimiter_token in the vocabulary. Returns `None` if the token has
not been set.
"""
if self._word_delimiter_token is None:
......@@ -244,7 +244,7 @@ class Wav2Vec2PhonemeCTCTokenizer(PreTrainedTokenizer):
@property
def phone_delimiter_token(self) -> str:
"""
:obj:`str`: Word delimiter token. Log an error if used while not having been set.
`str`: Word delimiter token. Log an error if used while not having been set.
"""
if self._phone_delimiter_token is None and self.verbose:
logger.error("Using phone_delimiter_token, but it is not set yet.")
......@@ -254,7 +254,7 @@ class Wav2Vec2PhonemeCTCTokenizer(PreTrainedTokenizer):
@property
def phone_delimiter_token_id(self) -> Optional[int]:
"""
:obj:`Optional[int]`: Id of the phone_delimiter_token in the vocabulary. Returns :obj:`None` if the token has
`Optional[int]`: Id of the phone_delimiter_token in the vocabulary. Returns `None` if the token has
not been set.
"""
if self._phone_delimiter_token is None:
......@@ -357,17 +357,18 @@ class Wav2Vec2PhonemeCTCTokenizer(PreTrainedTokenizer):
it with indices starting from length of the current vocabulary.
Args:
new_tokens (:obj:`List[str]`or :obj:`List[tokenizers.AddedToken]`):
new_tokens (`List[str]`or `List[tokenizers.AddedToken]`):
Token(s) to add in vocabulary. A token is only added if it's not already in the vocabulary (tested by
checking if the tokenizer assign the index of the ``unk_token`` to them).
special_tokens (:obj:`bool`, `optional`, defaults to :obj:`False`):
checking if the tokenizer assign the index of the `unk_token` to them).
special_tokens (`bool`, *optional*, defaults to `False`):
Whether or not the tokens should be added as special tokens.
Returns:
:obj:`int`: The number of tokens actually added to the vocabulary.
`int`: The number of tokens actually added to the vocabulary.
Examples::
Examples:
```python
# Let's see how to increase the vocabulary of Bert model and tokenizer
tokenizer = Wav2Vec2PhonemeCTCTokenizer.from_pretrained('facebook/wav2vec2-lv-60-espeak-cv-ft')
model = Wav2Vec2PhonemeForCTC.from_pretrained('facebook/wav2vec2-lv-60-espeak-cv-ft')
......@@ -376,7 +377,7 @@ class Wav2Vec2PhonemeCTCTokenizer(PreTrainedTokenizer):
print('We have added', num_added_toks, 'tokens')
# Note: resize_token_embeddings expects to receive the full size of the new vocabulary, i.e. the length of the tokenizer.
model.resize_token_embeddings(len(tokenizer))
"""
```"""
new_tokens = [str(tok) for tok in new_tokens]
tokens_to_add = []
......
src/transformers/models/wav2vec2_with_lm/processing_wav2vec2_with_lm.py
View file @ 27b3031d
......@@ -37,10 +37,10 @@ if TYPE_CHECKING:
@dataclass
class Wav2Vec2DecoderWithLMOutput(ModelOutput):
"""
Output type of :class:`~transformers.Wav2Vec2DecoderWithLM`, with transcription.
Output type of [`Wav2Vec2DecoderWithLM`], with transcription.
Args:
text (list of :obj:`str`):
text (list of `str`):
Decoded logits in text from. Usually the speech transcription.
"""
......@@ -53,12 +53,12 @@ class Wav2Vec2ProcessorWithLM:
with language model support into a single processor for language model boosted speech recognition decoding.
Args:
feature_extractor (:class:`~transformers.Wav2Vec2FeatureExtractor`):
An instance of :class:`~transformers.Wav2Vec2FeatureExtractor`. The feature extractor is a required input.
tokenizer (:class:`~transformers.Wav2Vec2CTCTokenizer`):
An instance of :class:`~transformers.Wav2Vec2CTCTokenizer`. The tokenizer is a required input.
decoder (:obj:`pyctcdecode.BeamSearchDecoderCTC`):
An instance of :class:`pyctcdecode.BeamSearchDecoderCTC`. The decoder is a required input.
feature_extractor ([`Wav2Vec2FeatureExtractor`]):
An instance of [`Wav2Vec2FeatureExtractor`]. The feature extractor is a required input.
tokenizer ([`Wav2Vec2CTCTokenizer`]):
An instance of [`Wav2Vec2CTCTokenizer`]. The tokenizer is a required input.
decoder (`pyctcdecode.BeamSearchDecoderCTC`):
An instance of [`pyctcdecode.BeamSearchDecoderCTC`]. The decoder is a required input.
"""
def __init__(
......@@ -98,20 +98,22 @@ class Wav2Vec2ProcessorWithLM:
def save_pretrained(self, save_directory):
"""
Save the Wav2Vec2 feature_extractor, a tokenizer object and a pyctcdecode decoder to the directory
``save_directory``, so that they can be re-loaded using the
:func:`~transformers.Wav2Vec2ProcessorWithLM.from_pretrained` class method.
`save_directory`, so that they can be re-loaded using the
[`~Wav2Vec2ProcessorWithLM.from_pretrained`] class method.
.. note::
<Tip>
This class method is simply calling
:meth:`~transformers.feature_extraction_utils.FeatureExtractionMixin.save_pretrained,`
:meth:`~transformers.tokenization_utils_base.PreTrainedTokenizer.save_pretrained` and pyctcdecode's
:meth:`pyctcdecode.BeamSearchDecoderCTC.save_to_dir`.
[`~feature_extraction_utils.FeatureExtractionMixin.save_pretrained,`]
[`~tokenization_utils_base.PreTrainedTokenizer.save_pretrained`] and pyctcdecode's
[`pyctcdecode.BeamSearchDecoderCTC.save_to_dir`].
Please refer to the docstrings of the methods above for more information.
</Tip>
Args:
save_directory (:obj:`str` or :obj:`os.PathLike`):
save_directory (`str` or `os.PathLike`):
Directory where the feature extractor JSON file and the tokenizer files will be saved (directory will
be created if it does not exist).
"""
......@@ -122,32 +124,34 @@ class Wav2Vec2ProcessorWithLM:
@classmethod
def from_pretrained(cls, pretrained_model_name_or_path, **kwargs):
r"""
Instantiate a :class:`~transformers.Wav2Vec2ProcessorWithLM` from a pretrained Wav2Vec2 processor.
Instantiate a [`Wav2Vec2ProcessorWithLM`] from a pretrained Wav2Vec2 processor.
.. note::
<Tip>
This class method is simply calling Wav2Vec2FeatureExtractor's
:meth:`~transformers.feature_extraction_utils.FeatureExtractionMixin.from_pretrained`,
Wav2Vec2CTCTokenizer's :meth:`~transformers.tokenization_utils_base.PreTrainedTokenizer.from_pretrained`,
and :meth:`pyctcdecode.BeamSearchDecoderCTC.load_from_hf_hub`.
[`~feature_extraction_utils.FeatureExtractionMixin.from_pretrained`],
Wav2Vec2CTCTokenizer's [`~tokenization_utils_base.PreTrainedTokenizer.from_pretrained`],
and [`pyctcdecode.BeamSearchDecoderCTC.load_from_hf_hub`].
Please refer to the docstrings of the methods above for more information.
</Tip>
Args:
pretrained_model_name_or_path (:obj:`str` or :obj:`os.PathLike`):
pretrained_model_name_or_path (`str` or `os.PathLike`):
This can be either:
- a string, the `model id` of a pretrained feature_extractor hosted inside a model repo on
huggingface.co. Valid model ids can be located at the root-level, like ``bert-base-uncased``, or
namespaced under a user or organization name, like ``dbmdz/bert-base-german-cased``.
- a path to a `directory` containing a feature extractor file saved using the
:meth:`~transformers.SequenceFeatureExtractor.save_pretrained` method, e.g.,
``./my_model_directory/``.
- a path or url to a saved feature extractor JSON `file`, e.g.,
``./my_model_directory/preprocessor_config.json``.
- a string, the *model id* of a pretrained feature_extractor hosted inside a model repo on
huggingface.co. Valid model ids can be located at the root-level, like `bert-base-uncased`, or
namespaced under a user or organization name, like `dbmdz/bert-base-german-cased`.
- a path to a *directory* containing a feature extractor file saved using the
[`~SequenceFeatureExtractor.save_pretrained`] method, e.g.,
`./my_model_directory/`.
- a path or url to a saved feature extractor JSON *file*, e.g.,
`./my_model_directory/preprocessor_config.json`.
**kwargs
Additional keyword arguments passed along to both :class:`~transformers.SequenceFeatureExtractor` and
:class:`~transformers.PreTrainedTokenizer`
Additional keyword arguments passed along to both [`SequenceFeatureExtractor`] and
[`PreTrainedTokenizer`]
"""
requires_backends(cls, "pyctcdecode")
from pyctcdecode import BeamSearchDecoderCTC
......@@ -215,9 +219,9 @@ class Wav2Vec2ProcessorWithLM:
def __call__(self, *args, **kwargs):
"""
When used in normal mode, this method forwards all its arguments to Wav2Vec2FeatureExtractor's
:meth:`~transformers.Wav2Vec2FeatureExtractor.__call__` and returns its output. If used in the context
:meth:`~transformers.Wav2Vec2ProcessorWithLM.as_target_processor` this method forwards all its arguments to
Wav2Vec2CTCTokenizer's :meth:`~transformers.Wav2Vec2CTCTokenizer.__call__`. Please refer to the docstring of
[`~Wav2Vec2FeatureExtractor.__call__`] and returns its output. If used in the context
[`~Wav2Vec2ProcessorWithLM.as_target_processor`] this method forwards all its arguments to
Wav2Vec2CTCTokenizer's [`~Wav2Vec2CTCTokenizer.__call__`]. Please refer to the docstring of
the above two methods for more information.
"""
return self.current_processor(*args, **kwargs)
......@@ -225,9 +229,9 @@ class Wav2Vec2ProcessorWithLM:
def pad(self, *args, **kwargs):
"""
When used in normal mode, this method forwards all its arguments to Wav2Vec2FeatureExtractor's
:meth:`~transformers.Wav2Vec2FeatureExtractor.pad` and returns its output. If used in the context
:meth:`~transformers.Wav2Vec2ProcessorWithLM.as_target_processor` this method forwards all its arguments to
Wav2Vec2CTCTokenizer's :meth:`~transformers.Wav2Vec2CTCTokenizer.pad`. Please refer to the docstring of the
[`~Wav2Vec2FeatureExtractor.pad`] and returns its output. If used in the context
[`~Wav2Vec2ProcessorWithLM.as_target_processor`] this method forwards all its arguments to
Wav2Vec2CTCTokenizer's [`~Wav2Vec2CTCTokenizer.pad`]. Please refer to the docstring of the
above two methods for more information.
"""
return self.current_processor.pad(*args, **kwargs)
......@@ -245,30 +249,32 @@ class Wav2Vec2ProcessorWithLM:
"""
Batch decode output logits to audio transcription with language model support.
.. note::
<Tip>
This function makes use of Python's multiprocessing.
</Tip>
Args:
logits (:obj:`np.ndarray`):
logits (`np.ndarray`):
The logits output vector of the model representing the log probabilities for each token.
num_processes (:obj:`int`, `optional`):
num_processes (`int`, *optional*):
Number of processes on which the function should be parallelized over. Defaults to the number of
available CPUs.
beam_width (:obj:`int`, `optional`):
beam_width (`int`, *optional*):
Maximum number of beams at each step in decoding. Defaults to pyctcdecode's DEFAULT_BEAM_WIDTH.
beam_prune_logp (:obj:`int`, `optional`):
beam_prune_logp (`int`, *optional*):
Beams that are much worse than best beam will be pruned Defaults to pyctcdecode's DEFAULT_PRUNE_LOGP.
token_min_logp (:obj:`int`, `optional`):
token_min_logp (`int`, *optional*):
Tokens below this logp are skipped unless they are argmax of frame Defaults to pyctcdecode's
DEFAULT_MIN_TOKEN_LOGP.
hotwords (:obj:`List[str]`, `optional`):
hotwords (`List[str]`, *optional*):
List of words with extra importance, can be OOV for LM
hotword_weight (:obj:`int`, `optional`):
hotword_weight (`int`, *optional*):
Weight factor for hotword importance Defaults to pyctcdecode's DEFAULT_HOTWORD_WEIGHT.
Returns:
:class:`~transformers.models.wav2vec2.Wav2Vec2DecoderWithLMOutput` or :obj:`tuple`.
[`~models.wav2vec2.Wav2Vec2DecoderWithLMOutput`] or `tuple`.
"""
from pyctcdecode.constants import (
......@@ -318,23 +324,23 @@ class Wav2Vec2ProcessorWithLM:
Decode output logits to audio transcription with language model support.
Args:
logits (:obj:`np.ndarray`):
logits (`np.ndarray`):
The logits output vector of the model representing the log probabilities for each token.
beam_width (:obj:`int`, `optional`):
beam_width (`int`, *optional*):
Maximum number of beams at each step in decoding. Defaults to pyctcdecode's DEFAULT_BEAM_WIDTH.
beam_prune_logp (:obj:`int`, `optional`):
beam_prune_logp (`int`, *optional*):
A threshold to prune beams with log-probs less than best_beam_logp + beam_prune_logp. The value should
be <= 0. Defaults to pyctcdecode's DEFAULT_PRUNE_LOGP.
token_min_logp (:obj:`int`, `optional`):
token_min_logp (`int`, *optional*):
Tokens with log-probs below token_min_logp are skipped unless they are have the maximum log-prob for an
utterance. Defaults to pyctcdecode's DEFAULT_MIN_TOKEN_LOGP.
hotwords (:obj:`List[str]`, `optional`):
hotwords (`List[str]`, *optional*):
List of words with extra importance which can be missing from the LM's vocabulary, e.g. ["huggingface"]
hotword_weight (:obj:`int`, `optional`):
hotword_weight (`int`, *optional*):
Weight multiplier that boosts hotword scores. Defaults to pyctcdecode's DEFAULT_HOTWORD_WEIGHT.
Returns:
:class:`~transformers.models.wav2vec2.Wav2Vec2DecoderWithLMOutput` or :obj:`tuple`.
[`~models.wav2vec2.Wav2Vec2DecoderWithLMOutput`] or `tuple`.
"""
from pyctcdecode.constants import (
......
src/transformers/models/wavlm/configuration_wavlm.py
View file @ 27b3031d
......@@ -28,151 +28,151 @@ WAVLM_PRETRAINED_CONFIG_ARCHIVE_MAP = {
class WavLMConfig(PretrainedConfig):
r"""
This is the configuration class to store the configuration of a :class:`~transformers.WavLMModel`. It is used to
This is the configuration class to store the configuration of a [`WavLMModel`]. It is used to
instantiate an WavLM model according to the specified arguments, defining the model architecture. Instantiating a
configuration with the defaults will yield a similar configuration to that of the WavLM `facebook/wavlm-base-960h
<https://huggingface.co/facebook/wavlm-base-960h>`__ architecture.
configuration with the defaults will yield a similar configuration to that of the WavLM [facebook/wavlm-base-960h](https://huggingface.co/facebook/wavlm-base-960h) architecture.
Configuration objects inherit from :class:`~transformers.PretrainedConfig` and can be used to control the model
outputs. Read the documentation from :class:`~transformers.PretrainedConfig` for more information.
Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model
outputs. Read the documentation from [`PretrainedConfig`] for more information.
Args:
vocab_size (:obj:`int`, `optional`, defaults to 32):
vocab_size (`int`, *optional*, defaults to 32):
Vocabulary size of the WavLM model. Defines the number of different tokens that can be represented by the
:obj:`inputs_ids` passed when calling :class:`~transformers.WavLMModel`. Vocabulary size of the model.
Defines the different tokens that can be represented by the `inputs_ids` passed to the forward method of
:class:`~transformers.WavLMModel`.
hidden_size (:obj:`int`, `optional`, defaults to 768):
`inputs_ids` passed when calling [`WavLMModel`]. Vocabulary size of the model.
Defines the different tokens that can be represented by the *inputs_ids* passed to the forward method of
[`WavLMModel`].
hidden_size (`int`, *optional*, defaults to 768):
Dimensionality of the encoder layers and the pooler layer.
num_hidden_layers (:obj:`int`, `optional`, defaults to 12):
num_hidden_layers (`int`, *optional*, defaults to 12):
Number of hidden layers in the Transformer encoder.
num_attention_heads (:obj:`int`, `optional`, defaults to 12):
num_attention_heads (`int`, *optional*, defaults to 12):
Number of attention heads for each attention layer in the Transformer encoder.
intermediate_size (:obj:`int`, `optional`, defaults to 3072):
intermediate_size (`int`, *optional*, defaults to 3072):
Dimensionality of the "intermediate" (i.e., feed-forward) layer in the Transformer encoder.
hidden_act (:obj:`str` or :obj:`function`, `optional`, defaults to :obj:`"gelu"`):
hidden_act (`str` or `function`, *optional*, defaults to `"gelu"`):
The non-linear activation function (function or string) in the encoder and pooler. If string,
:obj:`"gelu"`, :obj:`"relu"`, :obj:`"selu"` and :obj:`"gelu_new"` are supported.
hidden_dropout (:obj:`float`, `optional`, defaults to 0.1):
`"gelu"`, `"relu"`, `"selu"` and `"gelu_new"` are supported.
hidden_dropout (`float`, *optional*, defaults to 0.1):
The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
attention_dropout (:obj:`float`, `optional`, defaults to 0.1):
attention_dropout (`float`, *optional*, defaults to 0.1):
The dropout ratio for the attention probabilities.
final_dropout (:obj:`float`, `optional`, defaults to 0.1):
The dropout probability for the final projection layer of :class:`WavLMForCTC`.
initializer_range (:obj:`float`, `optional`, defaults to 0.02):
final_dropout (`float`, *optional*, defaults to 0.1):
The dropout probability for the final projection layer of [`WavLMForCTC`].
initializer_range (`float`, *optional*, defaults to 0.02):
The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
layer_norm_eps (:obj:`float`, `optional`, defaults to 1e-12):
layer_norm_eps (`float`, *optional*, defaults to 1e-12):
The epsilon used by the layer normalization layers.
feat_extract_norm (:obj:`str`, `optional`, defaults to :obj:`"group"`):
The norm to be applied to 1D convolutional layers in feature extractor. One of :obj:`"group"` for group
normalization of only the first 1D convolutional layer or :obj:`"layer"` for layer normalization of all 1D
feat_extract_norm (`str`, *optional*, defaults to `"group"`):
The norm to be applied to 1D convolutional layers in feature extractor. One of `"group"` for group
normalization of only the first 1D convolutional layer or `"layer"` for layer normalization of all 1D
convolutional layers.
feat_proj_dropout (:obj:`float`, `optional`, defaults to 0.0):
feat_proj_dropout (`float`, *optional*, defaults to 0.0):
The dropout probability for output of the feature extractor.
feat_extract_activation (:obj:`str, `optional`, defaults to :obj:`"gelu"`):
feat_extract_activation (`str, `optional`, defaults to `"gelu"`):
The non-linear activation function (function or string) in the 1D convolutional layers of the feature
extractor. If string, :obj:`"gelu"`, :obj:`"relu"`, :obj:`"selu"` and :obj:`"gelu_new"` are supported.
feat_quantizer_dropout (obj:`float`, `optional`, defaults to 0.0):
extractor. If string, `"gelu"`, `"relu"`, `"selu"` and `"gelu_new"` are supported.
feat_quantizer_dropout (obj:*float*, *optional*, defaults to 0.0):
The dropout probabilitiy for quantized feature extractor states.
conv_dim (:obj:`Tuple[int]`, `optional`, defaults to :obj:`(512, 512, 512, 512, 512, 512, 512)`):
conv_dim (`Tuple[int]`, *optional*, defaults to `(512, 512, 512, 512, 512, 512, 512)`):
A tuple of integers defining the number of input and output channels of each 1D convolutional layer in the
feature extractor. The length of `conv_dim` defines the number of 1D convolutional layers.
conv_stride (:obj:`Tuple[int]`, `optional`, defaults to :obj:`(5, 2, 2, 2, 2, 2, 2)`):
feature extractor. The length of *conv_dim* defines the number of 1D convolutional layers.
conv_stride (`Tuple[int]`, *optional*, defaults to `(5, 2, 2, 2, 2, 2, 2)`):
A tuple of integers defining the stride of each 1D convolutional layer in the feature extractor. The length
of `conv_stride` defines the number of convolutional layers and has to match the the length of `conv_dim`.
conv_kernel (:obj:`Tuple[int]`, `optional`, defaults to :obj:`(10, 3, 3, 3, 3, 3, 3)`):
of *conv_stride* defines the number of convolutional layers and has to match the the length of *conv_dim*.
conv_kernel (`Tuple[int]`, *optional*, defaults to `(10, 3, 3, 3, 3, 3, 3)`):
A tuple of integers defining the kernel size of each 1D convolutional layer in the feature extractor. The
length of `conv_kernel` defines the number of convolutional layers and has to match the the length of
`conv_dim`.
conv_bias (:obj:`bool`, `optional`, defaults to :obj:`False`):
length of *conv_kernel* defines the number of convolutional layers and has to match the the length of
*conv_dim*.
conv_bias (`bool`, *optional*, defaults to `False`):
Whether the 1D convolutional layers have a bias.
num_conv_pos_embeddings (:obj:`int`, `optional`, defaults to 128):
num_conv_pos_embeddings (`int`, *optional*, defaults to 128):
Number of convolutional positional embeddings. Defines the kernel size of 1D convolutional positional
embeddings layer.
num_conv_pos_embedding_groups (:obj:`int`, `optional`, defaults to 16):
num_conv_pos_embedding_groups (`int`, *optional*, defaults to 16):
Number of groups of 1D convolutional positional embeddings layer.
do_stable_layer_norm (:obj:`bool`, `optional`, defaults to :obj:`False`):
Whether to apply `stable` layer norm architecture of the Transformer encoder. ``do_stable_layer_norm is
True`` corresponds to applying layer norm before the attention layer, whereas ``do_stable_layer_norm is
False`` corresponds to applying layer norm after the attention layer.
apply_spec_augment (:obj:`bool`, `optional`, defaults to :obj:`True`):
do_stable_layer_norm (`bool`, *optional*, defaults to `False`):
Whether to apply *stable* layer norm architecture of the Transformer encoder. `do_stable_layer_norm is True` corresponds to applying layer norm before the attention layer, whereas `do_stable_layer_norm is False` corresponds to applying layer norm after the attention layer.
apply_spec_augment (`bool`, *optional*, defaults to `True`):
Whether to apply *SpecAugment* data augmentation to the outputs of the feature extractor. For reference see
`SpecAugment: A Simple Data Augmentation Method for Automatic Speech Recognition
<https://arxiv.org/abs/1904.08779>`__.
mask_time_prob (:obj:`float`, `optional`, defaults to 0.05):
[SpecAugment: A Simple Data Augmentation Method for Automatic Speech Recognition](https://arxiv.org/abs/1904.08779).
mask_time_prob (`float`, *optional*, defaults to 0.05):
Propability of each feature vector along the time axis to be chosen as the start of the vector span to be
masked. Approximately ``mask_time_prob * sequence_length // mask_time_length`` feature vectors will be
masked along the time axis. This is only relevant if ``apply_spec_augment is True``.
mask_time_length (:obj:`int`, `optional`, defaults to 10):
masked. Approximately `mask_time_prob * sequence_length // mask_time_length` feature vectors will be
masked along the time axis. This is only relevant if `apply_spec_augment is True`.
mask_time_length (`int`, *optional*, defaults to 10):
Length of vector span along the time axis.
mask_time_min_masks (:obj:`int`, `optional`, defaults to 2),:
The minimum number of masks of length ``mask_feature_length`` generated along the time axis, each time
step, irrespectively of ``mask_feature_prob``. Only relevant if
mask_time_min_masks (`int`, *optional*, defaults to 2),:
The minimum number of masks of length `mask_feature_length` generated along the time axis, each time
step, irrespectively of `mask_feature_prob`. Only relevant if
''mask_time_prob*len(time_axis)/mask_time_length < mask_time_min_masks''
mask_feature_prob (:obj:`float`, `optional`, defaults to 0.0):
mask_feature_prob (`float`, *optional*, defaults to 0.0):
Propability of each feature vector along the feature axis to be chosen as the start of the vector span to
be masked. Approximately ``mask_time_prob * hidden_size // mask_time_length`` feature vectors will be
masked along the time axis. This is only relevant if ``apply_spec_augment is True``.
mask_feature_length (:obj:`int`, `optional`, defaults to 10):
be masked. Approximately `mask_time_prob * hidden_size // mask_time_length` feature vectors will be
masked along the time axis. This is only relevant if `apply_spec_augment is True`.
mask_feature_length (`int`, *optional*, defaults to 10):
Length of vector span along the feature axis.
num_codevectors_per_group (:obj:`int`, `optional`, defaults to 320):
num_codevectors_per_group (`int`, *optional*, defaults to 320):
Number of entries in each quantization codebook (group).
num_codevector_groups (:obj:`int`, `optional`, defaults to 2):
num_codevector_groups (`int`, *optional*, defaults to 2):
Number of codevector groups for product codevector quantization.
contrastive_logits_temperature (:obj:`float`, `optional`, defaults to 0.1):
The temperature `kappa` in the contrastive loss.
feat_quantizer_dropout (:obj:`float`, `optional`, defaults to 0.0):
contrastive_logits_temperature (`float`, *optional*, defaults to 0.1):
The temperature *kappa* in the contrastive loss.
feat_quantizer_dropout (`float`, *optional*, defaults to 0.0):
The dropout probabilitiy for the output of the feature extractor that's used by the quantizer.
num_negatives (:obj:`int`, `optional`, defaults to 100):
num_negatives (`int`, *optional*, defaults to 100):
Number of negative samples for the contrastive loss.
codevector_dim (:obj:`int`, `optional`, defaults to 256):
codevector_dim (`int`, *optional*, defaults to 256):
Dimensionality of the quantized feature vectors.
proj_codevector_dim (:obj:`int`, `optional`, defaults to 256):
proj_codevector_dim (`int`, *optional*, defaults to 256):
Dimensionality of the final projection of both the quantized and the transformer features.
diversity_loss_weight (:obj:`int`, `optional`, defaults to 0.1):
diversity_loss_weight (`int`, *optional*, defaults to 0.1):
The weight of the codebook diversity loss component.
ctc_loss_reduction (:obj:`str`, `optional`, defaults to :obj:`"mean"`):
Specifies the reduction to apply to the output of ``torch.nn.CTCLoss``. Only relevant when training an
instance of :class:`~transformers.WavLMForCTC`.
ctc_zero_infinity (:obj:`bool`, `optional`, defaults to :obj:`False`):
Whether to zero infinite losses and the associated gradients of ``torch.nn.CTCLoss``. Infinite losses
ctc_loss_reduction (`str`, *optional*, defaults to `"mean"`):
Specifies the reduction to apply to the output of `torch.nn.CTCLoss`. Only relevant when training an
instance of [`WavLMForCTC`].
ctc_zero_infinity (`bool`, *optional*, defaults to `False`):
Whether to zero infinite losses and the associated gradients of `torch.nn.CTCLoss`. Infinite losses
mainly occur when the inputs are too short to be aligned to the targets. Only relevant when training an
instance of :class:`~transformers.WavLMForCTC`.
use_weighted_layer_sum (:obj:`bool`, `optional`, defaults to :obj:`False`):
instance of [`WavLMForCTC`].
use_weighted_layer_sum (`bool`, *optional*, defaults to `False`):
Whether to use a weighted average of layer outputs with learned weights. Only relevant when using an
instance of :class:`~transformers.WavLMForSequenceClassification`.
classifier_proj_size (:obj:`int`, `optional`, defaults to 256):
instance of [`WavLMForSequenceClassification`].
classifier_proj_size (`int`, *optional*, defaults to 256):
Dimensionality of the projection before token mean-pooling for classification.
tdnn_dim (:obj:`Tuple[int]`, `optional`, defaults to :obj:`(512, 512, 512, 512, 1500)`):
A tuple of integers defining the number of output channels of each 1D convolutional layer in the `TDNN`
module of the `XVector` model. The length of `tdnn_dim` defines the number of `TDNN` layers.
tdnn_kernel (:obj:`Tuple[int]`, `optional`, defaults to :obj:`(5, 3, 3, 1, 1)`):
A tuple of integers defining the kernel size of each 1D convolutional layer in the `TDNN` module of the
`XVector` model. The length of `tdnn_kernel` has to match the length of `tdnn_dim`.
tdnn_dilation (:obj:`Tuple[int]`, `optional`, defaults to :obj:`(1, 2, 3, 1, 1)`):
A tuple of integers defining the dilation factor of each 1D convolutional layer in `TDNN` module of the
`XVector` model. The length of `tdnn_dilation` has to match the length of `tdnn_dim`.
xvector_output_dim (:obj:`int`, `optional`, defaults to 512):
Dimensionality of the `XVector` embedding vectors.
add_adapter (:obj:`bool`, `optional`, defaults to :obj:`False`):
tdnn_dim (`Tuple[int]`, *optional*, defaults to `(512, 512, 512, 512, 1500)`):
A tuple of integers defining the number of output channels of each 1D convolutional layer in the *TDNN*
module of the *XVector* model. The length of *tdnn_dim* defines the number of *TDNN* layers.
tdnn_kernel (`Tuple[int]`, *optional*, defaults to `(5, 3, 3, 1, 1)`):
A tuple of integers defining the kernel size of each 1D convolutional layer in the *TDNN* module of the
*XVector* model. The length of *tdnn_kernel* has to match the length of *tdnn_dim*.
tdnn_dilation (`Tuple[int]`, *optional*, defaults to `(1, 2, 3, 1, 1)`):
A tuple of integers defining the dilation factor of each 1D convolutional layer in *TDNN* module of the
*XVector* model. The length of *tdnn_dilation* has to match the length of *tdnn_dim*.
xvector_output_dim (`int`, *optional*, defaults to 512):
Dimensionality of the *XVector* embedding vectors.
add_adapter (`bool`, *optional*, defaults to `False`):
Whether a convolutional network should be stacked on top of the Wav2Vec2 Encoder. Can be very useful for
warm-starting Wav2Vec2 for SpeechEncoderDecoder models.
adapter_kernel_size (:obj:`int`, `optional`, defaults to 3):
Kernel size of the convolutional layers in the adapter network. Only relevant if ``add_adapter is True``.
adapter_stride (:obj:`int`, `optional`, defaults to 2):
Stride of the convolutional layers in the adapter network. Only relevant if ``add_adapter is True``.
num_adapter_layers (:obj:`int`, `optional`, defaults to 3):
Number of convolutional layers that should be used in the adapter network. Only relevant if ``add_adapter
is True``.
output_hidden_size (:obj:`int`, `optional`):
Dimensionality of the encoder output layer. If not defined, this defaults to `hidden-size`. Only relevant
if ``add_adapter is True``.
adapter_kernel_size (`int`, *optional*, defaults to 3):
Kernel size of the convolutional layers in the adapter network. Only relevant if `add_adapter is True`.
adapter_stride (`int`, *optional*, defaults to 2):
Stride of the convolutional layers in the adapter network. Only relevant if `add_adapter is True`.
num_adapter_layers (`int`, *optional*, defaults to 3):
Number of convolutional layers that should be used in the adapter network. Only relevant if `add_adapter is True`.
output_hidden_size (`int`, *optional*):
Dimensionality of the encoder output layer. If not defined, this defaults to *hidden-size*. Only relevant
if `add_adapter is True`.
Example::
Example:
Example::
```python
```
Example:
```python
>>> from transformers import WavLMModel, WavLMConfig
>>> # Initializing a WavLM facebook/wavlm-base-960h style configuration
......@@ -183,7 +183,7 @@ class WavLMConfig(PretrainedConfig):
>>> # Accessing the model configuration
>>> configuration = model.config
"""
```"""
model_type = "wavlm"
def __init__(
......
src/transformers/models/xlm/configuration_xlm.py
View file @ 27b3031d
......@@ -36,103 +36,104 @@ XLM_PRETRAINED_CONFIG_ARCHIVE_MAP = {
class XLMConfig(PretrainedConfig):
"""
This is the configuration class to store the configuration of a :class:`~transformers.XLMModel` or a
:class:`~transformers.TFXLMModel`. It is used to instantiate a XLM model according to the specified arguments,
This is the configuration class to store the configuration of a [`XLMModel`] or a
[`TFXLMModel`]. It is used to instantiate a XLM model according to the specified arguments,
defining the model architecture. Instantiating a configuration with the defaults will yield a similar configuration
to that of the `xlm-mlm-en-2048 <https://huggingface.co/xlm-mlm-en-2048>`__ architecture.
to that of the [xlm-mlm-en-2048](https://huggingface.co/xlm-mlm-en-2048) architecture.
Configuration objects inherit from :class:`~transformers.PretrainedConfig` and can be used to control the model
outputs. Read the documentation from :class:`~transformers.PretrainedConfig` for more information.
Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model
outputs. Read the documentation from [`PretrainedConfig`] for more information.
Args:
vocab_size (:obj:`int`, `optional`, defaults to 30145):
vocab_size (`int`, *optional*, defaults to 30145):
Vocabulary size of the BERT model. Defines the number of different tokens that can be represented by the
:obj:`inputs_ids` passed when calling :class:`~transformers.XLMModel` or :class:`~transformers.TFXLMModel`.
emb_dim (:obj:`int`, `optional`, defaults to 2048):
`inputs_ids` passed when calling [`XLMModel`] or [`TFXLMModel`].
emb_dim (`int`, *optional*, defaults to 2048):
Dimensionality of the encoder layers and the pooler layer.
n_layer (:obj:`int`, `optional`, defaults to 12):
n_layer (`int`, *optional*, defaults to 12):
Number of hidden layers in the Transformer encoder.
n_head (:obj:`int`, `optional`, defaults to 16):
n_head (`int`, *optional*, defaults to 16):
Number of attention heads for each attention layer in the Transformer encoder.
dropout (:obj:`float`, `optional`, defaults to 0.1):
dropout (`float`, *optional*, defaults to 0.1):
The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
attention_dropout (:obj:`float`, `optional`, defaults to 0.1):
attention_dropout (`float`, *optional*, defaults to 0.1):
The dropout probability for the attention mechanism
gelu_activation (:obj:`bool`, `optional`, defaults to :obj:`True`):
Whether or not to use `gelu` for the activations instead of `relu`.
sinusoidal_embeddings (:obj:`bool`, `optional`, defaults to :obj:`False`):
gelu_activation (`bool`, *optional*, defaults to `True`):
Whether or not to use *gelu* for the activations instead of *relu*.
sinusoidal_embeddings (`bool`, *optional*, defaults to `False`):
Whether or not to use sinusoidal positional embeddings instead of absolute positional embeddings.
causal (:obj:`bool`, `optional`, defaults to :obj:`False`):
causal (`bool`, *optional*, defaults to `False`):
Whether or not the model should behave in a causal manner. Causal models use a triangular attention mask in
order to only attend to the left-side context instead if a bidirectional context.
asm (:obj:`bool`, `optional`, defaults to :obj:`False`):
asm (`bool`, *optional*, defaults to `False`):
Whether or not to use an adaptive log softmax projection layer instead of a linear layer for the prediction
layer.
n_langs (:obj:`int`, `optional`, defaults to 1):
n_langs (`int`, *optional*, defaults to 1):
The number of languages the model handles. Set to 1 for monolingual models.
use_lang_emb (:obj:`bool`, `optional`, defaults to :obj:`True`)
Whether to use language embeddings. Some models use additional language embeddings, see `the multilingual
models page <http://huggingface.co/transformers/multilingual.html#xlm-language-embeddings>`__ for
use_lang_emb (`bool`, *optional*, defaults to `True`)
Whether to use language embeddings. Some models use additional language embeddings, see [the multilingual
models page](http://huggingface.co/transformers/multilingual.html#xlm-language-embeddings) for
information on how to use them.
max_position_embeddings (:obj:`int`, `optional`, defaults to 512):
max_position_embeddings (`int`, *optional*, defaults to 512):
The maximum sequence length that this model might ever be used with. Typically set this to something large
just in case (e.g., 512 or 1024 or 2048).
embed_init_std (:obj:`float`, `optional`, defaults to 2048^-0.5):
embed_init_std (`float`, *optional*, defaults to 2048^-0.5):
The standard deviation of the truncated_normal_initializer for initializing the embedding matrices.
init_std (:obj:`int`, `optional`, defaults to 50257):
init_std (`int`, *optional*, defaults to 50257):
The standard deviation of the truncated_normal_initializer for initializing all weight matrices except the
embedding matrices.
layer_norm_eps (:obj:`float`, `optional`, defaults to 1e-12):
layer_norm_eps (`float`, *optional*, defaults to 1e-12):
The epsilon used by the layer normalization layers.
bos_index (:obj:`int`, `optional`, defaults to 0):
bos_index (`int`, *optional*, defaults to 0):
The index of the beginning of sentence token in the vocabulary.
eos_index (:obj:`int`, `optional`, defaults to 1):
eos_index (`int`, *optional*, defaults to 1):
The index of the end of sentence token in the vocabulary.
pad_index (:obj:`int`, `optional`, defaults to 2):
pad_index (`int`, *optional*, defaults to 2):
The index of the padding token in the vocabulary.
unk_index (:obj:`int`, `optional`, defaults to 3):
unk_index (`int`, *optional*, defaults to 3):
The index of the unknown token in the vocabulary.
mask_index (:obj:`int`, `optional`, defaults to 5):
mask_index (`int`, *optional*, defaults to 5):
The index of the masking token in the vocabulary.
is_encoder(:obj:`bool`, `optional`, defaults to :obj:`True`):
is_encoder(`bool`, *optional*, defaults to `True`):
Whether or not the initialized model should be a transformer encoder or decoder as seen in Vaswani et al.
summary_type (:obj:`string`, `optional`, defaults to "first"):
summary_type (`string`, *optional*, defaults to "first"):
Argument used when doing sequence summary. Used in the sequence classification and multiple choice models.
Has to be one of the following options:
- :obj:`"last"`: Take the last token hidden state (like XLNet).
- :obj:`"first"`: Take the first token hidden state (like BERT).
- :obj:`"mean"`: Take the mean of all tokens hidden states.
- :obj:`"cls_index"`: Supply a Tensor of classification token position (like GPT/GPT-2).
- :obj:`"attn"`: Not implemented now, use multi-head attention.
summary_use_proj (:obj:`bool`, `optional`, defaults to :obj:`True`):
- `"last"`: Take the last token hidden state (like XLNet).
- `"first"`: Take the first token hidden state (like BERT).
- `"mean"`: Take the mean of all tokens hidden states.
- `"cls_index"`: Supply a Tensor of classification token position (like GPT/GPT-2).
- `"attn"`: Not implemented now, use multi-head attention.
summary_use_proj (`bool`, *optional*, defaults to `True`):
Argument used when doing sequence summary. Used in the sequence classification and multiple choice models.
Whether or not to add a projection after the vector extraction.
summary_activation (:obj:`str`, `optional`):
summary_activation (`str`, *optional*):
Argument used when doing sequence summary. Used in the sequence classification and multiple choice models.
Pass :obj:`"tanh"` for a tanh activation to the output, any other value will result in no activation.
summary_proj_to_labels (:obj:`bool`, `optional`, defaults to :obj:`True`):
Pass `"tanh"` for a tanh activation to the output, any other value will result in no activation.
summary_proj_to_labels (`bool`, *optional*, defaults to `True`):
Used in the sequence classification and multiple choice models.
Whether the projection outputs should have :obj:`config.num_labels` or :obj:`config.hidden_size` classes.
summary_first_dropout (:obj:`float`, `optional`, defaults to 0.1):
Whether the projection outputs should have `config.num_labels` or `config.hidden_size` classes.
summary_first_dropout (`float`, *optional*, defaults to 0.1):
Used in the sequence classification and multiple choice models.
The dropout ratio to be used after the projection and activation.
start_n_top (:obj:`int`, `optional`, defaults to 5):
start_n_top (`int`, *optional*, defaults to 5):
Used in the SQuAD evaluation script.
end_n_top (:obj:`int`, `optional`, defaults to 5):
end_n_top (`int`, *optional*, defaults to 5):
Used in the SQuAD evaluation script.
mask_token_id (:obj:`int`, `optional`, defaults to 0):
mask_token_id (`int`, *optional*, defaults to 0):
Model agnostic parameter to identify masked tokens when generating text in an MLM context.
lang_id (:obj:`int`, `optional`, defaults to 1):
lang_id (`int`, *optional*, defaults to 1):
The ID of the language used by the model. This parameter is used when generating text in a given language.
Examples::
Examples:
```python
>>> from transformers import XLMConfig, XLMModel
>>> # Initializing a XLM configuration
......@@ -143,7 +144,7 @@ class XLMConfig(PretrainedConfig):
>>> # Accessing the model configuration
>>> configuration = model.config
"""
```"""
model_type = "xlm"
attribute_map = {
......
src/transformers/models/xlm/tokenization_xlm.py
View file @ 27b3031d
......@@ -534,49 +534,52 @@ class XLMTokenizer(PreTrainedTokenizer):
- Moses preprocessing and tokenization for most supported languages.
- Language specific tokenization for Chinese (Jieba), Japanese (KyTea) and Thai (PyThaiNLP).
- Optionally lowercases and normalizes all inputs text.
- The arguments ``special_tokens`` and the function ``set_special_tokens``, can be used to add additional symbols
- The arguments `special_tokens` and the function `set_special_tokens`, can be used to add additional symbols
(like "__classify__") to a vocabulary.
- The :obj:`lang2id` attribute maps the languages supported by the model with their IDs if provided (automatically
- The `lang2id` attribute maps the languages supported by the model with their IDs if provided (automatically
set for pretrained vocabularies).
- The :obj:`id2lang` attributes does reverse mapping if provided (automatically set for pretrained vocabularies).
- The `id2lang` attributes does reverse mapping if provided (automatically set for pretrained vocabularies).
This tokenizer inherits from :class:`~transformers.PreTrainedTokenizer` which contains most of the main methods.
This tokenizer inherits from [`PreTrainedTokenizer`] which contains most of the main methods.
Users should refer to this superclass for more information regarding those methods.
Args:
vocab_file (:obj:`str`):
vocab_file (`str`):
Vocabulary file.
merges_file (:obj:`str`):
merges_file (`str`):
Merges file.
unk_token (:obj:`str`, `optional`, defaults to :obj:`"<unk>"`):
unk_token (`str`, *optional*, defaults to `"<unk>"`):
The unknown token. A token that is not in the vocabulary cannot be converted to an ID and is set to be this
token instead.
bos_token (:obj:`str`, `optional`, defaults to :obj:`"<s>"`):
bos_token (`str`, *optional*, defaults to `"<s>"`):
The beginning of sequence token that was used during pretraining. Can be used a sequence classifier token.
.. note::
<Tip>
When building a sequence using special tokens, this is not the token that is used for the beginning of
sequence. The token used is the :obj:`cls_token`.
sep_token (:obj:`str`, `optional`, defaults to :obj:`"</s>"`):
sequence. The token used is the `cls_token`.
</Tip>
sep_token (`str`, *optional*, defaults to `"</s>"`):
The separator token, which is used when building a sequence from multiple sequences, e.g. two sequences for
sequence classification or for a text and a question for question answering. It is also used as the last
token of a sequence built with special tokens.
pad_token (:obj:`str`, `optional`, defaults to :obj:`"<pad>"`):
pad_token (`str`, *optional*, defaults to `"<pad>"`):
The token used for padding, for example when batching sequences of different lengths.
cls_token (:obj:`str`, `optional`, defaults to :obj:`"</s>"`):
cls_token (`str`, *optional*, defaults to `"</s>"`):
The classifier token which is used when doing sequence classification (classification of the whole sequence
instead of per-token classification). It is the first token of the sequence when built with special tokens.
mask_token (:obj:`str`, `optional`, defaults to :obj:`"<special1>"`):
mask_token (`str`, *optional*, defaults to `"<special1>"`):
The token used for masking values. This is the token used when training this model with masked language
modeling. This is the token which the model will try to predict.
additional_special_tokens (:obj:`List[str]`, `optional`, defaults to :obj:`["<special0>","<special1>","<special2>","<special3>","<special4>","<special5>","<special6>","<special7>","<special8>","<special9>"]`):
additional_special_tokens (`List[str]`, *optional*, defaults to `["<special0>","<special1>","<special2>","<special3>","<special4>","<special5>","<special6>","<special7>","<special8>","<special9>"]`):
List of additional special tokens.
lang2id (:obj:`Dict[str, int]`, `optional`):
lang2id (`Dict[str, int]`, *optional*):
Dictionary mapping languages string identifiers to their IDs.
id2lang (:obj:`Dict[int, str]`, `optional`):
id2lang (`Dict[int, str]`, *optional*):
Dictionary mapping language IDs to their string identifiers.
do_lowercase_and_remove_accent (:obj:`bool`, `optional`, defaults to :obj:`True`):
do_lowercase_and_remove_accent (`bool`, *optional*, defaults to `True`):
Whether to lowercase and remove accents when tokenizing.
"""
......@@ -866,17 +869,17 @@ class XLMTokenizer(PreTrainedTokenizer):
Build model inputs from a sequence or a pair of sequence for sequence classification tasks by concatenating and
adding special tokens. An XLM sequence has the following format:
- single sequence: ``<s> X </s>``
- pair of sequences: ``<s> A </s> B </s>``
- single sequence: `<s> X </s>`
- pair of sequences: `<s> A </s> B </s>`
Args:
token_ids_0 (:obj:`List[int]`):
token_ids_0 (`List[int]`):
List of IDs to which the special tokens will be added.
token_ids_1 (:obj:`List[int]`, `optional`):
token_ids_1 (`List[int]`, *optional*):
Optional second list of IDs for sequence pairs.
Returns:
:obj:`List[int]`: List of `input IDs <../glossary.html#input-ids>`__ with the appropriate special tokens.
`List[int]`: List of [input IDs](../glossary#input-ids) with the appropriate special tokens.
"""
bos = [self.bos_token_id]
......@@ -891,18 +894,18 @@ class XLMTokenizer(PreTrainedTokenizer):
) -> List[int]:
"""
Retrieve sequence ids from a token list that has no special tokens added. This method is called when adding
special tokens using the tokenizer ``prepare_for_model`` method.
special tokens using the tokenizer `prepare_for_model` method.
Args:
token_ids_0 (:obj:`List[int]`):
token_ids_0 (`List[int]`):
List of IDs.
token_ids_1 (:obj:`List[int]`, `optional`):
token_ids_1 (`List[int]`, *optional*):
Optional second list of IDs for sequence pairs.
already_has_special_tokens (:obj:`bool`, `optional`, defaults to :obj:`False`):
already_has_special_tokens (`bool`, *optional*, defaults to `False`):
Whether or not the token list is already formatted with special tokens for the model.
Returns:
:obj:`List[int]`: A list of integers in the range [0, 1]: 1 for a special token, 0 for a sequence token.
`List[int]`: A list of integers in the range [0, 1]: 1 for a special token, 0 for a sequence token.
"""
if already_has_special_tokens:
......@@ -921,21 +924,21 @@ class XLMTokenizer(PreTrainedTokenizer):
Create a mask from the two sequences passed to be used in a sequence-pair classification task. An XLM sequence
pair mask has the following format:
::
```
0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1
| first sequence | second sequence |
```
If :obj:`token_ids_1` is :obj:`None`, this method only returns the first portion of the mask (0s).
If `token_ids_1` is `None`, this method only returns the first portion of the mask (0s).
Args:
token_ids_0 (:obj:`List[int]`):
token_ids_0 (`List[int]`):
List of IDs.
token_ids_1 (:obj:`List[int]`, `optional`):
token_ids_1 (`List[int]`, *optional*):
Optional second list of IDs for sequence pairs.
Returns:
:obj:`List[int]`: List of `token type IDs <../glossary.html#token-type-ids>`_ according to the given
`List[int]`: List of [token type IDs](../glossary#token-type-ids) according to the given
sequence(s).
"""
sep = [self.sep_token_id]
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment