Skip to content

GitLab

Unverified Commit 7af80f66 authored by Sylvain Gugger's avatar Sylvain Gugger Committed by GitHub
Browse files

Convert docstrings of modeling files (#14850) 

* Convert file_utils docstrings to Markdown

* Test on BERT

* Return block indent

* Temporarily disable doc styler

* Remove from quality checks as well

* Remove doc styler mess

* Remove check from circleCI

* Fix typo

* Convert file_utils docstrings to Markdown

* Test on BERT

* Return block indent

* Temporarily disable doc styler

* Remove from quality checks as well

* Remove doc styler mess

* Remove check from circleCI

* Fix typo

* Let's go on all other model files

* Add templates too

* Styling and quality
parent 2a337346
Hide whitespace changes
Inline Side-by-side
Showing with 1609 additions and 1669 deletions
src/transformers/models/vit/modeling_vit.py
View file @ 7af80f66
......@@ -437,39 +437,39 @@ class ViTPreTrainedModel(PreTrainedModel):
VIT_START_DOCSTRING = r"""
This model is a PyTorch `torch.nn.Module <https://pytorch.org/docs/stable/nn.html#torch.nn.Module>`_ subclass. Use
This model is a PyTorch [torch.nn.Module](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) subclass. Use
it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and
behavior.
Parameters:
config (:class:`~transformers.ViTConfig`): Model configuration class with all the parameters of the model.
config ([`ViTConfig`]): Model configuration class with all the parameters of the model.
Initializing with a config file does not load the weights associated with the model, only the
configuration. Check out the :meth:`~transformers.PreTrainedModel.from_pretrained` method to load the model
configuration. Check out the [`~PreTrainedModel.from_pretrained`] method to load the model
weights.
"""
VIT_INPUTS_DOCSTRING = r"""
Args:
pixel_values (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, num_channels, height, width)`):
Pixel values. Pixel values can be obtained using :class:`~transformers.ViTFeatureExtractor`. See
:meth:`transformers.ViTFeatureExtractor.__call__` for details.
pixel_values (`torch.FloatTensor` of shape `(batch_size, num_channels, height, width)`):
Pixel values. Pixel values can be obtained using [`ViTFeatureExtractor`]. See
[`ViTFeatureExtractor.__call__`] for details.
head_mask (:obj:`torch.FloatTensor` of shape :obj:`(num_heads,)` or :obj:`(num_layers, num_heads)`, `optional`):
Mask to nullify selected heads of the self-attention modules. Mask values selected in ``[0, 1]``:
head_mask (`torch.FloatTensor` of shape `(num_heads,)` or `(num_layers, num_heads)`, *optional*):
Mask to nullify selected heads of the self-attention modules. Mask values selected in `[0, 1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
output_attentions (:obj:`bool`, `optional`):
Whether or not to return the attentions tensors of all attention layers. See ``attentions`` under returned
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under returned
tensors for more detail.
output_hidden_states (:obj:`bool`, `optional`):
Whether or not to return the hidden states of all layers. See ``hidden_states`` under returned tensors for
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for
more detail.
interpolate_pos_encoding (:obj:`bool`, `optional`):
interpolate_pos_encoding (`bool`, *optional*):
Whether to interpolate the pre-trained position encodings.
return_dict (:obj:`bool`, `optional`):
Whether or not to return a :class:`~transformers.file_utils.ModelOutput` instead of a plain tuple.
return_dict (`bool`, *optional*):
Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple.
"""
......@@ -621,32 +621,32 @@ class ViTForImageClassification(ViTPreTrainedModel):
return_dict=None,
):
r"""
labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`):
Labels for computing the image classification/regression loss. Indices should be in :obj:`[0, ...,
config.num_labels - 1]`. If :obj:`config.num_labels == 1` a regression loss is computed (Mean-Square loss),
If :obj:`config.num_labels > 1` a classification loss is computed (Cross-Entropy).
labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
Labels for computing the image classification/regression loss. Indices should be in `[0, ..., config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss),
If `config.num_labels > 1` a classification loss is computed (Cross-Entropy).
Returns:
Examples::
Examples:
>>> from transformers import ViTFeatureExtractor, ViTForImageClassification
>>> from PIL import Image
>>> import requests
```python
>>> from transformers import ViTFeatureExtractor, ViTForImageClassification
>>> from PIL import Image
>>> import requests
>>> url = 'http://images.cocodataset.org/val2017/000000039769.jpg'
>>> image = Image.open(requests.get(url, stream=True).raw)
>>> url = 'http://images.cocodataset.org/val2017/000000039769.jpg'
>>> image = Image.open(requests.get(url, stream=True).raw)
>>> feature_extractor = ViTFeatureExtractor.from_pretrained('google/vit-base-patch16-224')
>>> model = ViTForImageClassification.from_pretrained('google/vit-base-patch16-224')
>>> feature_extractor = ViTFeatureExtractor.from_pretrained('google/vit-base-patch16-224')
>>> model = ViTForImageClassification.from_pretrained('google/vit-base-patch16-224')
>>> inputs = feature_extractor(images=image, return_tensors="pt")
>>> outputs = model(**inputs)
>>> logits = outputs.logits
>>> # model predicts one of the 1000 ImageNet classes
>>> predicted_class_idx = logits.argmax(-1).item()
>>> print("Predicted class:", model.config.id2label[predicted_class_idx])
"""
>>> inputs = feature_extractor(images=image, return_tensors="pt")
>>> outputs = model(**inputs)
>>> logits = outputs.logits
>>> # model predicts one of the 1000 ImageNet classes
>>> predicted_class_idx = logits.argmax(-1).item()
>>> print("Predicted class:", model.config.id2label[predicted_class_idx])
```"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
outputs = self.vit(
......
src/transformers/models/wav2vec2/modeling_flax_wav2vec2.py
View file @ 7af80f66
......@@ -45,22 +45,21 @@ logger = logging.get_logger(__name__)
@flax.struct.dataclass
class FlaxWav2Vec2BaseModelOutput(ModelOutput):
"""
Output type of :class:`~transformers.FlaxWav2Vec2BaseModelOutput`, with potential hidden states and attentions.
Output type of [`FlaxWav2Vec2BaseModelOutput`], with potential hidden states and attentions.
Args:
last_hidden_state (:obj:`jnp.ndarray` of shape :obj:`(batch_size, sequence_length, hidden_size)`):
last_hidden_state (`jnp.ndarray` of shape `(batch_size, sequence_length, hidden_size)`):
Sequence of hidden-states at the output of the last layer of the model.
extract_features (:obj:`jnp.ndarray` of shape :obj:`(batch_size, sequence_length, last_conv_dim)`):
Sequence of extracted feature vectors of the last convolutional layer of the model with ``last_conv_dim``
extract_features (`jnp.ndarray` of shape `(batch_size, sequence_length, last_conv_dim)`):
Sequence of extracted feature vectors of the last convolutional layer of the model with `last_conv_dim`
being the dimension of the last convolutional layer.
hidden_states (:obj:`tuple(jnp.ndarray)`, `optional`, returned when ``output_hidden_states=True`` is passed or when ``config.output_hidden_states=True``):
Tuple of :obj:`jnp.ndarray` (one for the output of the embeddings + one for the output of each layer) of
shape :obj:`(batch_size, sequence_length, hidden_size)`.
hidden_states (`tuple(jnp.ndarray)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `jnp.ndarray` (one for the output of the embeddings + one for the output of each layer) of
shape `(batch_size, sequence_length, hidden_size)`.
Hidden-states of the model at the output of each layer plus the initial embedding outputs.
attentions (:obj:`tuple(jnp.ndarray)`, `optional`, returned when ``output_attentions=True`` is passed or when ``config.output_attentions=True``):
Tuple of :obj:`jnp.ndarray` (one for each layer) of shape :obj:`(batch_size, num_heads, sequence_length,
sequence_length)`.
attentions (`tuple(jnp.ndarray)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `jnp.ndarray` (one for each layer) of shape `(batch_size, num_heads, sequence_length, sequence_length)`.
Attentions weights after the attention softmax, used to compute the weighted average in the self-attention
heads.
......@@ -75,27 +74,26 @@ class FlaxWav2Vec2BaseModelOutput(ModelOutput):
@flax.struct.dataclass
class FlaxWav2Vec2ForPreTrainingOutput(ModelOutput):
"""
Output type of :class:`~transformers.FlaxWav2Vec2ForPreTrainingOutput`, with potential hidden states and
Output type of [`FlaxWav2Vec2ForPreTrainingOutput`], with potential hidden states and
attentions.
Args:
loss (`optional`, returned when model is in train mode, ``jnp.ndarray`` of shape :obj:`(1,)`):
Total loss as the sum of the contrastive loss (L_m) and the diversity loss (L_d) as stated in the `official
paper <https://arxiv.org/pdf/2006.11477.pdf>`__ . (classification) loss.
projected_states (:obj:`jnp.ndarray` of shape :obj:`(batch_size, sequence_length, config.proj_codevector_dim)`):
Hidden-states of the model projected to `config.proj_codevector_dim` that can be used to predict the masked
loss (*optional*, returned when model is in train mode, `jnp.ndarray` of shape `(1,)`):
Total loss as the sum of the contrastive loss (L_m) and the diversity loss (L_d) as stated in the [official
paper](https://arxiv.org/pdf/2006.11477.pdf) . (classification) loss.
projected_states (`jnp.ndarray` of shape `(batch_size, sequence_length, config.proj_codevector_dim)`):
Hidden-states of the model projected to *config.proj_codevector_dim* that can be used to predict the masked
projected quantized states.
projected_quantized_states (:obj:`jnp.ndarray` of shape :obj:`(batch_size, sequence_length, config.proj_codevector_dim)`):
Quantized extracted feature vectors projected to `config.proj_codevector_dim` representing the positive
projected_quantized_states (`jnp.ndarray` of shape `(batch_size, sequence_length, config.proj_codevector_dim)`):
Quantized extracted feature vectors projected to *config.proj_codevector_dim* representing the positive
target vectors for contrastive loss.
hidden_states (:obj:`tuple(jnp.ndarray)`, `optional`, returned when ``output_hidden_states=True`` is passed or when ``config.output_hidden_states=True``):
Tuple of :obj:`jnp.ndarray` (one for the output of the embeddings + one for the output of each layer) of
shape :obj:`(batch_size, sequence_length, hidden_size)`.
hidden_states (`tuple(jnp.ndarray)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `jnp.ndarray` (one for the output of the embeddings + one for the output of each layer) of
shape `(batch_size, sequence_length, hidden_size)`.
Hidden-states of the model at the output of each layer plus the initial embedding outputs.
attentions (:obj:`tuple(jnp.ndarray)`, `optional`, returned when ``output_attentions=True`` is passed or when ``config.output_attentions=True``):
Tuple of :obj:`jnp.ndarray` (one for each layer) of shape :obj:`(batch_size, num_heads, sequence_length,
sequence_length)`.
attentions (`tuple(jnp.ndarray)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `jnp.ndarray` (one for each layer) of shape `(batch_size, num_heads, sequence_length, sequence_length)`.
Attentions weights after the attention softmax, used to compute the weighted average in the self-attention
heads.
......@@ -213,77 +211,73 @@ def _sample_negative_indices(features_shape: Tuple, num_negatives: int, attentio
WAV_2_VEC_2_START_DOCSTRING = r"""
Wav2Vec2 was proposed in `wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations
<https://arxiv.org/abs/2006.11477>`__ by Alexei Baevski, Henry Zhou, Abdelrahman Mohamed, Michael Auli.
Wav2Vec2 was proposed in [wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations](https://arxiv.org/abs/2006.11477) by Alexei Baevski, Henry Zhou, Abdelrahman Mohamed, Michael Auli.
This model inherits from :class:`~transformers.FlaxPreTrainedModel`. Check the superclass documentation for the
This model inherits from [`FlaxPreTrainedModel`]. Check the superclass documentation for the
generic methods the library implements for all its model (such as downloading or saving, resizing the input
embeddings, pruning heads etc.)
This model is also a Flax Linen `flax.nn.Module
<https://flax.readthedocs.io/en/latest/_autosummary/flax.nn.module.html>`__ subclass. Use it as a regular Flax
This model is also a Flax Linen [flax.nn.Module](https://flax.readthedocs.io/en/latest/_autosummary/flax.nn.module.html) subclass. Use it as a regular Flax
Module and refer to the Flax documentation for all matter related to general usage and behavior.
Finally, this model supports inherent JAX features such as:
- `Just-In-Time (JIT) compilation <https://jax.readthedocs.io/en/latest/jax.html#just-in-time-compilation-jit>`__
- `Automatic Differentiation <https://jax.readthedocs.io/en/latest/jax.html#automatic-differentiation>`__
- `Vectorization <https://jax.readthedocs.io/en/latest/jax.html#vectorization-vmap>`__
- `Parallelization <https://jax.readthedocs.io/en/latest/jax.html#parallelization-pmap>`__
- [Just-In-Time (JIT) compilation](https://jax.readthedocs.io/en/latest/jax.html#just-in-time-compilation-jit)
- [Automatic Differentiation](https://jax.readthedocs.io/en/latest/jax.html#automatic-differentiation)
- [Vectorization](https://jax.readthedocs.io/en/latest/jax.html#vectorization-vmap)
- [Parallelization](https://jax.readthedocs.io/en/latest/jax.html#parallelization-pmap)
Parameters:
config (:class:`~transformers.Wav2Vec2Config`): Model configuration class with all the parameters of the model.
config ([`Wav2Vec2Config`]): Model configuration class with all the parameters of the model.
Initializing with a config file does not load the weights associated with the model, only the
configuration. Check out the :meth:`~transformers.FlaxPreTrainedModel.from_pretrained` method to load the
configuration. Check out the [`~FlaxPreTrainedModel.from_pretrained`] method to load the
model weights.
dtype (:obj:`jax.numpy.dtype`, `optional`, defaults to :obj:`jax.numpy.float32`):
The data type of the computation. Can be one of :obj:`jax.numpy.float32`, :obj:`jax.numpy.float16` (on
GPUs) and :obj:`jax.numpy.bfloat16` (on TPUs).
dtype (`jax.numpy.dtype`, *optional*, defaults to `jax.numpy.float32`):
The data type of the computation. Can be one of `jax.numpy.float32`, `jax.numpy.float16` (on
GPUs) and `jax.numpy.bfloat16` (on TPUs).
This can be used to enable mixed-precision training or half-precision inference on GPUs or TPUs. If
specified all the computation will be performed with the given ``dtype``.
specified all the computation will be performed with the given `dtype`.
**Note that this only specifies the dtype of the computation and does not influence the dtype of model
parameters.**
If you wish to change the dtype of the model parameters, see
:meth:`~transformers.FlaxPreTrainedModel.to_fp16` and :meth:`~transformers.FlaxPreTrainedModel.to_bf16`.
[`~FlaxPreTrainedModel.to_fp16`] and [`~FlaxPreTrainedModel.to_bf16`].
"""
WAV_2_VEC_2_INPUTS_DOCSTRING = r"""
Args:
input_values (:obj:`jnp.ndarray` of shape :obj:`(batch_size, sequence_length)`):
Float values of input raw speech waveform. Values can be obtained by loading a `.flac` or `.wav` audio file
into an array of type `List[float]` or a `numpy.ndarray`, *e.g.* via the soundfile library (`pip install
soundfile`). To prepare the array into `input_values`, the :class:`~transformers.Wav2Vec2Processor` should
be used for padding and conversion into a tensor of type `jnp.ndarray`. See
:meth:`transformers.Wav2Vec2Processor.__call__` for details.
attention_mask (:obj:`jnp.ndarray` of shape :obj:`(batch_size, sequence_length)`, `optional`):
Mask to avoid performing convolution and attention on padding token indices. Mask values selected in ``[0,
1]``:
input_values (`jnp.ndarray` of shape `(batch_size, sequence_length)`):
Float values of input raw speech waveform. Values can be obtained by loading a *.flac* or *.wav* audio file
into an array of type *List[float]* or a *numpy.ndarray*, *e.g.* via the soundfile library (*pip install
soundfile*). To prepare the array into *input_values*, the [`Wav2Vec2Processor`] should
be used for padding and conversion into a tensor of type *jnp.ndarray*. See
[`Wav2Vec2Processor.__call__`] for details.
attention_mask (`jnp.ndarray` of shape `(batch_size, sequence_length)`, *optional*):
Mask to avoid performing convolution and attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
`What are attention masks? <../glossary.html#attention-mask>`__ .. warning:: :obj:`attention_mask` should
only be passed if the corresponding processor has ``config.return_attention_mask == True``. For all models
whose processor has ``config.return_attention_mask == False``, such as `wav2vec2-base
<https://huggingface.co/facebook/wav2vec2-base-960h>`__, :obj:`attention_mask` should **not** be passed to
avoid degraded performance when doing batched inference. For such models :obj:`input_values` should simply
be padded with 0 and passed without :obj:`attention_mask`. Be aware that these models also yield slightly
different results depending on whether :obj:`input_values` is padded or not.
mask_time_indices (:obj:`jnp.ndarray` of shape :obj:`(batch_size, sequence_length)`, `optional`):
[What are attention masks?](../glossary#attention-mask) .. warning:: `attention_mask` should
only be passed if the corresponding processor has `config.return_attention_mask == True`. For all models
whose processor has `config.return_attention_mask == False`, such as [wav2vec2-base](https://huggingface.co/facebook/wav2vec2-base-960h), `attention_mask` should **not** be passed to
avoid degraded performance when doing batched inference. For such models `input_values` should simply
be padded with 0 and passed without `attention_mask`. Be aware that these models also yield slightly
different results depending on whether `input_values` is padded or not.
mask_time_indices (`jnp.ndarray` of shape `(batch_size, sequence_length)`, *optional*):
Indices to mask extracted features for contrastive loss. When in training mode, model learns to predict
masked extracted features in `config.proj_codevector_dim` space.
output_attentions (:obj:`bool`, `optional`):
Whether or not to return the attentions tensors of all attention layers. See ``attentions`` under returned
masked extracted features in *config.proj_codevector_dim* space.
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under returned
tensors for more detail.
output_hidden_states (:obj:`bool`, `optional`):
Whether or not to return the hidden states of all layers. See ``hidden_states`` under returned tensors for
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for
more detail.
return_dict (:obj:`bool`, `optional`):
Whether or not to return a :class:`~transformers.file_utils.ModelOutput` instead of a plain tuple.
return_dict (`bool`, *optional*):
Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple.
"""
......
src/transformers/models/wav2vec2/modeling_tf_wav2vec2.py
View file @ 7af80f66
......@@ -61,14 +61,13 @@ LARGE_NEGATIVE = -1e8
def input_values_processing(func, config, input_values, **kwargs):
"""
Process the input of each TensorFlow model including the booleans. In case of a list of symbolic inputs, each input
has to be named accordingly to the parameters name, i.e. :obj:`input_values = tf.keras.Input(shape=(128,),
dtype='float32', name="input_values")` otherwise the order of the tensors will not be guaranteed during the
has to be named accordingly to the parameters name, i.e. `input_values = tf.keras.Input(shape=(128,), dtype='float32', name="input_values")` otherwise the order of the tensors will not be guaranteed during the
training.
Args:
func (:obj:`callable`):
func (`callable`):
The callable function of the TensorFlow model.
config (:class:`~transformers.PretrainedConfig`):
config ([`PretrainedConfig`]):
The config of the running model.
**kwargs:
The inputs of the model.
......@@ -1277,92 +1276,92 @@ class TFWav2Vec2PreTrainedModel(TFPreTrainedModel):
WAV_2_VEC_2_START_DOCSTRING = r"""
This model inherits from :class:`~transformers.TFPreTrainedModel`. Check the superclass documentation for the
This model inherits from [`TFPreTrainedModel`]. Check the superclass documentation for the
generic methods the library implements for all its model (such as downloading or saving, resizing the input
embeddings, pruning heads etc.)
This model is also a `tf.keras.Model <https://www.tensorflow.org/api_docs/python/tf/keras/Model>`__ subclass. Use
This model is also a [tf.keras.Model](https://www.tensorflow.org/api_docs/python/tf/keras/Model) subclass. Use
it as a regular TF 2.0 Keras Model and refer to the TF 2.0 documentation for all matter related to general usage
and behavior.
.. note::
<Tip>
TF 2.0 models accepts two formats as inputs:
TF 2.0 models accepts two formats as inputs:
- having all inputs as keyword arguments (like PyTorch models), or
- having all inputs as a list, tuple or dict in the first positional arguments.
- having all inputs as keyword arguments (like PyTorch models), or
- having all inputs as a list, tuple or dict in the first positional arguments.
This second option is useful when using :meth:`tf.keras.Model.fit` method which currently requires having all
the tensors in the first argument of the model call function: :obj:`model(inputs)`.
This second option is useful when using [`tf.keras.Model.fit`] method which currently requires having all
the tensors in the first argument of the model call function: `model(inputs)`.
If you choose this second option, there are three possibilities you can use to gather all the input Tensors in
the first positional argument :
If you choose this second option, there are three possibilities you can use to gather all the input Tensors in
the first positional argument :
- a single Tensor with :obj:`input_values` only and nothing else: :obj:`model(inputs_ids)`
- a list of varying length with one or several input Tensors IN THE ORDER given in the docstring:
:obj:`model([input_values, attention_mask])` or :obj:`model([input_values, attention_mask, token_type_ids])`
- a dictionary with one or several input Tensors associated to the input names given in the docstring:
:obj:`model({"input_values": input_values, "token_type_ids": token_type_ids})`
- a single Tensor with `input_values` only and nothing else: `model(inputs_ids)`
- a list of varying length with one or several input Tensors IN THE ORDER given in the docstring:
`model([input_values, attention_mask])` or `model([input_values, attention_mask, token_type_ids])`
- a dictionary with one or several input Tensors associated to the input names given in the docstring:
`model({"input_values": input_values, "token_type_ids": token_type_ids})`
</Tip>
Args:
config (:class:`~transformers.Wav2Vec2Config`): Model configuration class with all the parameters of the model.
config ([`Wav2Vec2Config`]): Model configuration class with all the parameters of the model.
Initializing with a config file does not load the weights associated with the model, only the
configuration. Check out the :meth:`~transformers.PreTrainedModel.from_pretrained` method to load the model
configuration. Check out the [`~PreTrainedModel.from_pretrained`] method to load the model
weights.
"""
WAV_2_VEC_2_INPUTS_DOCSTRING = r"""
Args:
input_values (:obj:`np.ndarray`, :obj:`tf.Tensor`, :obj:`List[tf.Tensor]` :obj:`Dict[str, tf.Tensor]` or :obj:`Dict[str, np.ndarray]` and each example must have the shape :obj:`({0})`):
input_values (`np.ndarray`, `tf.Tensor`, `List[tf.Tensor]` ``Dict[str, tf.Tensor]` or `Dict[str, np.ndarray]` and each example must have the shape `({0})`):
Indices of input sequence tokens in the vocabulary.
Indices can be obtained using :class:`~transformers.BertTokenizer`. See
:func:`transformers.PreTrainedTokenizer.__call__` and :func:`transformers.PreTrainedTokenizer.encode` for
Indices can be obtained using [`BertTokenizer`]. See
[`PreTrainedTokenizer.__call__`] and [`PreTrainedTokenizer.encode`] for
details.
`What are input IDs? <../glossary.html#input-ids>`__
attention_mask (:obj:`np.ndarray` or :obj:`tf.Tensor` of shape :obj:`({0})`, `optional`):
Mask to avoid performing attention on padding token indices. Mask values selected in ``[0, 1]``:
[What are input IDs?](../glossary#input-ids)
attention_mask (`np.ndarray` or `tf.Tensor` of shape `({0})`, *optional*):
Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
`What are attention masks? <../glossary.html#attention-mask>`__
token_type_ids (:obj:`np.ndarray` or :obj:`tf.Tensor` of shape :obj:`({0})`, `optional`):
Segment token indices to indicate first and second portions of the inputs. Indices are selected in ``[0,
1]``:
[What are attention masks?](../glossary#attention-mask)
token_type_ids (`np.ndarray` or `tf.Tensor` of shape `({0})`, *optional*):
Segment token indices to indicate first and second portions of the inputs. Indices are selected in `[0, 1]`:
- 0 corresponds to a `sentence A` token,
- 1 corresponds to a `sentence B` token.
- 0 corresponds to a *sentence A* token,
- 1 corresponds to a *sentence B* token.
`What are token type IDs? <../glossary.html#token-type-ids>`__
position_ids (:obj:`np.ndarray` or :obj:`tf.Tensor` of shape :obj:`({0})`, `optional`):
Indices of positions of each input sequence tokens in the position embeddings. Selected in the range ``[0,
config.max_position_embeddings - 1]``.
[What are token type IDs?](../glossary#token-type-ids)
position_ids (`np.ndarray` or `tf.Tensor` of shape `({0})`, *optional*):
Indices of positions of each input sequence tokens in the position embeddings. Selected in the range `[0, config.max_position_embeddings - 1]`.
`What are position IDs? <../glossary.html#position-ids>`__
head_mask (:obj:`np.ndarray` or :obj:`tf.Tensor` of shape :obj:`(num_heads,)` or :obj:`(num_layers, num_heads)`, `optional`):
Mask to nullify selected heads of the self-attention modules. Mask values selected in ``[0, 1]``:
[What are position IDs?](../glossary#position-ids)
head_mask (`np.ndarray` or `tf.Tensor` of shape `(num_heads,)` or `(num_layers, num_heads)`, *optional*):
Mask to nullify selected heads of the self-attention modules. Mask values selected in `[0, 1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
inputs_embeds (:obj:`np.ndarray` or :obj:`tf.Tensor` of shape :obj:`({0}, hidden_size)`, `optional`):
Optionally, instead of passing :obj:`input_values` you can choose to directly pass an embedded
representation. This is useful if you want more control over how to convert :obj:`input_values` indices
inputs_embeds (`np.ndarray` or `tf.Tensor` of shape `({0}, hidden_size)`, *optional*):
Optionally, instead of passing `input_values` you can choose to directly pass an embedded
representation. This is useful if you want more control over how to convert `input_values` indices
into associated vectors than the model's internal embedding lookup matrix.
output_attentions (:obj:`bool`, `optional`):
Whether or not to return the attentions tensors of all attention layers. See ``attentions`` under returned
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under returned
tensors for more detail. This argument can be used only in eager mode, in graph mode the value in the
config will be used instead.
output_hidden_states (:obj:`bool`, `optional`):
Whether or not to return the hidden states of all layers. See ``hidden_states`` under returned tensors for
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for
more detail. This argument can be used only in eager mode, in graph mode the value in the config will be
used instead.
return_dict (:obj:`bool`, `optional`):
Whether or not to return a :class:`~transformers.file_utils.ModelOutput` instead of a plain tuple. This
return_dict (`bool`, *optional*):
Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple. This
argument can be used in eager mode, in graph mode the value will always be set to True.
training (:obj:`bool`, `optional`, defaults to :obj:`False`):
training (`bool`, *optional*, defaults to `False``):
Whether or not to use the model in training mode (some modules like dropout modules have different
behaviors between training and evaluation).
"""
......@@ -1499,46 +1498,46 @@ class TFWav2Vec2ForCTC(TFWav2Vec2PreTrainedModel):
training: Optional[bool] = False,
) -> Union[TFCausalLMOutput, Tuple[tf.Tensor]]:
r"""
labels (:obj:`tf.Tensor` or :obj:`np.ndarray` of shape :obj:`(batch_size, sequence_length)`, `optional`):
Labels for computing the masked language modeling loss. Indices should be in ``[-100, 0, ...,
config.vocab_size]`` (see ``input_values`` docstring) Tokens with indices set to ``-100`` are ignored
(masked), the loss is only computed for the tokens with labels in ``[0, ..., config.vocab_size]``
labels (`tf.Tensor` or `np.ndarray` of shape `(batch_size, sequence_length)`, *optional*):
Labels for computing the masked language modeling loss. Indices should be in `[-100, 0, ..., config.vocab_size]` (see `input_values` docstring) Tokens with indices set to `-100` are ignored
(masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]`
Returns:
Example::
Example:
>>> import tensorflow as tf
>>> from transformers import Wav2Vec2Processor, TFWav2Vec2ForCTC
>>> from datasets import load_dataset
>>> import soundfile as sf
```python
>>> import tensorflow as tf
>>> from transformers import Wav2Vec2Processor, TFWav2Vec2ForCTC
>>> from datasets import load_dataset
>>> import soundfile as sf
>>> processor = Wav2Vec2Processor.from_pretrained("facebook/wav2vec2-base-960h")
>>> model = TFWav2Vec2ForCTC.from_pretrained("facebook/wav2vec2-base-960h")
>>> processor = Wav2Vec2Processor.from_pretrained("facebook/wav2vec2-base-960h")
>>> model = TFWav2Vec2ForCTC.from_pretrained("facebook/wav2vec2-base-960h")
>>> def map_to_array(batch):
>>> speech, _ = sf.read(batch["file"])
>>> batch["speech"] = speech
>>> return batch
>>> def map_to_array(batch):
>>> speech, _ = sf.read(batch["file"])
>>> batch["speech"] = speech
>>> return batch
>>> ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
>>> ds = ds.map(map_to_array)
>>> ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
>>> ds = ds.map(map_to_array)
>>> input_values = processor(ds["speech"][0], return_tensors="tf").input_values # Batch size 1
>>> logits = model(input_values).logits
>>> predicted_ids = tf.argmax(logits, axis=-1)
>>> input_values = processor(ds["speech"][0], return_tensors="tf").input_values # Batch size 1
>>> logits = model(input_values).logits
>>> predicted_ids = tf.argmax(logits, axis=-1)
>>> transcription = processor.decode(predicted_ids[0])
>>> transcription = processor.decode(predicted_ids[0])
>>> # compute loss
>>> target_transcription = "A MAN SAID TO THE UNIVERSE SIR I EXIST"
>>> # compute loss
>>> target_transcription = "A MAN SAID TO THE UNIVERSE SIR I EXIST"
>>> # wrap processor as target processor to encode labels
>>> with processor.as_target_processor():
>>> labels = processor(transcription, return_tensors="tf").input_ids
>>> # wrap processor as target processor to encode labels
>>> with processor.as_target_processor():
>>> labels = processor(transcription, return_tensors="tf").input_ids
>>> loss = model(input_values, labels=labels).loss
"""
>>> loss = model(input_values, labels=labels).loss
```"""
inputs = input_values_processing(
func=self.call,
config=self.config,
......
src/transformers/models/wav2vec2/modeling_wav2vec2.py
View file @ 7af80f66
......@@ -72,21 +72,20 @@ WAV_2_VEC_2_PRETRAINED_MODEL_ARCHIVE_LIST = [
@dataclass
class Wav2Vec2BaseModelOutput(ModelOutput):
"""
Output type of :class:`~transformers.Wav2Vec2BaseModelOutput`, with potential hidden states and attentions.
Output type of [`Wav2Vec2BaseModelOutput`], with potential hidden states and attentions.
Args:
last_hidden_state (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length, hidden_size)`):
last_hidden_state (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`):
Sequence of hidden-states at the output of the last layer of the model.
extract_features (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length, conv_dim[-1])`):
extract_features (`torch.FloatTensor` of shape `(batch_size, sequence_length, conv_dim[-1])`):
Sequence of extracted feature vectors of the last convolutional layer of the model.
hidden_states (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_hidden_states=True`` is passed or when ``config.output_hidden_states=True``):
Tuple of :obj:`torch.FloatTensor` (one for the output of the embeddings + one for the output of each layer)
of shape :obj:`(batch_size, sequence_length, hidden_size)`.
hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `torch.FloatTensor` (one for the output of the embeddings + one for the output of each layer)
of shape `(batch_size, sequence_length, hidden_size)`.
Hidden-states of the model at the output of each layer plus the initial embedding outputs.
attentions (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_attentions=True`` is passed or when ``config.output_attentions=True``):
Tuple of :obj:`torch.FloatTensor` (one for each layer) of shape :obj:`(batch_size, num_heads,
sequence_length, sequence_length)`.
attentions (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length, sequence_length)`.
Attentions weights after the attention softmax, used to compute the weighted average in the self-attention
heads.
......@@ -101,33 +100,32 @@ class Wav2Vec2BaseModelOutput(ModelOutput):
@dataclass
class Wav2Vec2ForPreTrainingOutput(ModelOutput):
"""
Output type of :class:`~transformers.Wav2Vec2ForPreTraining`, with potential hidden states and attentions.
Output type of [`Wav2Vec2ForPreTraining`], with potential hidden states and attentions.
Args:
loss (`optional`, returned when :obj:`sample_negative_indices` are passed, ``torch.FloatTensor`` of shape :obj:`(1,)`):
Total loss as the sum of the contrastive loss (L_m) and the diversity loss (L_d) as stated in the `official
paper <https://arxiv.org/pdf/2006.11477.pdf>`__ . (classification) loss.
projected_states (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length, config.proj_codevector_dim)`):
Hidden-states of the model projected to `config.proj_codevector_dim` that can be used to predict the masked
loss (*optional*, returned when `sample_negative_indices` are passed, `torch.FloatTensor` of shape `(1,)`):
Total loss as the sum of the contrastive loss (L_m) and the diversity loss (L_d) as stated in the [official
paper](https://arxiv.org/pdf/2006.11477.pdf) . (classification) loss.
projected_states (`torch.FloatTensor` of shape `(batch_size, sequence_length, config.proj_codevector_dim)`):
Hidden-states of the model projected to *config.proj_codevector_dim* that can be used to predict the masked
projected quantized states.
projected_quantized_states (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length, config.proj_codevector_dim)`):
Quantized extracted feature vectors projected to `config.proj_codevector_dim` representing the positive
projected_quantized_states (`torch.FloatTensor` of shape `(batch_size, sequence_length, config.proj_codevector_dim)`):
Quantized extracted feature vectors projected to *config.proj_codevector_dim* representing the positive
target vectors for contrastive loss.
hidden_states (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_hidden_states=True`` is passed or when ``config.output_hidden_states=True``):
Tuple of :obj:`torch.FloatTensor` (one for the output of the embeddings + one for the output of each layer)
of shape :obj:`(batch_size, sequence_length, hidden_size)`.
hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `torch.FloatTensor` (one for the output of the embeddings + one for the output of each layer)
of shape `(batch_size, sequence_length, hidden_size)`.
Hidden-states of the model at the output of each layer plus the initial embedding outputs.
attentions (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_attentions=True`` is passed or when ``config.output_attentions=True``):
Tuple of :obj:`torch.FloatTensor` (one for each layer) of shape :obj:`(batch_size, num_heads,
sequence_length, sequence_length)`.
attentions (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length, sequence_length)`.
Attentions weights after the attention softmax, used to compute the weighted average in the self-attention
heads.
contrastive_loss (`optional`, returned when :obj:`sample_negative_indices` are passed, ``torch.FloatTensor`` of shape :obj:`(1,)`):
The contrastive loss (L_m) as stated in the `official paper <https://arxiv.org/pdf/2006.11477.pdf>`__ .
diversity_loss (`optional`, returned when :obj:`sample_negative_indices` are passed, ``torch.FloatTensor`` of shape :obj:`(1,)`):
The diversity loss (L_d) as stated in the `official paper <https://arxiv.org/pdf/2006.11477.pdf>`__ .
contrastive_loss (*optional*, returned when `sample_negative_indices` are passed, `torch.FloatTensor` of shape `(1,)`):
The contrastive loss (L_m) as stated in the [official paper](https://arxiv.org/pdf/2006.11477.pdf) .
diversity_loss (*optional*, returned when `sample_negative_indices` are passed, `torch.FloatTensor` of shape `(1,)`):
The diversity loss (L_d) as stated in the [official paper](https://arxiv.org/pdf/2006.11477.pdf) .
"""
loss: Optional[torch.FloatTensor] = None
......@@ -143,23 +141,22 @@ class Wav2Vec2ForPreTrainingOutput(ModelOutput):
@dataclass
class XVectorOutput(ModelOutput):
"""
Output type of :class:`~transformers.Wav2Vec2ForXVector`.
Output type of [`Wav2Vec2ForXVector`].
Args:
loss (:obj:`torch.FloatTensor` of shape :obj:`(1,)`, `optional`, returned when :obj:`labels` is provided):
loss (`torch.FloatTensor` of shape `(1,)`, *optional*, returned when `labels` is provided):
Classification loss.
logits (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, config.xvector_output_dim)`):
logits (`torch.FloatTensor` of shape `(batch_size, config.xvector_output_dim)`):
Classification hidden states before AMSoftmax.
embeddings (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, config.xvector_output_dim)`):
embeddings (`torch.FloatTensor` of shape `(batch_size, config.xvector_output_dim)`):
Utterance embeddings used for vector similarity-based retrieval.
hidden_states (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_hidden_states=True`` is passed or when ``config.output_hidden_states=True``):
Tuple of :obj:`torch.FloatTensor` (one for the output of the embeddings + one for the output of each layer)
of shape :obj:`(batch_size, sequence_length, hidden_size)`.
hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `torch.FloatTensor` (one for the output of the embeddings + one for the output of each layer)
of shape `(batch_size, sequence_length, hidden_size)`.
Hidden-states of the model at the output of each layer plus the initial embedding outputs.
attentions (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_attentions=True`` is passed or when ``config.output_attentions=True``):
Tuple of :obj:`torch.FloatTensor` (one for each layer) of shape :obj:`(batch_size, num_heads,
sequence_length, sequence_length)`.
attentions (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length, sequence_length)`.
Attentions weights after the attention softmax, used to compute the weighted average in the self-attention
heads.
......@@ -1130,58 +1127,58 @@ class Wav2Vec2PreTrainedModel(PreTrainedModel):
WAV_2_VEC_2_START_DOCSTRING = r"""
Wav2Vec2 was proposed in `wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations
<https://arxiv.org/abs/2006.11477>`__ by Alexei Baevski, Henry Zhou, Abdelrahman Mohamed, Michael Auli.
Wav2Vec2 was proposed in [wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations](https://arxiv.org/abs/2006.11477) by Alexei Baevski, Henry Zhou, Abdelrahman Mohamed, Michael Auli.
This model inherits from :class:`~transformers.PreTrainedModel`. Check the superclass documentation for the generic
This model inherits from [`PreTrainedModel`]. Check the superclass documentation for the generic
methods the library implements for all its model (such as downloading or saving etc.).
This model is a PyTorch `torch.nn.Module <https://pytorch.org/docs/stable/nn.html#torch.nn.Module>`_ sub-class. Use
This model is a PyTorch [torch.nn.Module](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) sub-class. Use
it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and
behavior.
Parameters:
config (:class:`~transformers.Wav2Vec2Config`): Model configuration class with all the parameters of the model.
config ([`Wav2Vec2Config`]): Model configuration class with all the parameters of the model.
Initializing with a config file does not load the weights associated with the model, only the
configuration. Check out the :meth:`~transformers.PreTrainedModel.from_pretrained` method to load the model
configuration. Check out the [`~PreTrainedModel.from_pretrained`] method to load the model
weights.
"""
WAV_2_VEC_2_INPUTS_DOCSTRING = r"""
Args:
input_values (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length)`):
Float values of input raw speech waveform. Values can be obtained by loading a `.flac` or `.wav` audio file
into an array of type `List[float]` or a `numpy.ndarray`, *e.g.* via the soundfile library (`pip install
soundfile`). To prepare the array into `input_values`, the :class:`~transformers.Wav2Vec2Processor` should
be used for padding and conversion into a tensor of type `torch.FloatTensor`. See
:meth:`transformers.Wav2Vec2Processor.__call__` for details.
attention_mask (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`):
Mask to avoid performing convolution and attention on padding token indices. Mask values selected in ``[0,
1]``:
input_values (`torch.FloatTensor` of shape `(batch_size, sequence_length)`):
Float values of input raw speech waveform. Values can be obtained by loading a *.flac* or *.wav* audio file
into an array of type *List[float]* or a *numpy.ndarray*, *e.g.* via the soundfile library (*pip install
soundfile*). To prepare the array into *input_values*, the [`Wav2Vec2Processor`] should
be used for padding and conversion into a tensor of type *torch.FloatTensor*. See
[`Wav2Vec2Processor.__call__`] for details.
attention_mask (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
Mask to avoid performing convolution and attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
`What are attention masks? <../glossary.html#attention-mask>`__
[What are attention masks?](../glossary#attention-mask)
.. warning::
:obj:`attention_mask` should only be passed if the corresponding processor has
``config.return_attention_mask == True``. For all models whose processor has
``config.return_attention_mask == False``, such as `wav2vec2-base
<https://huggingface.co/facebook/wav2vec2-base-960h>`__, :obj:`attention_mask` should **not** be passed
to avoid degraded performance when doing batched inference. For such models :obj:`input_values` should
simply be padded with 0 and passed without :obj:`attention_mask`. Be aware that these models also yield
slightly different results depending on whether :obj:`input_values` is padded or not.
<Tip warning={true}>
output_attentions (:obj:`bool`, `optional`):
Whether or not to return the attentions tensors of all attention layers. See ``attentions`` under returned
`attention_mask` should only be passed if the corresponding processor has
`config.return_attention_mask == True`. For all models whose processor has
`config.return_attention_mask == False`, such as [wav2vec2-base](https://huggingface.co/facebook/wav2vec2-base-960h), `attention_mask` should **not** be passed
to avoid degraded performance when doing batched inference. For such models `input_values` should
simply be padded with 0 and passed without `attention_mask`. Be aware that these models also yield
slightly different results depending on whether `input_values` is padded or not.
</Tip>
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under returned
tensors for more detail.
output_hidden_states (:obj:`bool`, `optional`):
Whether or not to return the hidden states of all layers. See ``hidden_states`` under returned tensors for
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for
more detail.
return_dict (:obj:`bool`, `optional`):
Whether or not to return a :class:`~transformers.file_utils.ModelOutput` instead of a plain tuple.
return_dict (`bool`, *optional*):
Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple.
"""
......@@ -1378,58 +1375,59 @@ class Wav2Vec2ForPreTraining(Wav2Vec2PreTrainedModel):
return_dict=None,
):
r"""
mask_time_indices (:obj:`torch.BoolTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`):
mask_time_indices (`torch.BoolTensor` of shape `(batch_size, sequence_length)`, *optional*):
Indices to mask extracted features for contrastive loss. When in training mode, model learns to predict
masked extracted features in `config.proj_codevector_dim` space.
sampled_negative_indices (:obj:`torch.BoolTensor` of shape :obj:`(batch_size, sequence_length, num_negatives)`, `optional`):
masked extracted features in *config.proj_codevector_dim* space.
sampled_negative_indices (`torch.BoolTensor` of shape `(batch_size, sequence_length, num_negatives)`, *optional*):
Indices indicating which quantized target vectors are used as negative sampled vectors in contrastive loss.
Required input for pre-training.
Returns:
Example::
Example:
>>> import torch
>>> from transformers import Wav2Vec2FeatureExtractor, Wav2Vec2ForPreTraining
>>> from transformers.models.wav2vec2.modeling_wav2vec2 import _compute_mask_indices
>>> from datasets import load_dataset
>>> import soundfile as sf
```python
>>> import torch
>>> from transformers import Wav2Vec2FeatureExtractor, Wav2Vec2ForPreTraining
>>> from transformers.models.wav2vec2.modeling_wav2vec2 import _compute_mask_indices
>>> from datasets import load_dataset
>>> import soundfile as sf
>>> feature_extractor = Wav2Vec2FeatureExtractor.from_pretrained("patrickvonplaten/wav2vec2-base")
>>> model = Wav2Vec2ForPreTraining.from_pretrained("patrickvonplaten/wav2vec2-base")
>>> feature_extractor = Wav2Vec2FeatureExtractor.from_pretrained("patrickvonplaten/wav2vec2-base")
>>> model = Wav2Vec2ForPreTraining.from_pretrained("patrickvonplaten/wav2vec2-base")
>>> def map_to_array(batch):
... speech, _ = sf.read(batch["file"])
... batch["speech"] = speech
... return batch
>>> def map_to_array(batch):
... speech, _ = sf.read(batch["file"])
... batch["speech"] = speech
... return batch
>>> ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
>>> ds = ds.map(map_to_array)
>>> ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
>>> ds = ds.map(map_to_array)
>>> input_values = feature_extractor(ds["speech"][0], return_tensors="pt").input_values # Batch size 1
>>> input_values = feature_extractor(ds["speech"][0], return_tensors="pt").input_values # Batch size 1
>>> # compute masked indices
>>> batch_size, raw_sequence_length = input_values.shape
>>> sequence_length = model._get_feat_extract_output_lengths(raw_sequence_length)
>>> mask_time_indices = _compute_mask_indices((batch_size, sequence_length), mask_prob=0.2, mask_length=2)
>>> # compute masked indices
>>> batch_size, raw_sequence_length = input_values.shape
>>> sequence_length = model._get_feat_extract_output_lengths(raw_sequence_length)
>>> mask_time_indices = _compute_mask_indices((batch_size, sequence_length), mask_prob=0.2, mask_length=2)
>>> with torch.no_grad():
... outputs = model(input_values, mask_time_indices=mask_time_indices)
>>> with torch.no_grad():
... outputs = model(input_values, mask_time_indices=mask_time_indices)
>>> # compute cosine similarity between predicted (=projected_states) and target (=projected_quantized_states)
>>> cosine_sim = torch.cosine_similarity(
... outputs.projected_states, outputs.projected_quantized_states, dim=-1
... )
>>> # compute cosine similarity between predicted (=projected_states) and target (=projected_quantized_states)
>>> cosine_sim = torch.cosine_similarity(
... outputs.projected_states, outputs.projected_quantized_states, dim=-1
... )
>>> # show that cosine similarity is much higher than random
>>> assert cosine_sim[mask_time_indices].mean() > 0.5
>>> # show that cosine similarity is much higher than random
>>> assert cosine_sim[mask_time_indices].mean() > 0.5
>>> # for contrastive loss training model should be put into train mode
>>> model.train()
>>> loss = model(input_values, mask_time_indices=mask_time_indices).loss
"""
>>> # for contrastive loss training model should be put into train mode
>>> model.train()
>>> loss = model(input_values, mask_time_indices=mask_time_indices).loss
```"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
......@@ -1551,34 +1549,35 @@ class Wav2Vec2ForMaskedLM(Wav2Vec2PreTrainedModel):
labels=None,
):
r"""
labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length, hidden_size)`, `optional`):
labels (`torch.LongTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*):
TODO(PVP): Fill out when adding training
Returns:
Example::
Example:
>>> from transformers import Wav2Vec2Processor, Wav2Vec2ForMaskedLM
>>> from datasets import load_dataset
>>> import soundfile as sf
```python
>>> from transformers import Wav2Vec2Processor, Wav2Vec2ForMaskedLM
>>> from datasets import load_dataset
>>> import soundfile as sf
>>> processor = Wav2Vec2Processor.from_pretrained("facebook/wav2vec2-base-960h")
>>> model = Wav2Vec2ForMaskedLM.from_pretrained("facebook/wav2vec2-base-960h")
>>> processor = Wav2Vec2Processor.from_pretrained("facebook/wav2vec2-base-960h")
>>> model = Wav2Vec2ForMaskedLM.from_pretrained("facebook/wav2vec2-base-960h")
>>> def map_to_array(batch):
>>> speech, _ = sf.read(batch["file"])
>>> batch["speech"] = speech
>>> return batch
>>> def map_to_array(batch):
>>> speech, _ = sf.read(batch["file"])
>>> batch["speech"] = speech
>>> return batch
>>> ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
>>> ds = ds.map(map_to_array)
>>> ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
>>> ds = ds.map(map_to_array)
>>> input_values = processor(ds["speech"][0], return_tensors="pt").input_values # Batch size 1
>>> logits = model(input_values).logits
>>> input_values = processor(ds["speech"][0], return_tensors="pt").input_values # Batch size 1
>>> logits = model(input_values).logits
>>> predicted_ids = torch.argmax(logits, dim=-1)
>>> transcription = processor.decode(predicted_ids[0])
"""
>>> predicted_ids = torch.argmax(logits, dim=-1)
>>> transcription = processor.decode(predicted_ids[0])
```"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
......@@ -1647,11 +1646,9 @@ class Wav2Vec2ForCTC(Wav2Vec2PreTrainedModel):
labels=None,
):
r"""
labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size, target_length)`, `optional`):
Labels for connectionist temporal classification. Note that ``target_length`` has to be smaller or equal to
the sequence length of the output logits. Indices are selected in ``[-100, 0, ..., config.vocab_size -
1]``. All labels set to ``-100`` are ignored (masked), the loss is only computed for labels in ``[0, ...,
config.vocab_size - 1]``.
labels (`torch.LongTensor` of shape `(batch_size, target_length)`, *optional*):
Labels for connectionist temporal classification. Note that `target_length` has to be smaller or equal to
the sequence length of the output logits. Indices are selected in `[-100, 0, ..., config.vocab_size - 1]`. All labels set to `-100` are ignored (masked), the loss is only computed for labels in `[0, ..., config.vocab_size - 1]`.
"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
......@@ -1764,10 +1761,9 @@ class Wav2Vec2ForSequenceClassification(Wav2Vec2PreTrainedModel):
labels=None,
):
r"""
labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`):
Labels for computing the sequence classification/regression loss. Indices should be in :obj:`[0, ...,
config.num_labels - 1]`. If :obj:`config.num_labels == 1` a regression loss is computed (Mean-Square loss),
If :obj:`config.num_labels > 1` a classification loss is computed (Cross-Entropy).
labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
Labels for computing the sequence classification/regression loss. Indices should be in `[0, ..., config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss),
If `config.num_labels > 1` a classification loss is computed (Cross-Entropy).
"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
......@@ -1866,10 +1862,9 @@ class Wav2Vec2ForAudioFrameClassification(Wav2Vec2PreTrainedModel):
return_dict=None,
):
r"""
labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`):
Labels for computing the sequence classification/regression loss. Indices should be in :obj:`[0, ...,
config.num_labels - 1]`. If :obj:`config.num_labels == 1` a regression loss is computed (Mean-Square loss),
If :obj:`config.num_labels > 1` a classification loss is computed (Cross-Entropy).
labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
Labels for computing the sequence classification/regression loss. Indices should be in `[0, ..., config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss),
If `config.num_labels > 1` a classification loss is computed (Cross-Entropy).
"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
......@@ -2028,10 +2023,9 @@ class Wav2Vec2ForXVector(Wav2Vec2PreTrainedModel):
labels=None,
):
r"""
labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`):
Labels for computing the sequence classification/regression loss. Indices should be in :obj:`[0, ...,
config.num_labels - 1]`. If :obj:`config.num_labels == 1` a regression loss is computed (Mean-Square loss),
If :obj:`config.num_labels > 1` a classification loss is computed (Cross-Entropy).
labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
Labels for computing the sequence classification/regression loss. Indices should be in `[0, ..., config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss),
If `config.num_labels > 1` a classification loss is computed (Cross-Entropy).
"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
......
src/transformers/models/wavlm/modeling_wavlm.py
View file @ 7af80f66
......@@ -65,21 +65,20 @@ WAVLM_PRETRAINED_MODEL_ARCHIVE_LIST = [
@dataclass
class WavLMBaseModelOutput(ModelOutput):
"""
Output type of :class:`~transformers.WavLMBaseModelOutput`, with potential hidden states and attentions.
Output type of [`WavLMBaseModelOutput`], with potential hidden states and attentions.
Args:
last_hidden_state (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length, hidden_size)`):
last_hidden_state (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`):
Sequence of hidden-states at the output of the last layer of the model.
extract_features (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length, conv_dim[-1])`):
extract_features (`torch.FloatTensor` of shape `(batch_size, sequence_length, conv_dim[-1])`):
Sequence of extracted feature vectors of the last convolutional layer of the model.
hidden_states (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_hidden_states=True`` is passed or when ``config.output_hidden_states=True``):
Tuple of :obj:`torch.FloatTensor` (one for the output of the embeddings + one for the output of each layer)
of shape :obj:`(batch_size, sequence_length, hidden_size)`.
hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `torch.FloatTensor` (one for the output of the embeddings + one for the output of each layer)
of shape `(batch_size, sequence_length, hidden_size)`.
Hidden-states of the model at the output of each layer plus the initial embedding outputs.
attentions (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_attentions=True`` is passed or when ``config.output_attentions=True``):
Tuple of :obj:`torch.FloatTensor` (one for each layer) of shape :obj:`(batch_size, num_heads,
sequence_length, sequence_length)`.
attentions (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length, sequence_length)`.
Attentions weights after the attention softmax, used to compute the weighted average in the self-attention
heads.
......@@ -94,23 +93,22 @@ class WavLMBaseModelOutput(ModelOutput):
@dataclass
class XVectorOutput(ModelOutput):
"""
Output type of :class:`~transformers.Wav2Vec2ForXVector`.
Output type of [`Wav2Vec2ForXVector`].
Args:
loss (:obj:`torch.FloatTensor` of shape :obj:`(1,)`, `optional`, returned when :obj:`labels` is provided):
loss (`torch.FloatTensor` of shape `(1,)`, *optional*, returned when `labels` is provided):
Classification loss.
logits (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, config.xvector_output_dim)`):
logits (`torch.FloatTensor` of shape `(batch_size, config.xvector_output_dim)`):
Classification hidden states before AMSoftmax.
embeddings (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, config.xvector_output_dim)`):
embeddings (`torch.FloatTensor` of shape `(batch_size, config.xvector_output_dim)`):
Utterance embeddings used for vector similarity-based retrieval.
hidden_states (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_hidden_states=True`` is passed or when ``config.output_hidden_states=True``):
Tuple of :obj:`torch.FloatTensor` (one for the output of the embeddings + one for the output of each layer)
of shape :obj:`(batch_size, sequence_length, hidden_size)`.
hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `torch.FloatTensor` (one for the output of the embeddings + one for the output of each layer)
of shape `(batch_size, sequence_length, hidden_size)`.
Hidden-states of the model at the output of each layer plus the initial embedding outputs.
attentions (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_attentions=True`` is passed or when ``config.output_attentions=True``):
Tuple of :obj:`torch.FloatTensor` (one for each layer) of shape :obj:`(batch_size, num_heads,
sequence_length, sequence_length)`.
attentions (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length, sequence_length)`.
Attentions weights after the attention softmax, used to compute the weighted average in the self-attention
heads.
......@@ -1082,58 +1080,59 @@ class WavLMPreTrainedModel(PreTrainedModel):
WAVLM_START_DOCSTRING = r"""
WavLM was proposed in `WavLM: Unified Speech Representation Learning with Labeled and Unlabeled Data
<https://arxiv.org/abs/2101.07597>`__ by Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei,
WavLM was proposed in [WavLM: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597) by Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei,
Michael Zeng, Xuedong Huang.
This model inherits from :class:`~transformers.PreTrainedModel`. Check the superclass documentation for the generic
This model inherits from [`PreTrainedModel`]. Check the superclass documentation for the generic
methods the library implements for all its model (such as downloading or saving etc.).
This model is a PyTorch `torch.nn.Module <https://pytorch.org/docs/stable/nn.html#torch.nn.Module>`_ sub-class. Use
This model is a PyTorch [torch.nn.Module](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) sub-class. Use
it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and
behavior.
Parameters:
config (:class:`~transformers.WavLMConfig`): Model configuration class with all the parameters of the model.
config ([`WavLMConfig`]): Model configuration class with all the parameters of the model.
Initializing with a config file does not load the weights associated with the model, only the
configuration. Check out the :meth:`~transformers.PreTrainedModel.from_pretrained` method to load the model
configuration. Check out the [`~PreTrainedModel.from_pretrained`] method to load the model
weights.
"""
WAVLM_INPUTS_DOCSTRING = r"""
Args:
input_values (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length)`):
Float values of input raw speech waveform. Values can be obtained by loading a `.flac` or `.wav` audio file
into an array of type `List[float]` or a `numpy.ndarray`, *e.g.* via the soundfile library (`pip install
soundfile`). To prepare the array into `input_values`, the :class:`~transformers.WavLMProcessor` should be
used for padding and conversion into a tensor of type `torch.FloatTensor`. See
:meth:`transformers.WavLMProcessor.__call__` for details.
attention_mask (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`):
Mask to avoid performing convolution and attention on padding token indices. Mask values selected in ``[0,
1]``:
input_values (`torch.FloatTensor` of shape `(batch_size, sequence_length)`):
Float values of input raw speech waveform. Values can be obtained by loading a *.flac* or *.wav* audio file
into an array of type *List[float]* or a *numpy.ndarray*, *e.g.* via the soundfile library (*pip install
soundfile*). To prepare the array into *input_values*, the [`WavLMProcessor`] should be
used for padding and conversion into a tensor of type *torch.FloatTensor*. See
[`WavLMProcessor.__call__`] for details.
attention_mask (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
Mask to avoid performing convolution and attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
`What are attention masks? <../glossary.html#attention-mask>`__
[What are attention masks?](../glossary#attention-mask)
.. warning::
:obj:`attention_mask` should only be passed if the corresponding processor has
``config.return_attention_mask == True``. For all models whose processor has
``config.return_attention_mask == False``, :obj:`attention_mask` should **not** be passed to avoid
degraded performance when doing batched inference. For such models :obj:`input_values` should simply be
padded with 0 and passed without :obj:`attention_mask`. Be aware that these models also yield slightly
different results depending on whether :obj:`input_values` is padded or not.
<Tip warning={true}>
output_attentions (:obj:`bool`, `optional`):
Whether or not to return the attentions tensors of all attention layers. See ``attentions`` under returned
`attention_mask` should only be passed if the corresponding processor has
`config.return_attention_mask == True`. For all models whose processor has
`config.return_attention_mask == False`, `attention_mask` should **not** be passed to avoid
degraded performance when doing batched inference. For such models `input_values` should simply be
padded with 0 and passed without `attention_mask`. Be aware that these models also yield slightly
different results depending on whether `input_values` is padded or not.
</Tip>
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under returned
tensors for more detail.
output_hidden_states (:obj:`bool`, `optional`):
Whether or not to return the hidden states of all layers. See ``hidden_states`` under returned tensors for
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for
more detail.
return_dict (:obj:`bool`, `optional`):
Whether or not to return a :class:`~transformers.file_utils.ModelOutput` instead of a plain tuple.
return_dict (`bool`, *optional*):
Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple.
"""
......@@ -1316,11 +1315,9 @@ class WavLMForCTC(WavLMPreTrainedModel):
labels=None,
):
r"""
labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size, target_length)`, `optional`):
Labels for connectionist temporal classification. Note that ``target_length`` has to be smaller or equal to
the sequence length of the output logits. Indices are selected in ``[-100, 0, ..., config.vocab_size -
1]``. All labels set to ``-100`` are ignored (masked), the loss is only computed for labels in ``[0, ...,
config.vocab_size - 1]``.
labels (`torch.LongTensor` of shape `(batch_size, target_length)`, *optional*):
Labels for connectionist temporal classification. Note that `target_length` has to be smaller or equal to
the sequence length of the output logits. Indices are selected in `[-100, 0, ..., config.vocab_size - 1]`. All labels set to `-100` are ignored (masked), the loss is only computed for labels in `[0, ..., config.vocab_size - 1]`.
"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
......@@ -1434,10 +1431,9 @@ class WavLMForSequenceClassification(WavLMPreTrainedModel):
labels=None,
):
r"""
labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`):
Labels for computing the sequence classification/regression loss. Indices should be in :obj:`[0, ...,
config.num_labels - 1]`. If :obj:`config.num_labels == 1` a regression loss is computed (Mean-Square loss),
If :obj:`config.num_labels > 1` a classification loss is computed (Cross-Entropy).
labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
Labels for computing the sequence classification/regression loss. Indices should be in `[0, ..., config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss),
If `config.num_labels > 1` a classification loss is computed (Cross-Entropy).
"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
......@@ -1537,10 +1533,9 @@ class WavLMForAudioFrameClassification(WavLMPreTrainedModel):
return_dict=None,
):
r"""
labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`):
Labels for computing the sequence classification/regression loss. Indices should be in :obj:`[0, ...,
config.num_labels - 1]`. If :obj:`config.num_labels == 1` a regression loss is computed (Mean-Square loss),
If :obj:`config.num_labels > 1` a classification loss is computed (Cross-Entropy).
labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
Labels for computing the sequence classification/regression loss. Indices should be in `[0, ..., config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss),
If `config.num_labels > 1` a classification loss is computed (Cross-Entropy).
"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
......@@ -1702,10 +1697,9 @@ class WavLMForXVector(WavLMPreTrainedModel):
labels=None,
):
r"""
labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`):
Labels for computing the sequence classification/regression loss. Indices should be in :obj:`[0, ...,
config.num_labels - 1]`. If :obj:`config.num_labels == 1` a regression loss is computed (Mean-Square loss),
If :obj:`config.num_labels > 1` a classification loss is computed (Cross-Entropy).
labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
Labels for computing the sequence classification/regression loss. Indices should be in `[0, ..., config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss),
If `config.num_labels > 1` a classification loss is computed (Cross-Entropy).
"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
......
src/transformers/models/xlm/modeling_tf_xlm.py
View file @ 7af80f66
......@@ -557,19 +557,18 @@ class TFXLMPreTrainedModel(TFPreTrainedModel):
@dataclass
class TFXLMWithLMHeadModelOutput(ModelOutput):
"""
Base class for :class:`~transformers.TFXLMWithLMHeadModel` outputs.
Base class for [`TFXLMWithLMHeadModel`] outputs.
Args:
logits (:obj:`tf.Tensor` of shape :obj:`(batch_size, sequence_length, config.vocab_size)`):
logits (`tf.Tensor` of shape `(batch_size, sequence_length, config.vocab_size)`):
Prediction scores of the language modeling head (scores for each vocabulary token before SoftMax).
hidden_states (:obj:`tuple(tf.Tensor)`, `optional`, returned when ``output_hidden_states=True`` is passed or when ``config.output_hidden_states=True``):
Tuple of :obj:`tf.Tensor` (one for the output of the embeddings + one for the output of each layer) of
shape :obj:`(batch_size, sequence_length, hidden_size)`.
hidden_states (`tuple(tf.Tensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `tf.Tensor` (one for the output of the embeddings + one for the output of each layer) of
shape `(batch_size, sequence_length, hidden_size)`.
Hidden-states of the model at the output of each layer plus the initial embedding outputs.
attentions (:obj:`tuple(tf.Tensor)`, `optional`, returned when ``output_attentions=True`` is passed or when ``config.output_attentions=True``):
Tuple of :obj:`tf.Tensor` (one for each layer) of shape :obj:`(batch_size, num_heads, sequence_length,
sequence_length)`.
attentions (`tuple(tf.Tensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `tf.Tensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length, sequence_length)`.
Attentions weights after the attention softmax, used to compute the weighted average in the self-attention
heads.
......@@ -582,111 +581,111 @@ class TFXLMWithLMHeadModelOutput(ModelOutput):
XLM_START_DOCSTRING = r"""
This model inherits from :class:`~transformers.TFPreTrainedModel`. Check the superclass documentation for the
This model inherits from [`TFPreTrainedModel`]. Check the superclass documentation for the
generic methods the library implements for all its model (such as downloading or saving, resizing the input
embeddings, pruning heads etc.)
This model is also a `tf.keras.Model <https://www.tensorflow.org/api_docs/python/tf/keras/Model>`__ subclass. Use
This model is also a [tf.keras.Model](https://www.tensorflow.org/api_docs/python/tf/keras/Model) subclass. Use
it as a regular TF 2.0 Keras Model and refer to the TF 2.0 documentation for all matter related to general usage
and behavior.
.. note::
<Tip>
TF 2.0 models accepts two formats as inputs:
TF 2.0 models accepts two formats as inputs:
- having all inputs as keyword arguments (like PyTorch models), or
- having all inputs as a list, tuple or dict in the first positional arguments.
- having all inputs as keyword arguments (like PyTorch models), or
- having all inputs as a list, tuple or dict in the first positional arguments.
This second option is useful when using :meth:`tf.keras.Model.fit` method which currently requires having all
the tensors in the first argument of the model call function: :obj:`model(inputs)`.
This second option is useful when using [`tf.keras.Model.fit`] method which currently requires having all
the tensors in the first argument of the model call function: `model(inputs)`.
If you choose this second option, there are three possibilities you can use to gather all the input Tensors in
the first positional argument :
If you choose this second option, there are three possibilities you can use to gather all the input Tensors in
the first positional argument :
- a single Tensor with :obj:`input_ids` only and nothing else: :obj:`model(inputs_ids)`
- a list of varying length with one or several input Tensors IN THE ORDER given in the docstring:
:obj:`model([input_ids, attention_mask])` or :obj:`model([input_ids, attention_mask, token_type_ids])`
- a dictionary with one or several input Tensors associated to the input names given in the docstring:
:obj:`model({"input_ids": input_ids, "token_type_ids": token_type_ids})`
- a single Tensor with `input_ids` only and nothing else: `model(inputs_ids)`
- a list of varying length with one or several input Tensors IN THE ORDER given in the docstring:
`model([input_ids, attention_mask])` or `model([input_ids, attention_mask, token_type_ids])`
- a dictionary with one or several input Tensors associated to the input names given in the docstring:
`model({"input_ids": input_ids, "token_type_ids": token_type_ids})`
</Tip>
Parameters:
config (:class:`~transformers.XLMConfig`): Model configuration class with all the parameters of the model.
config ([`XLMConfig`]): Model configuration class with all the parameters of the model.
Initializing with a config file does not load the weights associated with the model, only the
configuration. Check out the :meth:`~transformers.PreTrainedModel.from_pretrained` method to load the model
configuration. Check out the [`~PreTrainedModel.from_pretrained`] method to load the model
weights.
"""
XLM_INPUTS_DOCSTRING = r"""
Args:
input_ids (:obj:`Numpy array` or :obj:`tf.Tensor` of shape :obj:`({0})`):
input_ids (`Numpy array` or `tf.Tensor` of shape `({0})`):
Indices of input sequence tokens in the vocabulary.
Indices can be obtained using :class:`~transformers.BertTokenizer`. See
:func:`transformers.PreTrainedTokenizer.__call__` and :func:`transformers.PreTrainedTokenizer.encode` for
Indices can be obtained using [`BertTokenizer`]. See
[`PreTrainedTokenizer.__call__`] and [`PreTrainedTokenizer.encode`] for
details.
`What are input IDs? <../glossary.html#input-ids>`__
attention_mask (:obj:`Numpy array` or :obj:`tf.Tensor` of shape :obj:`({0})`, `optional`):
Mask to avoid performing attention on padding token indices. Mask values selected in ``[0, 1]``:
[What are input IDs?](../glossary#input-ids)
attention_mask (`Numpy array` or `tf.Tensor` of shape `({0})`, *optional*):
Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
`What are attention masks? <../glossary.html#attention-mask>`__
langs (:obj:`tf.Tensor` or :obj:`Numpy array` of shape :obj:`({0})`, `optional`):
[What are attention masks?](../glossary#attention-mask)
langs (`tf.Tensor` or `Numpy array` of shape `({0})`, *optional*):
A parallel sequence of tokens to be used to indicate the language of each token in the input. Indices are
languages ids which can be obtained from the language names by using two conversion mappings provided in
the configuration of the model (only provided for multilingual models). More precisely, the `language name
to language id` mapping is in :obj:`model.config.lang2id` (which is a dictionary string to int) and the
`language id to language name` mapping is in :obj:`model.config.id2lang` (dictionary int to string).
the configuration of the model (only provided for multilingual models). More precisely, the *language name
to language id* mapping is in `model.config.lang2id` (which is a dictionary string to int) and the
*language id to language name* mapping is in `model.config.id2lang` (dictionary int to string).
See usage examples detailed in the :doc:`multilingual documentation <../multilingual>`.
token_type_ids (:obj:`Numpy array` or :obj:`tf.Tensor` of shape :obj:`({0})`, `optional`):
Segment token indices to indicate first and second portions of the inputs. Indices are selected in ``[0,
1]``:
See usage examples detailed in the [multilingual documentation](../multilingual).
token_type_ids (`Numpy array` or `tf.Tensor` of shape `({0})`, *optional*):
Segment token indices to indicate first and second portions of the inputs. Indices are selected in `[0, 1]`:
- 0 corresponds to a `sentence A` token,
- 1 corresponds to a `sentence B` token.
- 0 corresponds to a *sentence A* token,
- 1 corresponds to a *sentence B* token.
`What are token type IDs? <../glossary.html#token-type-ids>`__
position_ids (:obj:`Numpy array` or :obj:`tf.Tensor` of shape :obj:`({0})`, `optional`):
Indices of positions of each input sequence tokens in the position embeddings. Selected in the range ``[0,
config.max_position_embeddings - 1]``.
[What are token type IDs?](../glossary#token-type-ids)
position_ids (`Numpy array` or `tf.Tensor` of shape `({0})`, *optional*):
Indices of positions of each input sequence tokens in the position embeddings. Selected in the range `[0, config.max_position_embeddings - 1]`.
`What are position IDs? <../glossary.html#position-ids>`__
lengths (:obj:`tf.Tensor` or :obj:`Numpy array` of shape :obj:`(batch_size,)`, `optional`):
[What are position IDs?](../glossary#position-ids)
lengths (`tf.Tensor` or `Numpy array` of shape `(batch_size,)`, *optional*):
Length of each sentence that can be used to avoid performing attention on padding token indices. You can
also use `attention_mask` for the same result (see above), kept here for compatibility. Indices selected in
``[0, ..., input_ids.size(-1)]``.
cache (:obj:`Dict[str, tf.Tensor]`, `optional`):
Dictionary string to ``torch.FloatTensor`` that contains precomputed hidden states (key and values in the
attention blocks) as computed by the model (see :obj:`cache` output below). Can be used to speed up
also use *attention_mask* for the same result (see above), kept here for compatibility. Indices selected in
`[0, ..., input_ids.size(-1)]`.
cache (`Dict[str, tf.Tensor]`, *optional*):
Dictionary string to `torch.FloatTensor` that contains precomputed hidden states (key and values in the
attention blocks) as computed by the model (see `cache` output below). Can be used to speed up
sequential decoding.
The dictionary object will be modified in-place during the forward pass to add newly computed
hidden-states.
head_mask (:obj:`Numpy array` or :obj:`tf.Tensor` of shape :obj:`(num_heads,)` or :obj:`(num_layers, num_heads)`, `optional`):
Mask to nullify selected heads of the self-attention modules. Mask values selected in ``[0, 1]``:
head_mask (`Numpy array` or `tf.Tensor` of shape `(num_heads,)` or `(num_layers, num_heads)`, *optional*):
Mask to nullify selected heads of the self-attention modules. Mask values selected in `[0, 1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
inputs_embeds (:obj:`tf.Tensor` of shape :obj:`({0}, hidden_size)`, `optional`):
Optionally, instead of passing :obj:`input_ids` you can choose to directly pass an embedded representation.
This is useful if you want more control over how to convert :obj:`input_ids` indices into associated
inputs_embeds (`tf.Tensor` of shape `({0}, hidden_size)`, *optional*):
Optionally, instead of passing `input_ids` you can choose to directly pass an embedded representation.
This is useful if you want more control over how to convert `input_ids` indices into associated
vectors than the model's internal embedding lookup matrix.
output_attentions (:obj:`bool`, `optional`):
Whether or not to return the attentions tensors of all attention layers. See ``attentions`` under returned
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under returned
tensors for more detail. This argument can be used only in eager mode, in graph mode the value in the
config will be used instead.
output_hidden_states (:obj:`bool`, `optional`):
Whether or not to return the hidden states of all layers. See ``hidden_states`` under returned tensors for
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for
more detail. This argument can be used only in eager mode, in graph mode the value in the config will be
used instead.
return_dict (:obj:`bool`, `optional`):
Whether or not to return a :class:`~transformers.file_utils.ModelOutput` instead of a plain tuple. This
return_dict (`bool`, *optional*):
Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple. This
argument can be used in eager mode, in graph mode the value will always be set to True.
training (:obj:`bool`, `optional`, defaults to :obj:`False`):
training (`bool`, *optional*, defaults to `False`):
Whether or not to use the model in training mode (some modules like dropout modules have different
behaviors between training and evaluation).
"""
......@@ -970,10 +969,9 @@ class TFXLMForSequenceClassification(TFXLMPreTrainedModel, TFSequenceClassificat
**kwargs,
):
r"""
labels (:obj:`tf.Tensor` of shape :obj:`(batch_size,)`, `optional`):
Labels for computing the sequence classification/regression loss. Indices should be in ``[0, ...,
config.num_labels - 1]``. If ``config.num_labels == 1`` a regression loss is computed (Mean-Square loss),
If ``config.num_labels > 1`` a classification loss is computed (Cross-Entropy).
labels (`tf.Tensor` of shape `(batch_size,)`, *optional*):
Labels for computing the sequence classification/regression loss. Indices should be in `[0, ..., config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss),
If `config.num_labels > 1` a classification loss is computed (Cross-Entropy).
"""
inputs = input_processing(
func=self.call,
......@@ -1246,9 +1244,8 @@ class TFXLMForTokenClassification(TFXLMPreTrainedModel, TFTokenClassificationLos
**kwargs,
):
r"""
labels (:obj:`tf.Tensor` of shape :obj:`(batch_size, sequence_length)`, `optional`):
Labels for computing the token classification loss. Indices should be in ``[0, ..., config.num_labels -
1]``.
labels (`tf.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
Labels for computing the token classification loss. Indices should be in `[0, ..., config.num_labels - 1]`.
"""
inputs = input_processing(
func=self.call,
......@@ -1352,13 +1349,13 @@ class TFXLMForQuestionAnsweringSimple(TFXLMPreTrainedModel, TFQuestionAnsweringL
**kwargs,
):
r"""
start_positions (:obj:`tf.Tensor` of shape :obj:`(batch_size,)`, `optional`):
start_positions (`tf.Tensor` of shape `(batch_size,)`, *optional*):
Labels for position (index) of the start of the labelled span for computing the token classification loss.
Positions are clamped to the length of the sequence (:obj:`sequence_length`). Position outside of the
Positions are clamped to the length of the sequence (`sequence_length`). Position outside of the
sequence are not taken into account for computing the loss.
end_positions (:obj:`tf.Tensor` of shape :obj:`(batch_size,)`, `optional`):
end_positions (`tf.Tensor` of shape `(batch_size,)`, *optional*):
Labels for position (index) of the end of the labelled span for computing the token classification loss.
Positions are clamped to the length of the sequence (:obj:`sequence_length`). Position outside of the
Positions are clamped to the length of the sequence (`sequence_length`). Position outside of the
sequence are not taken into account for computing the loss.
"""
inputs = input_processing(
......
src/transformers/models/xlm/modeling_xlm.py
View file @ 7af80f66
......@@ -269,31 +269,30 @@ class XLMPreTrainedModel(PreTrainedModel):
@dataclass
class XLMForQuestionAnsweringOutput(ModelOutput):
"""
Base class for outputs of question answering models using a :obj:`SquadHead`.
Base class for outputs of question answering models using a `SquadHead`.
Args:
loss (:obj:`torch.FloatTensor` of shape :obj:`(1,)`, `optional`, returned if both :obj:`start_positions` and :obj:`end_positions` are provided):
loss (`torch.FloatTensor` of shape `(1,)`, *optional*, returned if both `start_positions` and `end_positions` are provided):
Classification loss as the sum of start token, end token (and is_impossible if provided) classification
losses.
start_top_log_probs (``torch.FloatTensor`` of shape ``(batch_size, config.start_n_top)``, `optional`, returned if ``start_positions`` or ``end_positions`` is not provided):
start_top_log_probs (`torch.FloatTensor` of shape `(batch_size, config.start_n_top)`, *optional*, returned if `start_positions` or `end_positions` is not provided):
Log probabilities for the top config.start_n_top start token possibilities (beam-search).
start_top_index (``torch.LongTensor`` of shape ``(batch_size, config.start_n_top)``, `optional`, returned if ``start_positions`` or ``end_positions`` is not provided):
start_top_index (`torch.LongTensor` of shape `(batch_size, config.start_n_top)`, *optional*, returned if `start_positions` or `end_positions` is not provided):
Indices for the top config.start_n_top start token possibilities (beam-search).
end_top_log_probs (``torch.FloatTensor`` of shape ``(batch_size, config.start_n_top * config.end_n_top)``, `optional`, returned if ``start_positions`` or ``end_positions`` is not provided):
Log probabilities for the top ``config.start_n_top * config.end_n_top`` end token possibilities
end_top_log_probs (`torch.FloatTensor` of shape `(batch_size, config.start_n_top * config.end_n_top)`, *optional*, returned if `start_positions` or `end_positions` is not provided):
Log probabilities for the top `config.start_n_top * config.end_n_top` end token possibilities
(beam-search).
end_top_index (``torch.LongTensor`` of shape ``(batch_size, config.start_n_top * config.end_n_top)``, `optional`, returned if ``start_positions`` or ``end_positions`` is not provided):
Indices for the top ``config.start_n_top * config.end_n_top`` end token possibilities (beam-search).
cls_logits (``torch.FloatTensor`` of shape ``(batch_size,)``, `optional`, returned if ``start_positions`` or ``end_positions`` is not provided):
Log probabilities for the ``is_impossible`` label of the answers.
hidden_states (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_hidden_states=True`` is passed or when ``config.output_hidden_states=True``):
Tuple of :obj:`torch.FloatTensor` (one for the output of the embeddings + one for the output of each layer)
of shape :obj:`(batch_size, sequence_length, hidden_size)`.
end_top_index (`torch.LongTensor` of shape `(batch_size, config.start_n_top * config.end_n_top)`, *optional*, returned if `start_positions` or `end_positions` is not provided):
Indices for the top `config.start_n_top * config.end_n_top` end token possibilities (beam-search).
cls_logits (`torch.FloatTensor` of shape `(batch_size,)`, *optional*, returned if `start_positions` or `end_positions` is not provided):
Log probabilities for the `is_impossible` label of the answers.
hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `torch.FloatTensor` (one for the output of the embeddings + one for the output of each layer)
of shape `(batch_size, sequence_length, hidden_size)`.
Hidden-states of the model at the output of each layer plus the initial embedding outputs.
attentions (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_attentions=True`` is passed or when ``config.output_attentions=True``):
Tuple of :obj:`torch.FloatTensor` (one for each layer) of shape :obj:`(batch_size, num_heads,
sequence_length, sequence_length)`.
attentions (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length, sequence_length)`.
Attentions weights after the attention softmax, used to compute the weighted average in the self-attention
heads.
......@@ -311,88 +310,86 @@ class XLMForQuestionAnsweringOutput(ModelOutput):
XLM_START_DOCSTRING = r"""
This model inherits from :class:`~transformers.PreTrainedModel`. Check the superclass documentation for the generic
This model inherits from [`PreTrainedModel`]. Check the superclass documentation for the generic
methods the library implements for all its model (such as downloading or saving, resizing the input embeddings,
pruning heads etc.)
This model is also a PyTorch `torch.nn.Module <https://pytorch.org/docs/stable/nn.html#torch.nn.Module>`__
This model is also a PyTorch [torch.nn.Module](https://pytorch.org/docs/stable/nn.html#torch.nn.Module)
subclass. Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to
general usage and behavior.
Parameters:
config (:class:`~transformers.XLMConfig`): Model configuration class with all the parameters of the model.
config ([`XLMConfig`]): Model configuration class with all the parameters of the model.
Initializing with a config file does not load the weights associated with the model, only the
configuration. Check out the :meth:`~transformers.PreTrainedModel.from_pretrained` method to load the model
configuration. Check out the [`~PreTrainedModel.from_pretrained`] method to load the model
weights.
"""
XLM_INPUTS_DOCSTRING = r"""
Args:
input_ids (:obj:`torch.LongTensor` of shape :obj:`({0})`):
input_ids (`torch.LongTensor` of shape `({0})`):
Indices of input sequence tokens in the vocabulary.
Indices can be obtained using :class:`~transformers.XLMTokenizer`. See
:meth:`transformers.PreTrainedTokenizer.encode` and :meth:`transformers.PreTrainedTokenizer.__call__` for
Indices can be obtained using [`XLMTokenizer`]. See
[`PreTrainedTokenizer.encode`] and [`PreTrainedTokenizer.__call__`] for
details.
`What are input IDs? <../glossary.html#input-ids>`__
attention_mask (:obj:`torch.FloatTensor` of shape :obj:`({0})`, `optional`):
Mask to avoid performing attention on padding token indices. Mask values selected in ``[0, 1]``:
[What are input IDs?](../glossary#input-ids)
attention_mask (`torch.FloatTensor` of shape `({0})`, *optional*):
Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
`What are attention masks? <../glossary.html#attention-mask>`__
langs (:obj:`torch.LongTensor` of shape :obj:`({0})`, `optional`):
[What are attention masks?](../glossary#attention-mask)
langs (`torch.LongTensor` of shape `({0})`, *optional*):
A parallel sequence of tokens to be used to indicate the language of each token in the input. Indices are
languages ids which can be obtained from the language names by using two conversion mappings provided in
the configuration of the model (only provided for multilingual models). More precisely, the `language name
to language id` mapping is in :obj:`model.config.lang2id` (which is a dictionary string to int) and the
`language id to language name` mapping is in :obj:`model.config.id2lang` (dictionary int to string).
the configuration of the model (only provided for multilingual models). More precisely, the *language name
to language id* mapping is in `model.config.lang2id` (which is a dictionary string to int) and the
*language id to language name* mapping is in `model.config.id2lang` (dictionary int to string).
See usage examples detailed in the :doc:`multilingual documentation <../multilingual>`.
token_type_ids (:obj:`torch.LongTensor` of shape :obj:`({0})`, `optional`):
Segment token indices to indicate first and second portions of the inputs. Indices are selected in ``[0,
1]``:
See usage examples detailed in the [multilingual documentation](../multilingual).
token_type_ids (`torch.LongTensor` of shape `({0})`, *optional*):
Segment token indices to indicate first and second portions of the inputs. Indices are selected in `[0, 1]`:
- 0 corresponds to a `sentence A` token,
- 1 corresponds to a `sentence B` token.
- 0 corresponds to a *sentence A* token,
- 1 corresponds to a *sentence B* token.
`What are token type IDs? <../glossary.html#token-type-ids>`__
position_ids (:obj:`torch.LongTensor` of shape :obj:`({0})`, `optional`):
Indices of positions of each input sequence tokens in the position embeddings. Selected in the range ``[0,
config.max_position_embeddings - 1]``.
[What are token type IDs?](../glossary#token-type-ids)
position_ids (`torch.LongTensor` of shape `({0})`, *optional*):
Indices of positions of each input sequence tokens in the position embeddings. Selected in the range `[0, config.max_position_embeddings - 1]`.
`What are position IDs? <../glossary.html#position-ids>`__
lengths (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`):
[What are position IDs?](../glossary#position-ids)
lengths (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
Length of each sentence that can be used to avoid performing attention on padding token indices. You can
also use `attention_mask` for the same result (see above), kept here for compatibility. Indices selected in
``[0, ..., input_ids.size(-1)]``.
cache (:obj:`Dict[str, torch.FloatTensor]`, `optional`):
Dictionary string to ``torch.FloatTensor`` that contains precomputed hidden states (key and values in the
attention blocks) as computed by the model (see :obj:`cache` output below). Can be used to speed up
also use *attention_mask* for the same result (see above), kept here for compatibility. Indices selected in
`[0, ..., input_ids.size(-1)]`.
cache (`Dict[str, torch.FloatTensor]`, *optional*):
Dictionary string to `torch.FloatTensor` that contains precomputed hidden states (key and values in the
attention blocks) as computed by the model (see `cache` output below). Can be used to speed up
sequential decoding.
The dictionary object will be modified in-place during the forward pass to add newly computed
hidden-states.
head_mask (:obj:`torch.FloatTensor` of shape :obj:`(num_heads,)` or :obj:`(num_layers, num_heads)`, `optional`):
Mask to nullify selected heads of the self-attention modules. Mask values selected in ``[0, 1]``:
head_mask (`torch.FloatTensor` of shape `(num_heads,)` or `(num_layers, num_heads)`, *optional*):
Mask to nullify selected heads of the self-attention modules. Mask values selected in `[0, 1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
inputs_embeds (:obj:`torch.FloatTensor` of shape :obj:`({0}, hidden_size)`, `optional`):
Optionally, instead of passing :obj:`input_ids` you can choose to directly pass an embedded representation.
This is useful if you want more control over how to convert :obj:`input_ids` indices into associated
inputs_embeds (`torch.FloatTensor` of shape `({0}, hidden_size)`, *optional*):
Optionally, instead of passing `input_ids` you can choose to directly pass an embedded representation.
This is useful if you want more control over how to convert `input_ids` indices into associated
vectors than the model's internal embedding lookup matrix.
output_attentions (:obj:`bool`, `optional`):
Whether or not to return the attentions tensors of all attention layers. See ``attentions`` under returned
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under returned
tensors for more detail.
output_hidden_states (:obj:`bool`, `optional`):
Whether or not to return the hidden states of all layers. See ``hidden_states`` under returned tensors for
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for
more detail.
return_dict (:obj:`bool`, `optional`):
Whether or not to return a :class:`~transformers.file_utils.ModelOutput` instead of a plain tuple.
return_dict (`bool`, *optional*):
Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple.
"""
......@@ -735,10 +732,10 @@ class XLMWithLMHeadModel(XLMPreTrainedModel):
return_dict=None,
):
r"""
labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`):
labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
Labels for language modeling. Note that the labels **are shifted** inside the model, i.e. you can set
``labels = input_ids`` Indices are selected in ``[-100, 0, ..., config.vocab_size]`` All labels set to
``-100`` are ignored (masked), the loss is only computed for labels in ``[0, ..., config.vocab_size]``
`labels = input_ids` Indices are selected in `[-100, 0, ..., config.vocab_size]` All labels set to
`-100` are ignored (masked), the loss is only computed for labels in `[0, ..., config.vocab_size]`
"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
......@@ -814,10 +811,9 @@ class XLMForSequenceClassification(XLMPreTrainedModel):
return_dict=None,
):
r"""
labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`):
Labels for computing the sequence classification/regression loss. Indices should be in :obj:`[0, ...,
config.num_labels - 1]`. If :obj:`config.num_labels == 1` a regression loss is computed (Mean-Square loss),
If :obj:`config.num_labels > 1` a classification loss is computed (Cross-Entropy).
labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
Labels for computing the sequence classification/regression loss. Indices should be in `[0, ..., config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss),
If `config.num_labels > 1` a classification loss is computed (Cross-Entropy).
"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
......@@ -916,13 +912,13 @@ class XLMForQuestionAnsweringSimple(XLMPreTrainedModel):
return_dict=None,
):
r"""
start_positions (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`):
start_positions (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
Labels for position (index) of the start of the labelled span for computing the token classification loss.
Positions are clamped to the length of the sequence (:obj:`sequence_length`). Position outside of the
Positions are clamped to the length of the sequence (`sequence_length`). Position outside of the
sequence are not taken into account for computing the loss.
end_positions (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`):
end_positions (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
Labels for position (index) of the end of the labelled span for computing the token classification loss.
Positions are clamped to the length of the sequence (:obj:`sequence_length`). Position outside of the
Positions are clamped to the length of the sequence (`sequence_length`). Position outside of the
sequence are not taken into account for computing the loss.
"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
......@@ -1019,40 +1015,41 @@ class XLMForQuestionAnswering(XLMPreTrainedModel):
return_dict=None,
):
r"""
start_positions (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`):
start_positions (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
Labels for position (index) of the start of the labelled span for computing the token classification loss.
Positions are clamped to the length of the sequence (:obj:`sequence_length`). Position outside of the
Positions are clamped to the length of the sequence (`sequence_length`). Position outside of the
sequence are not taken into account for computing the loss.
end_positions (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`):
end_positions (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
Labels for position (index) of the end of the labelled span for computing the token classification loss.
Positions are clamped to the length of the sequence (:obj:`sequence_length`). Position outside of the
Positions are clamped to the length of the sequence (`sequence_length`). Position outside of the
sequence are not taken into account for computing the loss.
is_impossible (``torch.LongTensor`` of shape ``(batch_size,)``, `optional`):
is_impossible (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
Labels whether a question has an answer or no answer (SQuAD 2.0)
cls_index (``torch.LongTensor`` of shape ``(batch_size,)``, `optional`):
cls_index (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
Labels for position (index) of the classification token to use as input for computing plausibility of the
answer.
p_mask (``torch.FloatTensor`` of shape ``(batch_size, sequence_length)``, `optional`):
p_mask (`torch.FloatTensor` of shape `(batch_size, sequence_length)`, *optional*):
Optional mask of tokens which can't be in answers (e.g. [CLS], [PAD], ...). 1.0 means token should be
masked. 0.0 mean token is not masked.
Returns:
Example::
Example:
>>> from transformers import XLMTokenizer, XLMForQuestionAnswering
>>> import torch
```python
>>> from transformers import XLMTokenizer, XLMForQuestionAnswering
>>> import torch
>>> tokenizer = XLMTokenizer.from_pretrained('xlm-mlm-en-2048')
>>> model = XLMForQuestionAnswering.from_pretrained('xlm-mlm-en-2048')
>>> tokenizer = XLMTokenizer.from_pretrained('xlm-mlm-en-2048')
>>> model = XLMForQuestionAnswering.from_pretrained('xlm-mlm-en-2048')
>>> input_ids = torch.tensor(tokenizer.encode("Hello, my dog is cute", add_special_tokens=True)).unsqueeze(0) # Batch size 1
>>> start_positions = torch.tensor([1])
>>> end_positions = torch.tensor([3])
>>> input_ids = torch.tensor(tokenizer.encode("Hello, my dog is cute", add_special_tokens=True)).unsqueeze(0) # Batch size 1
>>> start_positions = torch.tensor([1])
>>> end_positions = torch.tensor([3])
>>> outputs = model(input_ids, start_positions=start_positions, end_positions=end_positions)
>>> loss = outputs.loss
"""
>>> outputs = model(input_ids, start_positions=start_positions, end_positions=end_positions)
>>> loss = outputs.loss
```"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
transformer_outputs = self.transformer(
......@@ -1140,9 +1137,8 @@ class XLMForTokenClassification(XLMPreTrainedModel):
return_dict=None,
):
r"""
labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`):
Labels for computing the token classification loss. Indices should be in ``[0, ..., config.num_labels -
1]``.
labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
Labels for computing the token classification loss. Indices should be in `[0, ..., config.num_labels - 1]`.
"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
......@@ -1234,10 +1230,9 @@ class XLMForMultipleChoice(XLMPreTrainedModel):
return_dict=None,
):
r"""
labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`):
Labels for computing the multiple choice classification loss. Indices should be in ``[0, ...,
num_choices-1]`` where :obj:`num_choices` is the size of the second dimension of the input tensors. (See
:obj:`input_ids` above)
labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
Labels for computing the multiple choice classification loss. Indices should be in `[0, ..., num_choices-1]` where `num_choices` is the size of the second dimension of the input tensors. (See
`input_ids` above)
"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
num_choices = input_ids.shape[1] if input_ids is not None else inputs_embeds.shape[1]
......
src/transformers/models/xlm_prophetnet/modeling_xlm_prophetnet.py
View file @ 7af80f66
......@@ -37,130 +37,135 @@ XLM_PROPHETNET_PRETRAINED_MODEL_ARCHIVE_LIST = [
class XLMProphetNetEncoder(ProphetNetEncoder):
r"""
This class overrides :class:`~transformers.ProphetNetEncoder`. Please check the superclass for the appropriate
This class overrides [`ProphetNetEncoder`]. Please check the superclass for the appropriate
documentation alongside usage examples.
Example::
Example:
>>> from transformers import XLMProphetNetTokenizer, XLMProphetNetEncoder
>>> import torch
```python
>>> from transformers import XLMProphetNetTokenizer, XLMProphetNetEncoder
>>> import torch
>>> tokenizer = XLMProphetNetTokenizer.from_pretrained('microsoft/xprophetnet-large-wiki100-cased')
>>> model = XLMProphetNetEncoder.from_pretrained('patrickvonplaten/xprophetnet-large-uncased-standalone')
>>> assert model.config.is_decoder, f"{model.__class__} has to be configured as a decoder."
>>> inputs = tokenizer("Hello, my dog is cute", return_tensors="pt")
>>> outputs = model(**inputs)
>>> tokenizer = XLMProphetNetTokenizer.from_pretrained('microsoft/xprophetnet-large-wiki100-cased')
>>> model = XLMProphetNetEncoder.from_pretrained('patrickvonplaten/xprophetnet-large-uncased-standalone')
>>> assert model.config.is_decoder, f"{model.__class__} has to be configured as a decoder."
>>> inputs = tokenizer("Hello, my dog is cute", return_tensors="pt")
>>> outputs = model(**inputs)
>>> last_hidden_states = outputs.last_hidden_state
"""
>>> last_hidden_states = outputs.last_hidden_state
```"""
config_class = XLMProphetNetConfig
class XLMProphetNetDecoder(ProphetNetDecoder):
r"""
This class overrides :class:`~transformers.ProphetNetDecoder`. Please check the superclass for the appropriate
This class overrides [`ProphetNetDecoder`]. Please check the superclass for the appropriate
documentation alongside usage examples.
Example::
Example:
>>> from transformers import XLMProphetNetTokenizer, XLMProphetNetDecoder
>>> import torch
```python
>>> from transformers import XLMProphetNetTokenizer, XLMProphetNetDecoder
>>> import torch
>>> tokenizer = XLMProphetNetTokenizer.from_pretrained('microsoft/xprophetnet-large-wiki100-cased')
>>> model = XLMProphetNetDecoder.from_pretrained('patrickvonplaten/xprophetnet-large-uncased-standalone', add_cross_attention=False)
>>> assert model.config.is_decoder, f"{model.__class__} has to be configured as a decoder."
>>> inputs = tokenizer("Hello, my dog is cute", return_tensors="pt")
>>> outputs = model(**inputs)
>>> tokenizer = XLMProphetNetTokenizer.from_pretrained('microsoft/xprophetnet-large-wiki100-cased')
>>> model = XLMProphetNetDecoder.from_pretrained('patrickvonplaten/xprophetnet-large-uncased-standalone', add_cross_attention=False)
>>> assert model.config.is_decoder, f"{model.__class__} has to be configured as a decoder."
>>> inputs = tokenizer("Hello, my dog is cute", return_tensors="pt")
>>> outputs = model(**inputs)
>>> last_hidden_states = outputs.last_hidden_state
"""
>>> last_hidden_states = outputs.last_hidden_state
```"""
config_class = XLMProphetNetConfig
class XLMProphetNetModel(ProphetNetModel):
r"""
This class overrides :class:`~transformers.ProphetNetModel`. Please check the superclass for the appropriate
This class overrides [`ProphetNetModel`]. Please check the superclass for the appropriate
documentation alongside usage examples.
Example::
Example:
>>> from transformers import XLMProphetNetTokenizer, XLMProphetNetModel
```python
>>> from transformers import XLMProphetNetTokenizer, XLMProphetNetModel
>>> tokenizer = XLMProphetNetTokenizer.from_pretrained('microsoft/xprophetnet-large-wiki100-cased')
>>> model = XLMProphetNetModel.from_pretrained('microsoft/xprophetnet-large-wiki100-cased')
>>> tokenizer = XLMProphetNetTokenizer.from_pretrained('microsoft/xprophetnet-large-wiki100-cased')
>>> model = XLMProphetNetModel.from_pretrained('microsoft/xprophetnet-large-wiki100-cased')
>>> input_ids = tokenizer("Studies have been shown that owning a dog is good for you", return_tensors="pt").input_ids # Batch size 1
>>> decoder_input_ids = tokenizer("Studies show that", return_tensors="pt").input_ids # Batch size 1
>>> outputs = model(input_ids=input_ids, decoder_input_ids=decoder_input_ids)
>>> input_ids = tokenizer("Studies have been shown that owning a dog is good for you", return_tensors="pt").input_ids # Batch size 1
>>> decoder_input_ids = tokenizer("Studies show that", return_tensors="pt").input_ids # Batch size 1
>>> outputs = model(input_ids=input_ids, decoder_input_ids=decoder_input_ids)
>>> last_hidden_states = outputs.last_hidden_state # main stream hidden states
>>> last_hidden_states_ngram = outputs.last_hidden_state_ngram # predict hidden states
"""
>>> last_hidden_states = outputs.last_hidden_state # main stream hidden states
>>> last_hidden_states_ngram = outputs.last_hidden_state_ngram # predict hidden states
```"""
config_class = XLMProphetNetConfig
class XLMProphetNetForConditionalGeneration(ProphetNetForConditionalGeneration):
r"""
This class overrides :class:`~transformers.ProphetNetForConditionalGeneration`. Please check the superclass for the
This class overrides [`ProphetNetForConditionalGeneration`]. Please check the superclass for the
appropriate documentation alongside usage examples.
Example::
Example:
>>> from transformers import XLMProphetNetTokenizer, XLMProphetNetForConditionalGeneration
```python
>>> from transformers import XLMProphetNetTokenizer, XLMProphetNetForConditionalGeneration
>>> tokenizer = XLMProphetNetTokenizer.from_pretrained('microsoft/xprophetnet-large-wiki100-cased')
>>> model = XLMProphetNetForConditionalGeneration.from_pretrained('microsoft/xprophetnet-large-wiki100-cased')
>>> tokenizer = XLMProphetNetTokenizer.from_pretrained('microsoft/xprophetnet-large-wiki100-cased')
>>> model = XLMProphetNetForConditionalGeneration.from_pretrained('microsoft/xprophetnet-large-wiki100-cased')
>>> input_ids = tokenizer("Studies have been shown that owning a dog is good for you", return_tensors="pt").input_ids # Batch size 1
>>> decoder_input_ids = tokenizer("Studies show that", return_tensors="pt").input_ids # Batch size 1
>>> outputs = model(input_ids=input_ids, decoder_input_ids=decoder_input_ids)
>>> input_ids = tokenizer("Studies have been shown that owning a dog is good for you", return_tensors="pt").input_ids # Batch size 1
>>> decoder_input_ids = tokenizer("Studies show that", return_tensors="pt").input_ids # Batch size 1
>>> outputs = model(input_ids=input_ids, decoder_input_ids=decoder_input_ids)
>>> logits_next_token = outputs.logits # logits to predict next token as usual
>>> logits_ngram_next_tokens = outputs.logits_ngram # logits to predict 2nd, 3rd, ... next tokens
"""
>>> logits_next_token = outputs.logits # logits to predict next token as usual
>>> logits_ngram_next_tokens = outputs.logits_ngram # logits to predict 2nd, 3rd, ... next tokens
```"""
config_class = XLMProphetNetConfig
class XLMProphetNetForCausalLM(ProphetNetForCausalLM):
r"""
This class overrides :class:`~transformers.ProphetNetForCausalLM`. Please check the superclass for the appropriate
This class overrides [`ProphetNetForCausalLM`]. Please check the superclass for the appropriate
documentation alongside usage examples.
Example::
Example:
>>> from transformers import XLMProphetNetTokenizer, XLMProphetNetForCausalLM
>>> import torch
```python
>>> from transformers import XLMProphetNetTokenizer, XLMProphetNetForCausalLM
>>> import torch
>>> tokenizer = XLMProphetNetTokenizer.from_pretrained('microsoft/xprophetnet-large-wiki100-cased')
>>> model = XLMProphetNetForCausalLM.from_pretrained('microsoft/xprophetnet-large-wiki100-cased')
>>> assert model.config.is_decoder, f"{model.__class__} has to be configured as a decoder."
>>> inputs = tokenizer("Hello, my dog is cute", return_tensors="pt")
>>> outputs = model(**inputs)
>>> tokenizer = XLMProphetNetTokenizer.from_pretrained('microsoft/xprophetnet-large-wiki100-cased')
>>> model = XLMProphetNetForCausalLM.from_pretrained('microsoft/xprophetnet-large-wiki100-cased')
>>> assert model.config.is_decoder, f"{model.__class__} has to be configured as a decoder."
>>> inputs = tokenizer("Hello, my dog is cute", return_tensors="pt")
>>> outputs = model(**inputs)
>>> logits = outputs.logits
>>> logits = outputs.logits
>>> # Model can also be used with EncoderDecoder framework
>>> from transformers import EncoderDecoderModel, XLMProphetNetTokenizer, XLMRobertaTokenizer
>>> import torch
>>> # Model can also be used with EncoderDecoder framework
>>> from transformers import EncoderDecoderModel, XLMProphetNetTokenizer, XLMRobertaTokenizer
>>> import torch
>>> tokenizer_enc = XLMRobertaTokenizer.from_pretrained('xlm-roberta-large')
>>> tokenizer_dec = XLMProphetNetTokenizer.from_pretrained('microsoft/xprophetnet-large-wiki100-cased')
>>> model = EncoderDecoderModel.from_encoder_decoder_pretrained("xlm-roberta-large", 'microsoft/xprophetnet-large-wiki100-cased')
>>> tokenizer_enc = XLMRobertaTokenizer.from_pretrained('xlm-roberta-large')
>>> tokenizer_dec = XLMProphetNetTokenizer.from_pretrained('microsoft/xprophetnet-large-wiki100-cased')
>>> model = EncoderDecoderModel.from_encoder_decoder_pretrained("xlm-roberta-large", 'microsoft/xprophetnet-large-wiki100-cased')
>>> ARTICLE = (
... "the us state department said wednesday it had received no "
... "formal word from bolivia that it was expelling the us ambassador there "
... "but said the charges made against him are `` baseless ."
... )
>>> input_ids = tokenizer_enc(ARTICLE, return_tensors="pt").input_ids
>>> labels = tokenizer_dec("us rejects charges against its ambassador in bolivia", return_tensors="pt").input_ids
>>> outputs = model(input_ids=input_ids, decoder_input_ids=labels[:, :-1], labels=labels[:, 1:])
>>> ARTICLE = (
... "the us state department said wednesday it had received no "
... "formal word from bolivia that it was expelling the us ambassador there "
... "but said the charges made against him are `` baseless ."
... )
>>> input_ids = tokenizer_enc(ARTICLE, return_tensors="pt").input_ids
>>> labels = tokenizer_dec("us rejects charges against its ambassador in bolivia", return_tensors="pt").input_ids
>>> outputs = model(input_ids=input_ids, decoder_input_ids=labels[:, :-1], labels=labels[:, 1:])
>>> loss = outputs.loss
"""
>>> loss = outputs.loss
```"""
config_class = XLMProphetNetConfig
src/transformers/models/xlm_roberta/modeling_tf_xlm_roberta.py
View file @ 7af80f66
......@@ -38,37 +38,39 @@ TF_XLM_ROBERTA_PRETRAINED_MODEL_ARCHIVE_LIST = [
XLM_ROBERTA_START_DOCSTRING = r"""
This model inherits from :class:`~transformers.TFPreTrainedModel`. Check the superclass documentation for the
This model inherits from [`TFPreTrainedModel`]. Check the superclass documentation for the
generic methods the library implements for all its model (such as downloading or saving, resizing the input
embeddings, pruning heads etc.)
This model is also a `tf.keras.Model <https://www.tensorflow.org/api_docs/python/tf/keras/Model>`__ subclass. Use
This model is also a [tf.keras.Model](https://www.tensorflow.org/api_docs/python/tf/keras/Model) subclass. Use
it as a regular TF 2.0 Keras Model and refer to the TF 2.0 documentation for all matter related to general usage
and behavior.
.. note::
<Tip>
TF 2.0 models accepts two formats as inputs:
TF 2.0 models accepts two formats as inputs:
- having all inputs as keyword arguments (like PyTorch models), or
- having all inputs as a list, tuple or dict in the first positional arguments.
- having all inputs as keyword arguments (like PyTorch models), or
- having all inputs as a list, tuple or dict in the first positional arguments.
This second option is useful when using :meth:`tf.keras.Model.fit` method which currently requires having all
the tensors in the first argument of the model call function: :obj:`model(inputs)`.
This second option is useful when using [`tf.keras.Model.fit`] method which currently requires having all
the tensors in the first argument of the model call function: `model(inputs)`.
If you choose this second option, there are three possibilities you can use to gather all the input Tensors in
the first positional argument :
If you choose this second option, there are three possibilities you can use to gather all the input Tensors in
the first positional argument :
- a single Tensor with :obj:`input_ids` only and nothing else: :obj:`model(inputs_ids)`
- a list of varying length with one or several input Tensors IN THE ORDER given in the docstring:
:obj:`model([input_ids, attention_mask])` or :obj:`model([input_ids, attention_mask, token_type_ids])`
- a dictionary with one or several input Tensors associated to the input names given in the docstring:
:obj:`model({"input_ids": input_ids, "token_type_ids": token_type_ids})`
- a single Tensor with `input_ids` only and nothing else: `model(inputs_ids)`
- a list of varying length with one or several input Tensors IN THE ORDER given in the docstring:
`model([input_ids, attention_mask])` or `model([input_ids, attention_mask, token_type_ids])`
- a dictionary with one or several input Tensors associated to the input names given in the docstring:
`model({"input_ids": input_ids, "token_type_ids": token_type_ids})`
</Tip>
Parameters:
config (:class:`~transformers.XLMRobertaConfig`): Model configuration class with all the parameters of the
config ([`XLMRobertaConfig`]): Model configuration class with all the parameters of the
model. Initializing with a config file does not load the weights associated with the model, only the
configuration. Check out the :meth:`~transformers.PreTrainedModel.from_pretrained` method to load the model
configuration. Check out the [`~PreTrainedModel.from_pretrained`] method to load the model
weights.
"""
......@@ -79,7 +81,7 @@ XLM_ROBERTA_START_DOCSTRING = r"""
)
class TFXLMRobertaModel(TFRobertaModel):
"""
This class overrides :class:`~transformers.TFRobertaModel`. Please check the superclass for the appropriate
This class overrides [`TFRobertaModel`]. Please check the superclass for the appropriate
documentation alongside usage examples.
"""
......@@ -92,7 +94,7 @@ class TFXLMRobertaModel(TFRobertaModel):
)
class XLMRobertaForCausalLM(TFRobertaForCausalLM):
"""
This class overrides :class:`~transformers.TFRobertaForCausalLM`. Please check the superclass for the appropriate
This class overrides [`TFRobertaForCausalLM`]. Please check the superclass for the appropriate
documentation alongside usage examples.
"""
......@@ -105,7 +107,7 @@ class XLMRobertaForCausalLM(TFRobertaForCausalLM):
)
class TFXLMRobertaForMaskedLM(TFRobertaForMaskedLM):
"""
This class overrides :class:`~transformers.TFRobertaForMaskedLM`. Please check the superclass for the appropriate
This class overrides [`TFRobertaForMaskedLM`]. Please check the superclass for the appropriate
documentation alongside usage examples.
"""
......@@ -121,7 +123,7 @@ class TFXLMRobertaForMaskedLM(TFRobertaForMaskedLM):
)
class TFXLMRobertaForSequenceClassification(TFRobertaForSequenceClassification):
"""
This class overrides :class:`~transformers.TFRobertaForSequenceClassification`. Please check the superclass for the
This class overrides [`TFRobertaForSequenceClassification`]. Please check the superclass for the
appropriate documentation alongside usage examples.
"""
......@@ -137,7 +139,7 @@ class TFXLMRobertaForSequenceClassification(TFRobertaForSequenceClassification):
)
class TFXLMRobertaForTokenClassification(TFRobertaForTokenClassification):
"""
This class overrides :class:`~transformers.TFRobertaForTokenClassification`. Please check the superclass for the
This class overrides [`TFRobertaForTokenClassification`]. Please check the superclass for the
appropriate documentation alongside usage examples.
"""
......@@ -153,7 +155,7 @@ layers on top of the hidden-states output to compute `span start logits` and `sp
)
class TFXLMRobertaForQuestionAnswering(TFRobertaForQuestionAnswering):
"""
This class overrides :class:`~transformers.TFRobertaForQuestionAnsweringSimple`. Please check the superclass for
This class overrides [`TFRobertaForQuestionAnsweringSimple`]. Please check the superclass for
the appropriate documentation alongside usage examples.
"""
......@@ -169,7 +171,7 @@ class TFXLMRobertaForQuestionAnswering(TFRobertaForQuestionAnswering):
)
class TFXLMRobertaForMultipleChoice(TFRobertaForMultipleChoice):
"""
This class overrides :class:`~transformers.TFRobertaForMultipleChoice`. Please check the superclass for the
This class overrides [`TFRobertaForMultipleChoice`]. Please check the superclass for the
appropriate documentation alongside usage examples.
"""
......
src/transformers/models/xlm_roberta/modeling_xlm_roberta.py
View file @ 7af80f66
......@@ -44,18 +44,18 @@ XLM_ROBERTA_PRETRAINED_MODEL_ARCHIVE_LIST = [
XLM_ROBERTA_START_DOCSTRING = r"""
This model inherits from :class:`~transformers.PreTrainedModel`. Check the superclass documentation for the generic
This model inherits from [`PreTrainedModel`]. Check the superclass documentation for the generic
methods the library implements for all its model (such as downloading or saving, resizing the input embeddings,
pruning heads etc.)
This model is also a PyTorch `torch.nn.Module <https://pytorch.org/docs/stable/nn.html#torch.nn.Module>`__
This model is also a PyTorch [torch.nn.Module](https://pytorch.org/docs/stable/nn.html#torch.nn.Module)
subclass. Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to
general usage and behavior.
Parameters:
config (:class:`~transformers.XLMRobertaConfig`): Model configuration class with all the parameters of the
config ([`XLMRobertaConfig`]): Model configuration class with all the parameters of the
model. Initializing with a config file does not load the weights associated with the model, only the
configuration. Check out the :meth:`~transformers.PreTrainedModel.from_pretrained` method to load the model
configuration. Check out the [`~PreTrainedModel.from_pretrained`] method to load the model
weights.
"""
......@@ -66,7 +66,7 @@ XLM_ROBERTA_START_DOCSTRING = r"""
)
class XLMRobertaModel(RobertaModel):
"""
This class overrides :class:`~transformers.RobertaModel`. Please check the superclass for the appropriate
This class overrides [`RobertaModel`]. Please check the superclass for the appropriate
documentation alongside usage examples.
"""
......@@ -79,7 +79,7 @@ class XLMRobertaModel(RobertaModel):
)
class XLMRobertaForCausalLM(RobertaForCausalLM):
"""
This class overrides :class:`~transformers.RobertaForCausalLM`. Please check the superclass for the appropriate
This class overrides [`RobertaForCausalLM`]. Please check the superclass for the appropriate
documentation alongside usage examples.
"""
......@@ -92,7 +92,7 @@ class XLMRobertaForCausalLM(RobertaForCausalLM):
)
class XLMRobertaForMaskedLM(RobertaForMaskedLM):
"""
This class overrides :class:`~transformers.RobertaForMaskedLM`. Please check the superclass for the appropriate
This class overrides [`RobertaForMaskedLM`]. Please check the superclass for the appropriate
documentation alongside usage examples.
"""
......@@ -108,7 +108,7 @@ class XLMRobertaForMaskedLM(RobertaForMaskedLM):
)
class XLMRobertaForSequenceClassification(RobertaForSequenceClassification):
"""
This class overrides :class:`~transformers.RobertaForSequenceClassification`. Please check the superclass for the
This class overrides [`RobertaForSequenceClassification`]. Please check the superclass for the
appropriate documentation alongside usage examples.
"""
......@@ -124,7 +124,7 @@ class XLMRobertaForSequenceClassification(RobertaForSequenceClassification):
)
class XLMRobertaForMultipleChoice(RobertaForMultipleChoice):
"""
This class overrides :class:`~transformers.RobertaForMultipleChoice`. Please check the superclass for the
This class overrides [`RobertaForMultipleChoice`]. Please check the superclass for the
appropriate documentation alongside usage examples.
"""
......@@ -140,7 +140,7 @@ class XLMRobertaForMultipleChoice(RobertaForMultipleChoice):
)
class XLMRobertaForTokenClassification(RobertaForTokenClassification):
"""
This class overrides :class:`~transformers.RobertaForTokenClassification`. Please check the superclass for the
This class overrides [`RobertaForTokenClassification`]. Please check the superclass for the
appropriate documentation alongside usage examples.
"""
......@@ -156,7 +156,7 @@ class XLMRobertaForTokenClassification(RobertaForTokenClassification):
)
class XLMRobertaForQuestionAnswering(RobertaForQuestionAnswering):
"""
This class overrides :class:`~transformers.RobertaForQuestionAnswering`. Please check the superclass for the
This class overrides [`RobertaForQuestionAnswering`]. Please check the superclass for the
appropriate documentation alongside usage examples.
"""
......
src/transformers/models/xlnet/modeling_tf_xlnet.py
View file @ 7af80f66
......@@ -826,26 +826,25 @@ class TFXLNetPreTrainedModel(TFPreTrainedModel):
@dataclass
class TFXLNetModelOutput(ModelOutput):
"""
Output type of :class:`~transformers.TFXLNetModel`.
Output type of [`TFXLNetModel`].
Args:
last_hidden_state (:obj:`tf.Tensor` of shape :obj:`(batch_size, num_predict, hidden_size)`):
last_hidden_state (`tf.Tensor` of shape `(batch_size, num_predict, hidden_size)`):
Sequence of hidden-states at the last layer of the model.
``num_predict`` corresponds to ``target_mapping.shape[1]``. If ``target_mapping`` is ``None``, then
``num_predict`` corresponds to ``sequence_length``.
mems (:obj:`List[tf.Tensor]` of length :obj:`config.n_layers`):
Contains pre-computed hidden-states. Can be used (see :obj:`mems` input) to speed up sequential decoding.
The token ids which have their past given to this model should not be passed as :obj:`input_ids` as they
`num_predict` corresponds to `target_mapping.shape[1]`. If `target_mapping` is `None`, then
`num_predict` corresponds to `sequence_length`.
mems (`List[tf.Tensor]` of length `config.n_layers`):
Contains pre-computed hidden-states. Can be used (see `mems` input) to speed up sequential decoding.
The token ids which have their past given to this model should not be passed as `input_ids` as they
have already been computed.
hidden_states (:obj:`tuple(tf.Tensor)`, `optional`, returned when ``output_hidden_states=True`` is passed or when ``config.output_hidden_states=True``):
Tuple of :obj:`tf.Tensor` (one for the output of the embeddings + one for the output of each layer) of
shape :obj:`(batch_size, sequence_length, hidden_size)`.
hidden_states (`tuple(tf.Tensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `tf.Tensor` (one for the output of the embeddings + one for the output of each layer) of
shape `(batch_size, sequence_length, hidden_size)`.
Hidden-states of the model at the output of each layer plus the initial embedding outputs.
attentions (:obj:`tuple(tf.Tensor)`, `optional`, returned when ``output_attentions=True`` is passed or when ``config.output_attentions=True``):
Tuple of :obj:`tf.Tensor` (one for each layer) of shape :obj:`(batch_size, num_heads, sequence_length,
sequence_length)`.
attentions (`tuple(tf.Tensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `tf.Tensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length, sequence_length)`.
Attentions weights after the attention softmax, used to compute the weighted average in the self-attention
heads.
......@@ -860,28 +859,27 @@ class TFXLNetModelOutput(ModelOutput):
@dataclass
class TFXLNetLMHeadModelOutput(ModelOutput):
"""
Output type of :class:`~transformers.TFXLNetLMHeadModel`.
Output type of [`TFXLNetLMHeadModel`].
Args:
loss (:obj:`tf.Tensor` of shape `(1,)`, `optional`, returned when ``labels`` is provided)
loss (`tf.Tensor` of shape *(1,)*, *optional*, returned when `labels` is provided)
Language modeling loss (for next-token prediction).
logits (:obj:`tf.Tensor` of shape :obj:`(batch_size, num_predict, config.vocab_size)`):
logits (`tf.Tensor` of shape `(batch_size, num_predict, config.vocab_size)`):
Prediction scores of the language modeling head (scores for each vocabulary token before SoftMax).
``num_predict`` corresponds to ``target_mapping.shape[1]``. If ``target_mapping`` is ``None``, then
``num_predict`` corresponds to ``sequence_length``.
mems (:obj:`List[tf.Tensor]` of length :obj:`config.n_layers`):
Contains pre-computed hidden-states. Can be used (see :obj:`mems` input) to speed up sequential decoding.
The token ids which have their past given to this model should not be passed as :obj:`input_ids` as they
`num_predict` corresponds to `target_mapping.shape[1]`. If `target_mapping` is `None`, then
`num_predict` corresponds to `sequence_length`.
mems (`List[tf.Tensor]` of length `config.n_layers`):
Contains pre-computed hidden-states. Can be used (see `mems` input) to speed up sequential decoding.
The token ids which have their past given to this model should not be passed as `input_ids` as they
have already been computed.
hidden_states (:obj:`tuple(tf.Tensor)`, `optional`, returned when ``output_hidden_states=True`` is passed or when ``config.output_hidden_states=True``):
Tuple of :obj:`tf.Tensor` (one for the output of the embeddings + one for the output of each layer) of
shape :obj:`(batch_size, sequence_length, hidden_size)`.
hidden_states (`tuple(tf.Tensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `tf.Tensor` (one for the output of the embeddings + one for the output of each layer) of
shape `(batch_size, sequence_length, hidden_size)`.
Hidden-states of the model at the output of each layer plus the initial embedding outputs.
attentions (:obj:`tuple(tf.Tensor)`, `optional`, returned when ``output_attentions=True`` is passed or when ``config.output_attentions=True``):
Tuple of :obj:`tf.Tensor` (one for each layer) of shape :obj:`(batch_size, num_heads, sequence_length,
sequence_length)`.
attentions (`tuple(tf.Tensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `tf.Tensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length, sequence_length)`.
Attentions weights after the attention softmax, used to compute the weighted average in the self-attention
heads.
......@@ -897,25 +895,24 @@ class TFXLNetLMHeadModelOutput(ModelOutput):
@dataclass
class TFXLNetForSequenceClassificationOutput(ModelOutput):
"""
Output type of :class:`~transformers.TFXLNetForSequenceClassification`.
Output type of [`TFXLNetForSequenceClassification`].
Args:
loss (:obj:`tf.Tensor` of shape :obj:`(1,)`, `optional`, returned when :obj:`label` is provided):
loss (`tf.Tensor` of shape `(1,)`, *optional*, returned when `label` is provided):
Classification (or regression if config.num_labels==1) loss.
logits (:obj:`tf.Tensor` of shape :obj:`(batch_size, config.num_labels)`):
logits (`tf.Tensor` of shape `(batch_size, config.num_labels)`):
Classification (or regression if config.num_labels==1) scores (before SoftMax).
mems (:obj:`List[tf.Tensor]` of length :obj:`config.n_layers`):
Contains pre-computed hidden-states. Can be used (see :obj:`mems` input) to speed up sequential decoding.
The token ids which have their past given to this model should not be passed as :obj:`input_ids` as they
mems (`List[tf.Tensor]` of length `config.n_layers`):
Contains pre-computed hidden-states. Can be used (see `mems` input) to speed up sequential decoding.
The token ids which have their past given to this model should not be passed as `input_ids` as they
have already been computed.
hidden_states (:obj:`tuple(tf.Tensor)`, `optional`, returned when ``output_hidden_states=True`` is passed or when ``config.output_hidden_states=True``):
Tuple of :obj:`tf.Tensor` (one for the output of the embeddings + one for the output of each layer) of
shape :obj:`(batch_size, sequence_length, hidden_size)`.
hidden_states (`tuple(tf.Tensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `tf.Tensor` (one for the output of the embeddings + one for the output of each layer) of
shape `(batch_size, sequence_length, hidden_size)`.
Hidden-states of the model at the output of each layer plus the initial embedding outputs.
attentions (:obj:`tuple(tf.Tensor)`, `optional`, returned when ``output_attentions=True`` is passed or when ``config.output_attentions=True``):
Tuple of :obj:`tf.Tensor` (one for each layer) of shape :obj:`(batch_size, num_heads, sequence_length,
sequence_length)`.
attentions (`tuple(tf.Tensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `tf.Tensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length, sequence_length)`.
Attentions weights after the attention softmax, used to compute the weighted average in the self-attention
heads.
......@@ -931,25 +928,24 @@ class TFXLNetForSequenceClassificationOutput(ModelOutput):
@dataclass
class TFXLNetForTokenClassificationOutput(ModelOutput):
"""
Output type of :class:`~transformers.TFXLNetForTokenClassificationOutput`.
Output type of [`TFXLNetForTokenClassificationOutput`].
Args:
loss (:obj:`tf.Tensor` of shape :obj:`(1,)`, `optional`, returned when ``labels`` is provided) :
loss (`tf.Tensor` of shape `(1,)`, *optional*, returned when `labels` is provided) :
Classification loss.
logits (:obj:`tf.Tensor` of shape :obj:`(batch_size, sequence_length, config.num_labels)`):
logits (`tf.Tensor` of shape `(batch_size, sequence_length, config.num_labels)`):
Classification scores (before SoftMax).
mems (:obj:`List[tf.Tensor]` of length :obj:`config.n_layers`):
Contains pre-computed hidden-states. Can be used (see :obj:`mems` input) to speed up sequential decoding.
The token ids which have their past given to this model should not be passed as :obj:`input_ids` as they
mems (`List[tf.Tensor]` of length `config.n_layers`):
Contains pre-computed hidden-states. Can be used (see `mems` input) to speed up sequential decoding.
The token ids which have their past given to this model should not be passed as `input_ids` as they
have already been computed.
hidden_states (:obj:`tuple(tf.Tensor)`, `optional`, returned when ``output_hidden_states=True`` is passed or when ``config.output_hidden_states=True``):
Tuple of :obj:`tf.Tensor` (one for the output of the embeddings + one for the output of each layer) of
shape :obj:`(batch_size, sequence_length, hidden_size)`.
hidden_states (`tuple(tf.Tensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `tf.Tensor` (one for the output of the embeddings + one for the output of each layer) of
shape `(batch_size, sequence_length, hidden_size)`.
Hidden-states of the model at the output of each layer plus the initial embedding outputs.
attentions (:obj:`tuple(tf.Tensor)`, `optional`, returned when ``output_attentions=True`` is passed or when ``config.output_attentions=True``):
Tuple of :obj:`tf.Tensor` (one for each layer) of shape :obj:`(batch_size, num_heads, sequence_length,
sequence_length)`.
attentions (`tuple(tf.Tensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `tf.Tensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length, sequence_length)`.
Attentions weights after the attention softmax, used to compute the weighted average in the self-attention
heads.
......@@ -965,27 +961,26 @@ class TFXLNetForTokenClassificationOutput(ModelOutput):
@dataclass
class TFXLNetForMultipleChoiceOutput(ModelOutput):
"""
Output type of :class:`~transformers.TFXLNetForMultipleChoice`.
Output type of [`TFXLNetForMultipleChoice`].
Args:
loss (:obj:`tf.Tensor` of shape `(1,)`, `optional`, returned when :obj:`labels` is provided):
loss (`tf.Tensor` of shape *(1,)*, *optional*, returned when `labels` is provided):
Classification loss.
logits (:obj:`tf.Tensor` of shape :obj:`(batch_size, num_choices)`):
`num_choices` is the second dimension of the input tensors. (see `input_ids` above).
logits (`tf.Tensor` of shape `(batch_size, num_choices)`):
*num_choices* is the second dimension of the input tensors. (see *input_ids* above).
Classification scores (before SoftMax).
mems (:obj:`List[tf.Tensor]` of length :obj:`config.n_layers`):
Contains pre-computed hidden-states. Can be used (see :obj:`mems` input) to speed up sequential decoding.
The token ids which have their past given to this model should not be passed as :obj:`input_ids` as they
mems (`List[tf.Tensor]` of length `config.n_layers`):
Contains pre-computed hidden-states. Can be used (see `mems` input) to speed up sequential decoding.
The token ids which have their past given to this model should not be passed as `input_ids` as they
have already been computed.
hidden_states (:obj:`tuple(tf.Tensor)`, `optional`, returned when ``output_hidden_states=True`` is passed or when ``config.output_hidden_states=True``):
Tuple of :obj:`tf.Tensor` (one for the output of the embeddings + one for the output of each layer) of
shape :obj:`(batch_size, sequence_length, hidden_size)`.
hidden_states (`tuple(tf.Tensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `tf.Tensor` (one for the output of the embeddings + one for the output of each layer) of
shape `(batch_size, sequence_length, hidden_size)`.
Hidden-states of the model at the output of each layer plus the initial embedding outputs.
attentions (:obj:`tuple(tf.Tensor)`, `optional`, returned when ``output_attentions=True`` is passed or when ``config.output_attentions=True``):
Tuple of :obj:`tf.Tensor` (one for each layer) of shape :obj:`(batch_size, num_heads, sequence_length,
sequence_length)`.
attentions (`tuple(tf.Tensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `tf.Tensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length, sequence_length)`.
Attentions weights after the attention softmax, used to compute the weighted average in the self-attention
heads.
......@@ -1001,27 +996,26 @@ class TFXLNetForMultipleChoiceOutput(ModelOutput):
@dataclass
class TFXLNetForQuestionAnsweringSimpleOutput(ModelOutput):
"""
Output type of :class:`~transformers.TFXLNetForQuestionAnsweringSimple`.
Output type of [`TFXLNetForQuestionAnsweringSimple`].
Args:
loss (:obj:`tf.Tensor` of shape :obj:`(1,)`, `optional`, returned when :obj:`labels` is provided):
loss (`tf.Tensor` of shape `(1,)`, *optional*, returned when `labels` is provided):
Total span extraction loss is the sum of a Cross-Entropy for the start and end positions.
start_logits (:obj:`tf.Tensor` of shape :obj:`(batch_size, sequence_length,)`):
start_logits (`tf.Tensor` of shape `(batch_size, sequence_length,)`):
Span-start scores (before SoftMax).
end_logits (:obj:`tf.Tensor` of shape :obj:`(batch_size, sequence_length,)`):
end_logits (`tf.Tensor` of shape `(batch_size, sequence_length,)`):
Span-end scores (before SoftMax).
mems (:obj:`List[tf.Tensor]` of length :obj:`config.n_layers`):
Contains pre-computed hidden-states. Can be used (see :obj:`mems` input) to speed up sequential decoding.
The token ids which have their past given to this model should not be passed as :obj:`input_ids` as they
mems (`List[tf.Tensor]` of length `config.n_layers`):
Contains pre-computed hidden-states. Can be used (see `mems` input) to speed up sequential decoding.
The token ids which have their past given to this model should not be passed as `input_ids` as they
have already been computed.
hidden_states (:obj:`tuple(tf.Tensor)`, `optional`, returned when ``output_hidden_states=True`` is passed or when ``config.output_hidden_states=True``):
Tuple of :obj:`tf.Tensor` (one for the output of the embeddings + one for the output of each layer) of
shape :obj:`(batch_size, sequence_length, hidden_size)`.
hidden_states (`tuple(tf.Tensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `tf.Tensor` (one for the output of the embeddings + one for the output of each layer) of
shape `(batch_size, sequence_length, hidden_size)`.
Hidden-states of the model at the output of each layer plus the initial embedding outputs.
attentions (:obj:`tuple(tf.Tensor)`, `optional`, returned when ``output_attentions=True`` is passed or when ``config.output_attentions=True``):
Tuple of :obj:`tf.Tensor` (one for each layer) of shape :obj:`(batch_size, num_heads, sequence_length,
sequence_length)`.
attentions (`tuple(tf.Tensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `tf.Tensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length, sequence_length)`.
Attentions weights after the attention softmax, used to compute the weighted average in the self-attention
heads.
......@@ -1037,114 +1031,115 @@ class TFXLNetForQuestionAnsweringSimpleOutput(ModelOutput):
XLNET_START_DOCSTRING = r"""
This model inherits from :class:`~transformers.TFPreTrainedModel`. Check the superclass documentation for the
This model inherits from [`TFPreTrainedModel`]. Check the superclass documentation for the
generic methods the library implements for all its model (such as downloading or saving, resizing the input
embeddings, pruning heads etc.)
This model is also a `tf.keras.Model <https://www.tensorflow.org/api_docs/python/tf/keras/Model>`__ subclass. Use
This model is also a [tf.keras.Model](https://www.tensorflow.org/api_docs/python/tf/keras/Model) subclass. Use
it as a regular TF 2.0 Keras Model and refer to the TF 2.0 documentation for all matter related to general usage
and behavior.
.. note::
<Tip>
TF 2.0 models accepts two formats as inputs:
TF 2.0 models accepts two formats as inputs:
- having all inputs as keyword arguments (like PyTorch models), or
- having all inputs as a list, tuple or dict in the first positional arguments.
- having all inputs as keyword arguments (like PyTorch models), or
- having all inputs as a list, tuple or dict in the first positional arguments.
This second option is useful when using :meth:`tf.keras.Model.fit` method which currently requires having all
the tensors in the first argument of the model call function: :obj:`model(inputs)`.
This second option is useful when using [`tf.keras.Model.fit`] method which currently requires having all
the tensors in the first argument of the model call function: `model(inputs)`.
If you choose this second option, there are three possibilities you can use to gather all the input Tensors in
the first positional argument :
If you choose this second option, there are three possibilities you can use to gather all the input Tensors in
the first positional argument :
- a single Tensor with :obj:`input_ids` only and nothing else: :obj:`model(inputs_ids)`
- a list of varying length with one or several input Tensors IN THE ORDER given in the docstring:
:obj:`model([input_ids, attention_mask])` or :obj:`model([input_ids, attention_mask, token_type_ids])`
- a dictionary with one or several input Tensors associated to the input names given in the docstring:
:obj:`model({"input_ids": input_ids, "token_type_ids": token_type_ids})`
- a single Tensor with `input_ids` only and nothing else: `model(inputs_ids)`
- a list of varying length with one or several input Tensors IN THE ORDER given in the docstring:
`model([input_ids, attention_mask])` or `model([input_ids, attention_mask, token_type_ids])`
- a dictionary with one or several input Tensors associated to the input names given in the docstring:
`model({"input_ids": input_ids, "token_type_ids": token_type_ids})`
</Tip>
Parameters:
config (:class:`~transformers.XLNetConfig`): Model configuration class with all the parameters of the model.
config ([`XLNetConfig`]): Model configuration class with all the parameters of the model.
Initializing with a config file does not load the weights associated with the model, only the
configuration. Check out the :meth:`~transformers.PreTrainedModel.from_pretrained` method to load the model
configuration. Check out the [`~PreTrainedModel.from_pretrained`] method to load the model
weights.
"""
XLNET_INPUTS_DOCSTRING = r"""
Args:
input_ids (:obj:`Numpy array` or :obj:`tf.Tensor` of shape :obj:`({0})`):
input_ids (`Numpy array` or `tf.Tensor` of shape `({0})`):
Indices of input sequence tokens in the vocabulary.
Indices can be obtained using :class:`~transformers.BertTokenizer`. See
:func:`transformers.PreTrainedTokenizer.__call__` and :func:`transformers.PreTrainedTokenizer.encode` for
Indices can be obtained using [`BertTokenizer`]. See
[`PreTrainedTokenizer.__call__`] and [`PreTrainedTokenizer.encode`] for
details.
`What are input IDs? <../glossary.html#input-ids>`__
attention_mask (:obj:`Numpy array` or :obj:`tf.Tensor` of shape :obj:`({0})`, `optional`):
Mask to avoid performing attention on padding token indices. Mask values selected in ``[0, 1]``:
[What are input IDs?](../glossary#input-ids)
attention_mask (`Numpy array` or `tf.Tensor` of shape `({0})`, *optional*):
Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
`What are attention masks? <../glossary.html#attention-mask>`__
mems (:obj:`List[torch.FloatTensor]` of length :obj:`config.n_layers`):
Contains pre-computed hidden-states (see :obj:`mems` output below) . Can be used to speed up sequential
decoding. The token ids which have their past given to this model should not be passed as :obj:`input_ids`
[What are attention masks?](../glossary#attention-mask)
mems (`List[torch.FloatTensor]` of length `config.n_layers`):
Contains pre-computed hidden-states (see `mems` output below) . Can be used to speed up sequential
decoding. The token ids which have their past given to this model should not be passed as `input_ids`
as they have already been computed.
:obj::obj:`use_mems` has to be set to :obj:`True` to make use of :obj:`mems`.
perm_mask (:obj:`tf.Tensor` or :obj:`Numpy array` of shape :obj:`(batch_size, sequence_length, sequence_length)`, `optional`):
Mask to indicate the attention pattern for each input token with values selected in ``[0, 1]``:
:obj:`use_mems` has to be set to `True` to make use of `mems`.
perm_mask (`tf.Tensor` or `Numpy array` of shape `(batch_size, sequence_length, sequence_length)`, *optional*):
Mask to indicate the attention pattern for each input token with values selected in `[0, 1]`:
- if ``perm_mask[k, i, j] = 0``, i attend to j in batch k;
- if ``perm_mask[k, i, j] = 1``, i does not attend to j in batch k.
- if `perm_mask[k, i, j] = 0`, i attend to j in batch k;
- if `perm_mask[k, i, j] = 1`, i does not attend to j in batch k.
If not set, each token attends to all the others (full bidirectional attention). Only used during
pretraining (to define factorization order) or for sequential decoding (generation).
target_mapping (:obj:`tf.Tensor` or :obj:`Numpy array` of shape :obj:`(batch_size, num_predict, sequence_length)`, `optional`):
Mask to indicate the output tokens to use. If ``target_mapping[k, i, j] = 1``, the i-th predict in batch k
target_mapping (`tf.Tensor` or `Numpy array` of shape `(batch_size, num_predict, sequence_length)`, *optional*):
Mask to indicate the output tokens to use. If `target_mapping[k, i, j] = 1`, the i-th predict in batch k
is on the j-th token.
token_type_ids (:obj:`Numpy array` or :obj:`tf.Tensor` of shape :obj:`({0})`, `optional`):
Segment token indices to indicate first and second portions of the inputs. Indices are selected in ``[0,
1]``:
token_type_ids (`Numpy array` or `tf.Tensor` of shape `({0})`, *optional*):
Segment token indices to indicate first and second portions of the inputs. Indices are selected in `[0, 1]`:
- 0 corresponds to a `sentence A` token,
- 1 corresponds to a `sentence B` token.
- 0 corresponds to a *sentence A* token,
- 1 corresponds to a *sentence B* token.
`What are token type IDs? <../glossary.html#token-type-ids>`__
input_mask (:obj:`tf.Tensor` or :obj:`Numpy array` of shape :obj:`({0})`, `optional`):
Mask to avoid performing attention on padding token indices. Negative of :obj:`attention_mask`, i.e. with 0
[What are token type IDs?](../glossary#token-type-ids)
input_mask (`tf.Tensor` or `Numpy array` of shape `({0})`, *optional*):
Mask to avoid performing attention on padding token indices. Negative of `attention_mask`, i.e. with 0
for real tokens and 1 for padding which is kept for compatibility with the original code base.
Mask values selected in ``[0, 1]``:
Mask values selected in `[0, 1]`:
- 1 for tokens that are **masked**,
- 0 for tokens that are **not masked**.
You can only uses one of :obj:`input_mask` and :obj:`attention_mask`.
head_mask (:obj:`Numpy array` or :obj:`tf.Tensor` of shape :obj:`(num_heads,)` or :obj:`(num_layers, num_heads)`, `optional`):
Mask to nullify selected heads of the self-attention modules. Mask values selected in ``[0, 1]``:
You can only uses one of `input_mask` and `attention_mask`.
head_mask (`Numpy array` or `tf.Tensor` of shape `(num_heads,)` or `(num_layers, num_heads)`, *optional*):
Mask to nullify selected heads of the self-attention modules. Mask values selected in `[0, 1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
inputs_embeds (:obj:`tf.Tensor` of shape :obj:`({0}, hidden_size)`, `optional`):
Optionally, instead of passing :obj:`input_ids` you can choose to directly pass an embedded representation.
This is useful if you want more control over how to convert :obj:`input_ids` indices into associated
inputs_embeds (`tf.Tensor` of shape `({0}, hidden_size)`, *optional*):
Optionally, instead of passing `input_ids` you can choose to directly pass an embedded representation.
This is useful if you want more control over how to convert `input_ids` indices into associated
vectors than the model's internal embedding lookup matrix.
output_attentions (:obj:`bool`, `optional`):
Whether or not to return the attentions tensors of all attention layers. See ``attentions`` under returned
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under returned
tensors for more detail. This argument can be used only in eager mode, in graph mode the value in the
config will be used instead.
output_hidden_states (:obj:`bool`, `optional`):
Whether or not to return the hidden states of all layers. See ``hidden_states`` under returned tensors for
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for
more detail. This argument can be used only in eager mode, in graph mode the value in the config will be
used instead.
return_dict (:obj:`bool`, `optional`):
Whether or not to return a :class:`~transformers.file_utils.ModelOutput` instead of a plain tuple. This
return_dict (`bool`, *optional*):
Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple. This
argument can be used in eager mode, in graph mode the value will always be set to True.
training (:obj:`bool`, `optional`, defaults to :obj:`False`):
training (`bool`, *optional*, defaults to `False`):
Whether or not to use the model in training mode (some modules like dropout modules have different
behaviors between training and evaluation).
"""
......@@ -1313,35 +1308,34 @@ class TFXLNetLMHeadModel(TFXLNetPreTrainedModel, TFCausalLanguageModelingLoss):
**kwargs,
):
r"""
labels (:obj:`tf.Tensor` of shape :obj:`(batch_size, sequence_length)`, `optional`):
Labels for computing the cross entropy classification loss. Indices should be in ``[0, ...,
config.vocab_size - 1]``.
labels (`tf.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
Labels for computing the cross entropy classification loss. Indices should be in `[0, ..., config.vocab_size - 1]`.
Return:
Examples::
>>> import tensorflow as tf
>>> import numpy as np
>>> from transformers import XLNetTokenizer, TFXLNetLMHeadModel
Examples:
>>> tokenizer = XLNetTokenizer.from_pretrained('xlnet-large-cased')
>>> model = TFXLNetLMHeadModel.from_pretrained('xlnet-large-cased')
```python
>>> import tensorflow as tf
>>> import numpy as np
>>> from transformers import XLNetTokenizer, TFXLNetLMHeadModel
>>> # We show how to setup inputs to predict a next token using a bi-directional context.
>>> input_ids = tf.constant(tokenizer.encode("Hello, my dog is very <mask>", add_special_tokens=True))[None, :] # We will predict the masked token
>>> tokenizer = XLNetTokenizer.from_pretrained('xlnet-large-cased')
>>> model = TFXLNetLMHeadModel.from_pretrained('xlnet-large-cased')
>>> perm_mask = np.zeros((1, input_ids.shape[1], input_ids.shape[1]))
>>> perm_mask[:, :, -1] = 1.0 # Previous tokens don't see last token
>>> # We show how to setup inputs to predict a next token using a bi-directional context.
>>> input_ids = tf.constant(tokenizer.encode("Hello, my dog is very <mask>", add_special_tokens=True))[None, :] # We will predict the masked token
>>> target_mapping = np.zeros((1, 1, input_ids.shape[1])) # Shape [1, 1, seq_length] => let's predict one token
>>> target_mapping[0, 0, -1] = 1.0 # Our first (and only) prediction will be the last token of the sequence (the masked token)
>>> perm_mask = np.zeros((1, input_ids.shape[1], input_ids.shape[1]))
>>> perm_mask[:, :, -1] = 1.0 # Previous tokens don't see last token
>>> outputs = model(input_ids, perm_mask=tf.constant(perm_mask, dtype=tf.float32), target_mapping=tf.constant(target_mapping, dtype=tf.float32))
>>> target_mapping = np.zeros((1, 1, input_ids.shape[1])) # Shape [1, 1, seq_length] => let's predict one token
>>> target_mapping[0, 0, -1] = 1.0 # Our first (and only) prediction will be the last token of the sequence (the masked token)
>>> next_token_logits = outputs[0] # Output has shape [target_mapping.size(0), target_mapping.size(1), config.vocab_size]
>>> outputs = model(input_ids, perm_mask=tf.constant(perm_mask, dtype=tf.float32), target_mapping=tf.constant(target_mapping, dtype=tf.float32))
"""
>>> next_token_logits = outputs[0] # Output has shape [target_mapping.size(0), target_mapping.size(1), config.vocab_size]
```"""
inputs = input_processing(
func=self.call,
config=self.config,
......@@ -1455,10 +1449,9 @@ class TFXLNetForSequenceClassification(TFXLNetPreTrainedModel, TFSequenceClassif
**kwargs,
):
r"""
labels (:obj:`tf.Tensor` of shape :obj:`(batch_size,)`, `optional`):
Labels for computing the sequence classification/regression loss. Indices should be in ``[0, ...,
config.num_labels - 1]``. If ``config.num_labels == 1`` a regression loss is computed (Mean-Square loss),
If ``config.num_labels > 1`` a classification loss is computed (Cross-Entropy).
labels (`tf.Tensor` of shape `(batch_size,)`, *optional*):
Labels for computing the sequence classification/regression loss. Indices should be in `[0, ..., config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss),
If `config.num_labels > 1` a classification loss is computed (Cross-Entropy).
"""
inputs = input_processing(
func=self.call,
......@@ -1581,10 +1574,9 @@ class TFXLNetForMultipleChoice(TFXLNetPreTrainedModel, TFMultipleChoiceLoss):
**kwargs,
):
r"""
labels (:obj:`tf.Tensor` of shape :obj:`(batch_size,)`, `optional`):
Labels for computing the multiple choice classification loss. Indices should be in ``[0, ...,
num_choices]`` where :obj:`num_choices` is the size of the second dimension of the input tensors. (See
:obj:`input_ids` above)
labels (`tf.Tensor` of shape `(batch_size,)`, *optional*):
Labels for computing the multiple choice classification loss. Indices should be in `[0, ..., num_choices]` where `num_choices` is the size of the second dimension of the input tensors. (See
`input_ids` above)
"""
inputs = input_processing(
......@@ -1730,9 +1722,8 @@ class TFXLNetForTokenClassification(TFXLNetPreTrainedModel, TFTokenClassificatio
**kwargs,
):
r"""
labels (:obj:`tf.Tensor` of shape :obj:`(batch_size, sequence_length)`, `optional`):
Labels for computing the token classification loss. Indices should be in ``[0, ..., config.num_labels -
1]``.
labels (`tf.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
Labels for computing the token classification loss. Indices should be in `[0, ..., config.num_labels - 1]`.
"""
inputs = input_processing(
......@@ -1838,13 +1829,13 @@ class TFXLNetForQuestionAnsweringSimple(TFXLNetPreTrainedModel, TFQuestionAnswer
**kwargs,
):
r"""
start_positions (:obj:`tf.Tensor` of shape :obj:`(batch_size,)`, `optional`):
start_positions (`tf.Tensor` of shape `(batch_size,)`, *optional*):
Labels for position (index) of the start of the labelled span for computing the token classification loss.
Positions are clamped to the length of the sequence (:obj:`sequence_length`). Position outside of the
Positions are clamped to the length of the sequence (`sequence_length`). Position outside of the
sequence are not taken into account for computing the loss.
end_positions (:obj:`tf.Tensor` of shape :obj:`(batch_size,)`, `optional`):
end_positions (`tf.Tensor` of shape `(batch_size,)`, *optional*):
Labels for position (index) of the end of the labelled span for computing the token classification loss.
Positions are clamped to the length of the sequence (:obj:`sequence_length`). Position outside of the
Positions are clamped to the length of the sequence (`sequence_length`). Position outside of the
sequence are not taken into account for computing the loss.
"""
inputs = input_processing(
......
src/transformers/models/xlnet/modeling_xlnet.py
View file @ 7af80f66
......@@ -585,26 +585,25 @@ class XLNetPreTrainedModel(PreTrainedModel):
@dataclass
class XLNetModelOutput(ModelOutput):
"""
Output type of :class:`~transformers.XLNetModel`.
Output type of [`XLNetModel`].
Args:
last_hidden_state (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, num_predict, hidden_size)`):
last_hidden_state (`torch.FloatTensor` of shape `(batch_size, num_predict, hidden_size)`):
Sequence of hidden-states at the last layer of the model.
``num_predict`` corresponds to ``target_mapping.shape[1]``. If ``target_mapping`` is ``None``, then
``num_predict`` corresponds to ``sequence_length``.
mems (:obj:`List[torch.FloatTensor]` of length :obj:`config.n_layers`):
Contains pre-computed hidden-states. Can be used (see :obj:`mems` input) to speed up sequential decoding.
The token ids which have their past given to this model should not be passed as :obj:`input_ids` as they
`num_predict` corresponds to `target_mapping.shape[1]`. If `target_mapping` is `None`, then
`num_predict` corresponds to `sequence_length`.
mems (`List[torch.FloatTensor]` of length `config.n_layers`):
Contains pre-computed hidden-states. Can be used (see `mems` input) to speed up sequential decoding.
The token ids which have their past given to this model should not be passed as `input_ids` as they
have already been computed.
hidden_states (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_hidden_states=True`` is passed or when ``config.output_hidden_states=True``):
Tuple of :obj:`torch.FloatTensor` (one for the output of the embeddings + one for the output of each layer)
of shape :obj:`(batch_size, sequence_length, hidden_size)`.
hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `torch.FloatTensor` (one for the output of the embeddings + one for the output of each layer)
of shape `(batch_size, sequence_length, hidden_size)`.
Hidden-states of the model at the output of each layer plus the initial embedding outputs.
attentions (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_attentions=True`` is passed or when ``config.output_attentions=True``):
Tuple of :obj:`torch.FloatTensor` (one for each layer) of shape :obj:`(batch_size, num_heads,
sequence_length, sequence_length)`.
attentions (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length, sequence_length)`.
Attentions weights after the attention softmax, used to compute the weighted average in the self-attention
heads.
......@@ -619,28 +618,27 @@ class XLNetModelOutput(ModelOutput):
@dataclass
class XLNetLMHeadModelOutput(ModelOutput):
"""
Output type of :class:`~transformers.XLNetLMHeadModel`.
Output type of [`XLNetLMHeadModel`].
Args:
loss (:obj:`torch.FloatTensor` of shape `(1,)`, `optional`, returned when ``labels`` is provided)
loss (`torch.FloatTensor` of shape *(1,)*, *optional*, returned when `labels` is provided)
Language modeling loss (for next-token prediction).
logits (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, num_predict, config.vocab_size)`):
logits (`torch.FloatTensor` of shape `(batch_size, num_predict, config.vocab_size)`):
Prediction scores of the language modeling head (scores for each vocabulary token before SoftMax).
``num_predict`` corresponds to ``target_mapping.shape[1]``. If ``target_mapping`` is ``None``, then
``num_predict`` corresponds to ``sequence_length``.
mems (:obj:`List[torch.FloatTensor]` of length :obj:`config.n_layers`):
Contains pre-computed hidden-states. Can be used (see :obj:`mems` input) to speed up sequential decoding.
The token ids which have their past given to this model should not be passed as :obj:`input_ids` as they
`num_predict` corresponds to `target_mapping.shape[1]`. If `target_mapping` is `None`, then
`num_predict` corresponds to `sequence_length`.
mems (`List[torch.FloatTensor]` of length `config.n_layers`):
Contains pre-computed hidden-states. Can be used (see `mems` input) to speed up sequential decoding.
The token ids which have their past given to this model should not be passed as `input_ids` as they
have already been computed.
hidden_states (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_hidden_states=True`` is passed or when ``config.output_hidden_states=True``):
Tuple of :obj:`torch.FloatTensor` (one for the output of the embeddings + one for the output of each layer)
of shape :obj:`(batch_size, sequence_length, hidden_size)`.
hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `torch.FloatTensor` (one for the output of the embeddings + one for the output of each layer)
of shape `(batch_size, sequence_length, hidden_size)`.
Hidden-states of the model at the output of each layer plus the initial embedding outputs.
attentions (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_attentions=True`` is passed or when ``config.output_attentions=True``):
Tuple of :obj:`torch.FloatTensor` (one for each layer) of shape :obj:`(batch_size, num_heads,
sequence_length, sequence_length)`.
attentions (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length, sequence_length)`.
Attentions weights after the attention softmax, used to compute the weighted average in the self-attention
heads.
......@@ -656,25 +654,24 @@ class XLNetLMHeadModelOutput(ModelOutput):
@dataclass
class XLNetForSequenceClassificationOutput(ModelOutput):
"""
Output type of :class:`~transformers.XLNetForSequenceClassification`.
Output type of [`XLNetForSequenceClassification`].
Args:
loss (:obj:`torch.FloatTensor` of shape :obj:`(1,)`, `optional`, returned when :obj:`label` is provided):
loss (`torch.FloatTensor` of shape `(1,)`, *optional*, returned when `label` is provided):
Classification (or regression if config.num_labels==1) loss.
logits (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, config.num_labels)`):
logits (`torch.FloatTensor` of shape `(batch_size, config.num_labels)`):
Classification (or regression if config.num_labels==1) scores (before SoftMax).
mems (:obj:`List[torch.FloatTensor]` of length :obj:`config.n_layers`):
Contains pre-computed hidden-states. Can be used (see :obj:`mems` input) to speed up sequential decoding.
The token ids which have their past given to this model should not be passed as :obj:`input_ids` as they
mems (`List[torch.FloatTensor]` of length `config.n_layers`):
Contains pre-computed hidden-states. Can be used (see `mems` input) to speed up sequential decoding.
The token ids which have their past given to this model should not be passed as `input_ids` as they
have already been computed.
hidden_states (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_hidden_states=True`` is passed or when ``config.output_hidden_states=True``):
Tuple of :obj:`torch.FloatTensor` (one for the output of the embeddings + one for the output of each layer)
of shape :obj:`(batch_size, sequence_length, hidden_size)`.
hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `torch.FloatTensor` (one for the output of the embeddings + one for the output of each layer)
of shape `(batch_size, sequence_length, hidden_size)`.
Hidden-states of the model at the output of each layer plus the initial embedding outputs.
attentions (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_attentions=True`` is passed or when ``config.output_attentions=True``):
Tuple of :obj:`torch.FloatTensor` (one for each layer) of shape :obj:`(batch_size, num_heads,
sequence_length, sequence_length)`.
attentions (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length, sequence_length)`.
Attentions weights after the attention softmax, used to compute the weighted average in the self-attention
heads.
......@@ -690,25 +687,24 @@ class XLNetForSequenceClassificationOutput(ModelOutput):
@dataclass
class XLNetForTokenClassificationOutput(ModelOutput):
"""
Output type of :class:`~transformers.XLNetForTokenClassificationOutput`.
Output type of [`XLNetForTokenClassificationOutput`].
Args:
loss (:obj:`torch.FloatTensor` of shape :obj:`(1,)`, `optional`, returned when ``labels`` is provided) :
loss (`torch.FloatTensor` of shape `(1,)`, *optional*, returned when `labels` is provided) :
Classification loss.
logits (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length, config.num_labels)`):
logits (`torch.FloatTensor` of shape `(batch_size, sequence_length, config.num_labels)`):
Classification scores (before SoftMax).
mems (:obj:`List[torch.FloatTensor]` of length :obj:`config.n_layers`):
Contains pre-computed hidden-states. Can be used (see :obj:`mems` input) to speed up sequential decoding.
The token ids which have their past given to this model should not be passed as :obj:`input_ids` as they
mems (`List[torch.FloatTensor]` of length `config.n_layers`):
Contains pre-computed hidden-states. Can be used (see `mems` input) to speed up sequential decoding.
The token ids which have their past given to this model should not be passed as `input_ids` as they
have already been computed.
hidden_states (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_hidden_states=True`` is passed or when ``config.output_hidden_states=True``):
Tuple of :obj:`torch.FloatTensor` (one for the output of the embeddings + one for the output of each layer)
of shape :obj:`(batch_size, sequence_length, hidden_size)`.
hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `torch.FloatTensor` (one for the output of the embeddings + one for the output of each layer)
of shape `(batch_size, sequence_length, hidden_size)`.
Hidden-states of the model at the output of each layer plus the initial embedding outputs.
attentions (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_attentions=True`` is passed or when ``config.output_attentions=True``):
Tuple of :obj:`torch.FloatTensor` (one for each layer) of shape :obj:`(batch_size, num_heads,
sequence_length, sequence_length)`.
attentions (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length, sequence_length)`.
Attentions weights after the attention softmax, used to compute the weighted average in the self-attention
heads.
......@@ -724,27 +720,26 @@ class XLNetForTokenClassificationOutput(ModelOutput):
@dataclass
class XLNetForMultipleChoiceOutput(ModelOutput):
"""
Output type of :class:`~transformers.XLNetForMultipleChoice`.
Output type of [`XLNetForMultipleChoice`].
Args:
loss (:obj:`torch.FloatTensor` of shape `(1,)`, `optional`, returned when :obj:`labels` is provided):
loss (`torch.FloatTensor` of shape *(1,)*, *optional*, returned when `labels` is provided):
Classification loss.
logits (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, num_choices)`):
`num_choices` is the second dimension of the input tensors. (see `input_ids` above).
logits (`torch.FloatTensor` of shape `(batch_size, num_choices)`):
*num_choices* is the second dimension of the input tensors. (see *input_ids* above).
Classification scores (before SoftMax).
mems (:obj:`List[torch.FloatTensor]` of length :obj:`config.n_layers`):
Contains pre-computed hidden-states. Can be used (see :obj:`mems` input) to speed up sequential decoding.
The token ids which have their past given to this model should not be passed as :obj:`input_ids` as they
mems (`List[torch.FloatTensor]` of length `config.n_layers`):
Contains pre-computed hidden-states. Can be used (see `mems` input) to speed up sequential decoding.
The token ids which have their past given to this model should not be passed as `input_ids` as they
have already been computed.
hidden_states (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_hidden_states=True`` is passed or when ``config.output_hidden_states=True``):
Tuple of :obj:`torch.FloatTensor` (one for the output of the embeddings + one for the output of each layer)
of shape :obj:`(batch_size, sequence_length, hidden_size)`.
hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `torch.FloatTensor` (one for the output of the embeddings + one for the output of each layer)
of shape `(batch_size, sequence_length, hidden_size)`.
Hidden-states of the model at the output of each layer plus the initial embedding outputs.
attentions (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_attentions=True`` is passed or when ``config.output_attentions=True``):
Tuple of :obj:`torch.FloatTensor` (one for each layer) of shape :obj:`(batch_size, num_heads,
sequence_length, sequence_length)`.
attentions (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length, sequence_length)`.
Attentions weights after the attention softmax, used to compute the weighted average in the self-attention
heads.
......@@ -760,27 +755,26 @@ class XLNetForMultipleChoiceOutput(ModelOutput):
@dataclass
class XLNetForQuestionAnsweringSimpleOutput(ModelOutput):
"""
Output type of :class:`~transformers.XLNetForQuestionAnsweringSimple`.
Output type of [`XLNetForQuestionAnsweringSimple`].
Args:
loss (:obj:`torch.FloatTensor` of shape :obj:`(1,)`, `optional`, returned when :obj:`labels` is provided):
loss (`torch.FloatTensor` of shape `(1,)`, *optional*, returned when `labels` is provided):
Total span extraction loss is the sum of a Cross-Entropy for the start and end positions.
start_logits (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length,)`):
start_logits (`torch.FloatTensor` of shape `(batch_size, sequence_length,)`):
Span-start scores (before SoftMax).
end_logits (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length,)`):
end_logits (`torch.FloatTensor` of shape `(batch_size, sequence_length,)`):
Span-end scores (before SoftMax).
mems (:obj:`List[torch.FloatTensor]` of length :obj:`config.n_layers`):
Contains pre-computed hidden-states. Can be used (see :obj:`mems` input) to speed up sequential decoding.
The token ids which have their past given to this model should not be passed as :obj:`input_ids` as they
mems (`List[torch.FloatTensor]` of length `config.n_layers`):
Contains pre-computed hidden-states. Can be used (see `mems` input) to speed up sequential decoding.
The token ids which have their past given to this model should not be passed as `input_ids` as they
have already been computed.
hidden_states (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_hidden_states=True`` is passed or when ``config.output_hidden_states=True``):
Tuple of :obj:`torch.FloatTensor` (one for the output of the embeddings + one for the output of each layer)
of shape :obj:`(batch_size, sequence_length, hidden_size)`.
hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `torch.FloatTensor` (one for the output of the embeddings + one for the output of each layer)
of shape `(batch_size, sequence_length, hidden_size)`.
Hidden-states of the model at the output of each layer plus the initial embedding outputs.
attentions (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_attentions=True`` is passed or when ``config.output_attentions=True``):
Tuple of :obj:`torch.FloatTensor` (one for each layer) of shape :obj:`(batch_size, num_heads,
sequence_length, sequence_length)`.
attentions (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length, sequence_length)`.
Attentions weights after the attention softmax, used to compute the weighted average in the self-attention
heads.
......@@ -797,35 +791,34 @@ class XLNetForQuestionAnsweringSimpleOutput(ModelOutput):
@dataclass
class XLNetForQuestionAnsweringOutput(ModelOutput):
"""
Output type of :class:`~transformers.XLNetForQuestionAnswering`.
Output type of [`XLNetForQuestionAnswering`].
Args:
loss (:obj:`torch.FloatTensor` of shape :obj:`(1,)`, `optional`, returned if both :obj:`start_positions` and :obj:`end_positions` are provided):
loss (`torch.FloatTensor` of shape `(1,)`, *optional*, returned if both `start_positions` and `end_positions` are provided):
Classification loss as the sum of start token, end token (and is_impossible if provided) classification
losses.
start_top_log_probs (``torch.FloatTensor`` of shape ``(batch_size, config.start_n_top)``, `optional`, returned if ``start_positions`` or ``end_positions`` is not provided):
start_top_log_probs (`torch.FloatTensor` of shape `(batch_size, config.start_n_top)`, *optional*, returned if `start_positions` or `end_positions` is not provided):
Log probabilities for the top config.start_n_top start token possibilities (beam-search).
start_top_index (``torch.LongTensor`` of shape ``(batch_size, config.start_n_top)``, `optional`, returned if ``start_positions`` or ``end_positions`` is not provided):
start_top_index (`torch.LongTensor` of shape `(batch_size, config.start_n_top)`, *optional*, returned if `start_positions` or `end_positions` is not provided):
Indices for the top config.start_n_top start token possibilities (beam-search).
end_top_log_probs (``torch.FloatTensor`` of shape ``(batch_size, config.start_n_top * config.end_n_top)``, `optional`, returned if ``start_positions`` or ``end_positions`` is not provided):
Log probabilities for the top ``config.start_n_top * config.end_n_top`` end token possibilities
end_top_log_probs (`torch.FloatTensor` of shape `(batch_size, config.start_n_top * config.end_n_top)`, *optional*, returned if `start_positions` or `end_positions` is not provided):
Log probabilities for the top `config.start_n_top * config.end_n_top` end token possibilities
(beam-search).
end_top_index (``torch.LongTensor`` of shape ``(batch_size, config.start_n_top * config.end_n_top)``, `optional`, returned if ``start_positions`` or ``end_positions`` is not provided):
Indices for the top ``config.start_n_top * config.end_n_top`` end token possibilities (beam-search).
cls_logits (``torch.FloatTensor`` of shape ``(batch_size,)``, `optional`, returned if ``start_positions`` or ``end_positions`` is not provided):
Log probabilities for the ``is_impossible`` label of the answers.
mems (:obj:`List[torch.FloatTensor]` of length :obj:`config.n_layers`):
Contains pre-computed hidden-states. Can be used (see :obj:`mems` input) to speed up sequential decoding.
The token ids which have their past given to this model should not be passed as :obj:`input_ids` as they
end_top_index (`torch.LongTensor` of shape `(batch_size, config.start_n_top * config.end_n_top)`, *optional*, returned if `start_positions` or `end_positions` is not provided):
Indices for the top `config.start_n_top * config.end_n_top` end token possibilities (beam-search).
cls_logits (`torch.FloatTensor` of shape `(batch_size,)`, *optional*, returned if `start_positions` or `end_positions` is not provided):
Log probabilities for the `is_impossible` label of the answers.
mems (`List[torch.FloatTensor]` of length `config.n_layers`):
Contains pre-computed hidden-states. Can be used (see `mems` input) to speed up sequential decoding.
The token ids which have their past given to this model should not be passed as `input_ids` as they
have already been computed.
hidden_states (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_hidden_states=True`` is passed or when ``config.output_hidden_states=True``):
Tuple of :obj:`torch.FloatTensor` (one for the output of the embeddings + one for the output of each layer)
of shape :obj:`(batch_size, sequence_length, hidden_size)`.
hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `torch.FloatTensor` (one for the output of the embeddings + one for the output of each layer)
of shape `(batch_size, sequence_length, hidden_size)`.
Hidden-states of the model at the output of each layer plus the initial embedding outputs.
attentions (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_attentions=True`` is passed or when ``config.output_attentions=True``):
Tuple of :obj:`torch.FloatTensor` (one for each layer) of shape :obj:`(batch_size, num_heads,
sequence_length, sequence_length)`.
attentions (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length, sequence_length)`.
Attentions weights after the attention softmax, used to compute the weighted average in the self-attention
heads.
......@@ -844,92 +837,91 @@ class XLNetForQuestionAnsweringOutput(ModelOutput):
XLNET_START_DOCSTRING = r"""
This model inherits from :class:`~transformers.PreTrainedModel`. Check the superclass documentation for the generic
This model inherits from [`PreTrainedModel`]. Check the superclass documentation for the generic
methods the library implements for all its model (such as downloading or saving, resizing the input embeddings,
pruning heads etc.)
This model is also a PyTorch `torch.nn.Module <https://pytorch.org/docs/stable/nn.html#torch.nn.Module>`__
This model is also a PyTorch [torch.nn.Module](https://pytorch.org/docs/stable/nn.html#torch.nn.Module)
subclass. Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to
general usage and behavior.
Parameters:
config (:class:`~transformers.XLNetConfig`): Model configuration class with all the parameters of the model.
config ([`XLNetConfig`]): Model configuration class with all the parameters of the model.
Initializing with a config file does not load the weights associated with the model, only the
configuration. Check out the :meth:`~transformers.PreTrainedModel.from_pretrained` method to load the model
configuration. Check out the [`~PreTrainedModel.from_pretrained`] method to load the model
weights.
"""
XLNET_INPUTS_DOCSTRING = r"""
Args:
input_ids (:obj:`torch.LongTensor` of shape :obj:`({0})`):
input_ids (`torch.LongTensor` of shape `({0})`):
Indices of input sequence tokens in the vocabulary.
Indices can be obtained using :class:`transformers.XLNetTokenizer`. See
:func:`transformers.PreTrainedTokenizer.encode` and :func:`transformers.PreTrainedTokenizer.__call__` for
Indices can be obtained using [`XLNetTokenizer`]. See
[`PreTrainedTokenizer.encode`] and [`PreTrainedTokenizer.__call__`] for
details.
`What are input IDs? <../glossary.html#input-ids>`__
attention_mask (:obj:`torch.FloatTensor` of shape :obj:`({0})`, `optional`):
Mask to avoid performing attention on padding token indices. Mask values selected in ``[0, 1]``:
[What are input IDs?](../glossary#input-ids)
attention_mask (`torch.FloatTensor` of shape `({0})`, *optional*):
Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
`What are attention masks? <../glossary.html#attention-mask>`__
mems (:obj:`List[torch.FloatTensor]` of length :obj:`config.n_layers`):
Contains pre-computed hidden-states (see :obj:`mems` output below) . Can be used to speed up sequential
decoding. The token ids which have their past given to this model should not be passed as :obj:`input_ids`
[What are attention masks?](../glossary#attention-mask)
mems (`List[torch.FloatTensor]` of length `config.n_layers`):
Contains pre-computed hidden-states (see `mems` output below) . Can be used to speed up sequential
decoding. The token ids which have their past given to this model should not be passed as `input_ids`
as they have already been computed.
:obj:`use_mems` has to be set to :obj:`True` to make use of :obj:`mems`.
perm_mask (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length, sequence_length)`, `optional`):
Mask to indicate the attention pattern for each input token with values selected in ``[0, 1]``:
`use_mems` has to be set to `True` to make use of `mems`.
perm_mask (`torch.FloatTensor` of shape `(batch_size, sequence_length, sequence_length)`, *optional*):
Mask to indicate the attention pattern for each input token with values selected in `[0, 1]`:
- if ``perm_mask[k, i, j] = 0``, i attend to j in batch k;
- if ``perm_mask[k, i, j] = 1``, i does not attend to j in batch k.
- if `perm_mask[k, i, j] = 0`, i attend to j in batch k;
- if `perm_mask[k, i, j] = 1`, i does not attend to j in batch k.
If not set, each token attends to all the others (full bidirectional attention). Only used during
pretraining (to define factorization order) or for sequential decoding (generation).
target_mapping (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, num_predict, sequence_length)`, `optional`):
Mask to indicate the output tokens to use. If ``target_mapping[k, i, j] = 1``, the i-th predict in batch k
target_mapping (`torch.FloatTensor` of shape `(batch_size, num_predict, sequence_length)`, *optional*):
Mask to indicate the output tokens to use. If `target_mapping[k, i, j] = 1`, the i-th predict in batch k
is on the j-th token. Only used during pretraining for partial prediction or for sequential decoding
(generation).
token_type_ids (:obj:`torch.LongTensor` of shape :obj:`({0})`, `optional`):
Segment token indices to indicate first and second portions of the inputs. Indices are selected in ``[0,
1]``:
token_type_ids (`torch.LongTensor` of shape `({0})`, *optional*):
Segment token indices to indicate first and second portions of the inputs. Indices are selected in `[0, 1]`:
- 0 corresponds to a `sentence A` token,
- 1 corresponds to a `sentence B` token.
- 0 corresponds to a *sentence A* token,
- 1 corresponds to a *sentence B* token.
`What are token type IDs? <../glossary.html#token-type-ids>`__
input_mask (:obj:`torch.FloatTensor` of shape :obj:`{0}`, `optional`):
Mask to avoid performing attention on padding token indices. Negative of :obj:`attention_mask`, i.e. with 0
[What are token type IDs?](../glossary#token-type-ids)
input_mask (`torch.FloatTensor` of shape `{0}`, *optional*):
Mask to avoid performing attention on padding token indices. Negative of `attention_mask`, i.e. with 0
for real tokens and 1 for padding which is kept for compatibility with the original code base.
Mask values selected in ``[0, 1]``:
Mask values selected in `[0, 1]`:
- 1 for tokens that are **masked**,
- 0 for tokens that are **not masked**.
You can only uses one of :obj:`input_mask` and :obj:`attention_mask`.
head_mask (:obj:`torch.FloatTensor` of shape :obj:`(num_heads,)` or :obj:`(num_layers, num_heads)`, `optional`):
Mask to nullify selected heads of the self-attention modules. Mask values selected in ``[0, 1]``:
You can only uses one of `input_mask` and `attention_mask`.
head_mask (`torch.FloatTensor` of shape `(num_heads,)` or `(num_layers, num_heads)`, *optional*):
Mask to nullify selected heads of the self-attention modules. Mask values selected in `[0, 1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
inputs_embeds (:obj:`torch.FloatTensor` of shape :obj:`({0}, hidden_size)`, `optional`):
Optionally, instead of passing :obj:`input_ids` you can choose to directly pass an embedded representation.
This is useful if you want more control over how to convert :obj:`input_ids` indices into associated
inputs_embeds (`torch.FloatTensor` of shape `({0}, hidden_size)`, *optional*):
Optionally, instead of passing `input_ids` you can choose to directly pass an embedded representation.
This is useful if you want more control over how to convert `input_ids` indices into associated
vectors than the model's internal embedding lookup matrix.
output_attentions (:obj:`bool`, `optional`):
Whether or not to return the attentions tensors of all attention layers. See ``attentions`` under returned
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under returned
tensors for more detail.
output_hidden_states (:obj:`bool`, `optional`):
Whether or not to return the hidden states of all layers. See ``hidden_states`` under returned tensors for
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for
more detail.
return_dict (:obj:`bool`, `optional`):
Whether or not to return a :class:`~transformers.file_utils.ModelOutput` instead of a plain tuple.
return_dict (`bool`, *optional*):
Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple.
"""
......@@ -1384,51 +1376,52 @@ class XLNetLMHeadModel(XLNetPreTrainedModel):
**kwargs, # delete when `use_cache` is removed in XLNetModel
):
r"""
labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size, num_predict)`, `optional`):
Labels for masked language modeling. :obj:`num_predict` corresponds to :obj:`target_mapping.shape[1]`. If
:obj:`target_mapping` is :obj`None`, then :obj:`num_predict` corresponds to :obj:`sequence_length`.
labels (`torch.LongTensor` of shape `(batch_size, num_predict)`, *optional*):
Labels for masked language modeling. `num_predict` corresponds to `target_mapping.shape[1]`. If
`target_mapping` is :obj*None*, then `num_predict` corresponds to `sequence_length`.
The labels should correspond to the masked input words that should be predicted and depends on
:obj:`target_mapping`. Note in order to perform standard auto-regressive language modeling a `<mask>` token
has to be added to the :obj:`input_ids` (see the :obj:`prepare_inputs_for_generation` function and examples
`target_mapping`. Note in order to perform standard auto-regressive language modeling a *<mask>* token
has to be added to the `input_ids` (see the `prepare_inputs_for_generation` function and examples
below)
Indices are selected in ``[-100, 0, ..., config.vocab_size]`` All labels set to ``-100`` are ignored, the
loss is only computed for labels in ``[0, ..., config.vocab_size]``
Indices are selected in `[-100, 0, ..., config.vocab_size]` All labels set to `-100` are ignored, the
loss is only computed for labels in `[0, ..., config.vocab_size]`
Return:
Examples::
>>> from transformers import XLNetTokenizer, XLNetLMHeadModel
>>> import torch
>>> tokenizer = XLNetTokenizer.from_pretrained('xlnet-large-cased')
>>> model = XLNetLMHeadModel.from_pretrained('xlnet-large-cased')
>>> # We show how to setup inputs to predict a next token using a bi-directional context.
>>> input_ids = torch.tensor(tokenizer.encode("Hello, my dog is very <mask>", add_special_tokens=False)).unsqueeze(0) # We will predict the masked token
>>> perm_mask = torch.zeros((1, input_ids.shape[1], input_ids.shape[1]), dtype=torch.float)
>>> perm_mask[:, :, -1] = 1.0 # Previous tokens don't see last token
>>> target_mapping = torch.zeros((1, 1, input_ids.shape[1]), dtype=torch.float) # Shape [1, 1, seq_length] => let's predict one token
>>> target_mapping[0, 0, -1] = 1.0 # Our first (and only) prediction will be the last token of the sequence (the masked token)
>>> outputs = model(input_ids, perm_mask=perm_mask, target_mapping=target_mapping)
>>> next_token_logits = outputs[0] # Output has shape [target_mapping.size(0), target_mapping.size(1), config.vocab_size]
>>> # The same way can the XLNetLMHeadModel be used to be trained by standard auto-regressive language modeling.
>>> input_ids = torch.tensor(tokenizer.encode("Hello, my dog is very <mask>", add_special_tokens=False)).unsqueeze(0) # We will predict the masked token
>>> labels = torch.tensor(tokenizer.encode("cute", add_special_tokens=False)).unsqueeze(0)
>>> assert labels.shape[0] == 1, 'only one word will be predicted'
>>> perm_mask = torch.zeros((1, input_ids.shape[1], input_ids.shape[1]), dtype=torch.float)
>>> perm_mask[:, :, -1] = 1.0 # Previous tokens don't see last token as is done in standard auto-regressive lm training
>>> target_mapping = torch.zeros((1, 1, input_ids.shape[1]), dtype=torch.float) # Shape [1, 1, seq_length] => let's predict one token
>>> target_mapping[0, 0, -1] = 1.0 # Our first (and only) prediction will be the last token of the sequence (the masked token)
>>> outputs = model(input_ids, perm_mask=perm_mask, target_mapping=target_mapping, labels=labels)
>>> loss = outputs.loss
>>> next_token_logits = outputs.logits # Logits have shape [target_mapping.size(0), target_mapping.size(1), config.vocab_size]
"""
Examples:
```python
>>> from transformers import XLNetTokenizer, XLNetLMHeadModel
>>> import torch
>>> tokenizer = XLNetTokenizer.from_pretrained('xlnet-large-cased')
>>> model = XLNetLMHeadModel.from_pretrained('xlnet-large-cased')
>>> # We show how to setup inputs to predict a next token using a bi-directional context.
>>> input_ids = torch.tensor(tokenizer.encode("Hello, my dog is very <mask>", add_special_tokens=False)).unsqueeze(0) # We will predict the masked token
>>> perm_mask = torch.zeros((1, input_ids.shape[1], input_ids.shape[1]), dtype=torch.float)
>>> perm_mask[:, :, -1] = 1.0 # Previous tokens don't see last token
>>> target_mapping = torch.zeros((1, 1, input_ids.shape[1]), dtype=torch.float) # Shape [1, 1, seq_length] => let's predict one token
>>> target_mapping[0, 0, -1] = 1.0 # Our first (and only) prediction will be the last token of the sequence (the masked token)
>>> outputs = model(input_ids, perm_mask=perm_mask, target_mapping=target_mapping)
>>> next_token_logits = outputs[0] # Output has shape [target_mapping.size(0), target_mapping.size(1), config.vocab_size]
>>> # The same way can the XLNetLMHeadModel be used to be trained by standard auto-regressive language modeling.
>>> input_ids = torch.tensor(tokenizer.encode("Hello, my dog is very <mask>", add_special_tokens=False)).unsqueeze(0) # We will predict the masked token
>>> labels = torch.tensor(tokenizer.encode("cute", add_special_tokens=False)).unsqueeze(0)
>>> assert labels.shape[0] == 1, 'only one word will be predicted'
>>> perm_mask = torch.zeros((1, input_ids.shape[1], input_ids.shape[1]), dtype=torch.float)
>>> perm_mask[:, :, -1] = 1.0 # Previous tokens don't see last token as is done in standard auto-regressive lm training
>>> target_mapping = torch.zeros((1, 1, input_ids.shape[1]), dtype=torch.float) # Shape [1, 1, seq_length] => let's predict one token
>>> target_mapping[0, 0, -1] = 1.0 # Our first (and only) prediction will be the last token of the sequence (the masked token)
>>> outputs = model(input_ids, perm_mask=perm_mask, target_mapping=target_mapping, labels=labels)
>>> loss = outputs.loss
>>> next_token_logits = outputs.logits # Logits have shape [target_mapping.size(0), target_mapping.size(1), config.vocab_size]
```"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
transformer_outputs = self.transformer(
......@@ -1471,8 +1464,8 @@ class XLNetLMHeadModel(XLNetPreTrainedModel):
@staticmethod
def _reorder_cache(mems: List[torch.Tensor], beam_idx: torch.Tensor) -> List[torch.Tensor]:
"""
This function is used to re-order the :obj:`mems` cache if :meth:`~transformers.PreTrainedModel.beam_search` or
:meth:`~transformers.PreTrainedModel.beam_sample` is called. This is required to match :obj:`mems` with the
This function is used to re-order the `mems` cache if [`~PreTrainedModel.beam_search`] or
[`~PreTrainedModel.beam_sample`] is called. This is required to match `mems` with the
correct beam_idx at every generation step.
"""
return [layer_past.index_select(1, beam_idx.to(layer_past.device)) for layer_past in mems]
......@@ -1524,10 +1517,9 @@ class XLNetForSequenceClassification(XLNetPreTrainedModel):
**kwargs, # delete when `use_cache` is removed in XLNetModel
):
r"""
labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`):
Labels for computing the sequence classification/regression loss. Indices should be in ``[0, ...,
config.num_labels - 1]``. If ``config.num_labels == 1`` a regression loss is computed (Mean-Square loss),
If ``config.num_labels > 1`` a classification loss is computed (Cross-Entropy).
labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
Labels for computing the sequence classification/regression loss. Indices should be in `[0, ..., config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss),
If `config.num_labels > 1` a classification loss is computed (Cross-Entropy).
"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
......@@ -1632,10 +1624,9 @@ class XLNetForTokenClassification(XLNetPreTrainedModel):
**kwargs, # delete when `use_cache` is removed in XLNetModel
):
r"""
labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`):
Labels for computing the multiple choice classification loss. Indices should be in ``[0, ...,
num_choices]`` where `num_choices` is the size of the second dimension of the input tensors. (see
`input_ids` above)
labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
Labels for computing the multiple choice classification loss. Indices should be in `[0, ..., num_choices]` where *num_choices* is the size of the second dimension of the input tensors. (see
*input_ids* above)
"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
......@@ -1730,10 +1721,9 @@ class XLNetForMultipleChoice(XLNetPreTrainedModel):
**kwargs, # delete when `use_cache` is removed in XLNetModel
):
r"""
labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`):
Labels for computing the multiple choice classification loss. Indices should be in ``[0, ...,
num_choices-1]`` where :obj:`num_choices` is the size of the second dimension of the input tensors. (See
:obj:`input_ids` above)
labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
Labels for computing the multiple choice classification loss. Indices should be in `[0, ..., num_choices-1]` where `num_choices` is the size of the second dimension of the input tensors. (See
`input_ids` above)
"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
......@@ -1835,13 +1825,13 @@ class XLNetForQuestionAnsweringSimple(XLNetPreTrainedModel):
**kwargs, # delete when `use_cache` is removed in XLNetModel
):
r"""
start_positions (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`):
start_positions (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
Labels for position (index) of the start of the labelled span for computing the token classification loss.
Positions are clamped to the length of the sequence (:obj:`sequence_length`). Position outside of the
Positions are clamped to the length of the sequence (`sequence_length`). Position outside of the
sequence are not taken into account for computing the loss.
end_positions (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`):
end_positions (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
Labels for position (index) of the end of the labelled span for computing the token classification loss.
Positions are clamped to the length of the sequence (:obj:`sequence_length`). Position outside of the
Positions are clamped to the length of the sequence (`sequence_length`). Position outside of the
sequence are not taken into account for computing the loss.
"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
......@@ -1947,40 +1937,41 @@ class XLNetForQuestionAnswering(XLNetPreTrainedModel):
**kwargs, # delete when `use_cache` is removed in XLNetModel
):
r"""
start_positions (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`):
start_positions (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
Labels for position (index) of the start of the labelled span for computing the token classification loss.
Positions are clamped to the length of the sequence (:obj:`sequence_length`). Position outside of the
Positions are clamped to the length of the sequence (`sequence_length`). Position outside of the
sequence are not taken into account for computing the loss.
end_positions (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`):
end_positions (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
Labels for position (index) of the end of the labelled span for computing the token classification loss.
Positions are clamped to the length of the sequence (:obj:`sequence_length`). Position outside of the
Positions are clamped to the length of the sequence (`sequence_length`). Position outside of the
sequence are not taken into account for computing the loss.
is_impossible (``torch.LongTensor`` of shape ``(batch_size,)``, `optional`):
is_impossible (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
Labels whether a question has an answer or no answer (SQuAD 2.0)
cls_index (``torch.LongTensor`` of shape ``(batch_size,)``, `optional`):
cls_index (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
Labels for position (index) of the classification token to use as input for computing plausibility of the
answer.
p_mask (``torch.FloatTensor`` of shape ``(batch_size, sequence_length)``, `optional`):
p_mask (`torch.FloatTensor` of shape `(batch_size, sequence_length)`, *optional*):
Optional mask of tokens which can't be in answers (e.g. [CLS], [PAD], ...). 1.0 means token should be
masked. 0.0 mean token is not masked.
Returns:
Example::
Example:
>>> from transformers import XLNetTokenizer, XLNetForQuestionAnswering
>>> import torch
```python
>>> from transformers import XLNetTokenizer, XLNetForQuestionAnswering
>>> import torch
>>> tokenizer = XLNetTokenizer.from_pretrained('xlnet-base-cased')
>>> model = XLNetForQuestionAnswering.from_pretrained('xlnet-base-cased')
>>> tokenizer = XLNetTokenizer.from_pretrained('xlnet-base-cased')
>>> model = XLNetForQuestionAnswering.from_pretrained('xlnet-base-cased')
>>> input_ids = torch.tensor(tokenizer.encode("Hello, my dog is cute", add_special_tokens=True)).unsqueeze(0) # Batch size 1
>>> start_positions = torch.tensor([1])
>>> end_positions = torch.tensor([3])
>>> outputs = model(input_ids, start_positions=start_positions, end_positions=end_positions)
>>> input_ids = torch.tensor(tokenizer.encode("Hello, my dog is cute", add_special_tokens=True)).unsqueeze(0) # Batch size 1
>>> start_positions = torch.tensor([1])
>>> end_positions = torch.tensor([3])
>>> outputs = model(input_ids, start_positions=start_positions, end_positions=end_positions)
>>> loss = outputs.loss
"""
>>> loss = outputs.loss
```"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
transformer_outputs = self.transformer(
......
templates/adding_a_new_model/cookiecutter-template-{{cookiecutter.modelname}}/modeling_flax_{{cookiecutter.lowercase_modelname}}.py
View file @ 7af80f66
......@@ -55,75 +55,71 @@ _CONFIG_FOR_DOC = "{{cookiecutter.camelcase_modelname}}Config"
_TOKENIZER_FOR_DOC = "{{cookiecutter.camelcase_modelname}}Tokenizer"
{{cookiecutter.uppercase_modelname}}_START_DOCSTRING = r"""
This model inherits from :class:`~transformers.FlaxPreTrainedModel`. Check the superclass documentation for the
This model inherits from [`FlaxPreTrainedModel`]. Check the superclass documentation for the
generic methods the library implements for all its model (such as downloading, saving and converting weights from
PyTorch models)
This model is also a Flax Linen `flax.linen.Module
<https://flax.readthedocs.io/en/latest/flax.linen.html#module>`__ subclass. Use it as a regular Flax linen Module
This model is also a Flax Linen [flax.linen.Module](https://flax.readthedocs.io/en/latest/flax.linen.html#module) subclass. Use it as a regular Flax linen Module
and refer to the Flax documentation for all matter related to general usage and behavior.
Finally, this model supports inherent JAX features such as:
- `Just-In-Time (JIT) compilation <https://jax.readthedocs.io/en/latest/jax.html#just-in-time-compilation-jit>`__
- `Automatic Differentiation <https://jax.readthedocs.io/en/latest/jax.html#automatic-differentiation>`__
- `Vectorization <https://jax.readthedocs.io/en/latest/jax.html#vectorization-vmap>`__
- `Parallelization <https://jax.readthedocs.io/en/latest/jax.html#parallelization-pmap>`__
- [Just-In-Time (JIT) compilation](https://jax.readthedocs.io/en/latest/jax.html#just-in-time-compilation-jit)
- [Automatic Differentiation](https://jax.readthedocs.io/en/latest/jax.html#automatic-differentiation)
- [Vectorization](https://jax.readthedocs.io/en/latest/jax.html#vectorization-vmap)
- [Parallelization](https://jax.readthedocs.io/en/latest/jax.html#parallelization-pmap)
Parameters:
config (:class:`~transformers.{{cookiecutter.uppercase_modelname}}Config`): Model configuration class with all the parameters of the model.
config ([`~{{cookiecutter.uppercase_modelname}}Config`]): Model configuration class with all the parameters of the model.
Initializing with a config file does not load the weights associated with the model, only the
configuration. Check out the :meth:`~transformers.FlaxPreTrainedModel.from_pretrained` method to load the
configuration. Check out the [`~FlaxPreTrainedModel.from_pretrained`] method to load the
model weights.
dtype (:obj:`jax.numpy.dtype`, `optional`, defaults to :obj:`jax.numpy.float32`):
The data type of the computation. Can be one of :obj:`jax.numpy.float32`, :obj:`jax.numpy.float16` (on
GPUs) and :obj:`jax.numpy.bfloat16` (on TPUs).
dtype (`jax.numpy.dtype`, *optional*, defaults to `jax.numpy.float32`):
The data type of the computation. Can be one of `jax.numpy.float32`, `jax.numpy.float16` (on
GPUs) and `jax.numpy.bfloat16` (on TPUs).
This can be used to enable mixed-precision training or half-precision inference on GPUs or TPUs. If
specified all the computation will be performed with the given ``dtype``.
specified all the computation will be performed with the given `dtype`.
**Note that this only specifies the dtype of the computation and does not influence the dtype of model
parameters.**
If you wish to change the dtype of the model parameters, see
:meth:`~transformers.FlaxPreTrainedModel.to_fp16` and :meth:`~transformers.FlaxPreTrainedModel.to_bf16`.
[`~FlaxPreTrainedModel.to_fp16`] and [`~FlaxPreTrainedModel.to_bf16`].
"""
{{cookiecutter.uppercase_modelname}}_INPUTS_DOCSTRING = r"""
Args:
input_ids (:obj:`numpy.ndarray` of shape :obj:`({0})`):
input_ids (`numpy.ndarray` of shape `({0})`):
Indices of input sequence tokens in the vocabulary.
Indices can be obtained using :class:`~transformers.{{cookiecutter.uppercase_modelname}}ConfiTokenizer`. See
:meth:`transformers.PreTrainedTokenizer.encode` and :func:`transformers.PreTrainedTokenizer.__call__` for
Indices can be obtained using [`~{{cookiecutter.uppercase_modelname}}ConfiTokenizer`]. See
[`PreTrainedTokenizer.encode`] and [`PreTrainedTokenizer.__call__`] for
details.
`What are input IDs? <../glossary.html#input-ids>`__
attention_mask (:obj:`numpy.ndarray` of shape :obj:`({0})`, `optional`):
Mask to avoid performing attention on padding token indices. Mask values selected in ``[0, 1]``:
[What are input IDs?](../glossary#input-ids)
attention_mask (`numpy.ndarray` of shape `({0})`, *optional*):
Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
`What are attention masks? <../glossary.html#attention-mask>`__
token_type_ids (:obj:`numpy.ndarray` of shape :obj:`({0})`, `optional`):
Segment token indices to indicate first and second portions of the inputs. Indices are selected in ``[0,
1]``:
[What are attention masks?](../glossary#attention-mask)
token_type_ids (`numpy.ndarray` of shape `({0})`, *optional*):
Segment token indices to indicate first and second portions of the inputs. Indices are selected in `[0, 1]`:
- 0 corresponds to a `sentence A` token,
- 1 corresponds to a `sentence B` token.
- 0 corresponds to a *sentence A* token,
- 1 corresponds to a *sentence B* token.
`What are token type IDs? <../glossary.html#token-type-ids>`__
position_ids (:obj:`numpy.ndarray` of shape :obj:`({0})`, `optional`):
Indices of positions of each input sequence tokens in the position embeddings. Selected in the range ``[0,
config.max_position_embeddings - 1]``.
head_mask (:obj:`numpy.ndarray` of shape :obj:`({0})`, `optional):
Mask to nullify selected heads of the attention modules. Mask values selected in ``[0, 1]``:
[What are token type IDs?](../glossary#token-type-ids)
position_ids (`numpy.ndarray` of shape `({0})`, *optional*):
Indices of positions of each input sequence tokens in the position embeddings. Selected in the range `[0, config.max_position_embeddings - 1]`.
head_mask (`numpy.ndarray` of shape `({0})`, `optional): Mask to nullify selected heads of the attention modules. Mask values selected in `[0, 1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
return_dict (:obj:`bool`, `optional`):
Whether or not to return a :class:`~transformers.file_utils.ModelOutput` instead of a plain tuple.
return_dict (`bool`, *optional*):
Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple.
"""
......@@ -1168,170 +1164,165 @@ _CONFIG_FOR_DOC = "{{cookiecutter.camelcase_modelname}}Config"
_TOKENIZER_FOR_DOC = "{{cookiecutter.camelcase_modelname}}Tokenizer"
{{cookiecutter.uppercase_modelname}}_START_DOCSTRING = r"""
This model inherits from :class:`~transformers.FlaxPreTrainedModel`. Check the superclass documentation for the
This model inherits from [`FlaxPreTrainedModel`]. Check the superclass documentation for the
generic methods the library implements for all its model (such as downloading or saving, resizing the input
embeddings, pruning heads etc.)
This model is also a Flax Linen `flax.nn.Module
<https://flax.readthedocs.io/en/latest/_autosummary/flax.nn.module.html>`__ subclass. Use it as a regular Flax
This model is also a Flax Linen [flax.nn.Module](https://flax.readthedocs.io/en/latest/_autosummary/flax.nn.module.html) subclass. Use it as a regular Flax
Module and refer to the Flax documentation for all matter related to general usage and behavior.
Finally, this model supports inherent JAX features such as:
- `Just-In-Time (JIT) compilation <https://jax.readthedocs.io/en/latest/jax.html#just-in-time-compilation-jit>`__
- `Automatic Differentiation <https://jax.readthedocs.io/en/latest/jax.html#automatic-differentiation>`__
- `Vectorization <https://jax.readthedocs.io/en/latest/jax.html#vectorization-vmap>`__
- `Parallelization <https://jax.readthedocs.io/en/latest/jax.html#parallelization-pmap>`__
- [Just-In-Time (JIT) compilation](https://jax.readthedocs.io/en/latest/jax.html#just-in-time-compilation-jit)
- [Automatic Differentiation](https://jax.readthedocs.io/en/latest/jax.html#automatic-differentiation)
- [Vectorization](https://jax.readthedocs.io/en/latest/jax.html#vectorization-vmap)
- [Parallelization](https://jax.readthedocs.io/en/latest/jax.html#parallelization-pmap)
Parameters:
config (:class:`~transformers.{{cookiecutter.camelcase_modelname}}Config`): Model configuration class with all the parameters of the model.
config ([`~{{cookiecutter.camelcase_modelname}}Config`]): Model configuration class with all the parameters of the model.
Initializing with a config file does not load the weights associated with the model, only the
configuration. Check out the :meth:`~transformers.FlaxPreTrainedModel.from_pretrained` method to load the
configuration. Check out the [`~FlaxPreTrainedModel.from_pretrained`] method to load the
model weights.
dtype (:obj:`jax.numpy.dtype`, `optional`, defaults to :obj:`jax.numpy.float32`):
The data type of the computation. Can be one of :obj:`jax.numpy.float32`, :obj:`jax.numpy.float16` (on
GPUs) and :obj:`jax.numpy.bfloat16` (on TPUs).
dtype (`jax.numpy.dtype`, *optional*, defaults to `jax.numpy.float32`):
The data type of the computation. Can be one of `jax.numpy.float32`, `jax.numpy.float16` (on
GPUs) and `jax.numpy.bfloat16` (on TPUs).
This can be used to enable mixed-precision training or half-precision inference on GPUs or TPUs. If
specified all the computation will be performed with the given ``dtype``.
specified all the computation will be performed with the given `dtype`.
**Note that this only specifies the dtype of the computation and does not influence the dtype of model
parameters.**
If you wish to change the dtype of the model parameters, see
:meth:`~transformers.FlaxPreTrainedModel.to_fp16` and :meth:`~transformers.FlaxPreTrainedModel.to_bf16`.
[`~FlaxPreTrainedModel.to_fp16`] and [`~FlaxPreTrainedModel.to_bf16`].
"""
{{cookiecutter.uppercase_modelname}}_INPUTS_DOCSTRING = r"""
Args:
input_ids (:obj:`jnp.ndarray` of shape :obj:`(batch_size, sequence_length)`):
input_ids (`jnp.ndarray` of shape `(batch_size, sequence_length)`):
Indices of input sequence tokens in the vocabulary. Padding will be ignored by default should you provide
it.
Indices can be obtained using :class:`~transformers.{{cookiecutter.camelcase_modelname}}Tokenizer`. See
:meth:`transformers.PreTrainedTokenizer.encode` and :meth:`transformers.PreTrainedTokenizer.__call__` for
Indices can be obtained using [`~{{cookiecutter.camelcase_modelname}}Tokenizer`]. See
[`PreTrainedTokenizer.encode`] and [`PreTrainedTokenizer.__call__`] for
details.
`What are input IDs? <../glossary.html#input-ids>`__
attention_mask (:obj:`jnp.ndarray` of shape :obj:`(batch_size, sequence_length)`, `optional`):
Mask to avoid performing attention on padding token indices. Mask values selected in ``[0, 1]``:
[What are input IDs?](../glossary#input-ids)
attention_mask (`jnp.ndarray` of shape `(batch_size, sequence_length)`, *optional*):
Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
`What are attention masks? <../glossary.html#attention-mask>`__
decoder_input_ids (:obj:`jnp.ndarray` of shape :obj:`(batch_size, target_sequence_length)`, `optional`):
[What are attention masks?](../glossary#attention-mask)
decoder_input_ids (`jnp.ndarray` of shape `(batch_size, target_sequence_length)`, *optional*):
Indices of decoder input sequence tokens in the vocabulary.
Indices can be obtained using :class:`~transformers.{{cookiecutter.camelcase_modelname}}Tokenizer`. See
:meth:`transformers.PreTrainedTokenizer.encode` and :meth:`transformers.PreTrainedTokenizer.__call__` for
Indices can be obtained using [`~{{cookiecutter.camelcase_modelname}}Tokenizer`]. See
[`PreTrainedTokenizer.encode`] and [`PreTrainedTokenizer.__call__`] for
details.
`What are decoder input IDs? <../glossary.html#decoder-input-ids>`__
[What are decoder input IDs?](../glossary#decoder-input-ids)
For translation and summarization training, :obj:`decoder_input_ids` should be provided. If no
:obj:`decoder_input_ids` is provided, the model will create this tensor by shifting the :obj:`input_ids` to
For translation and summarization training, `decoder_input_ids` should be provided. If no
`decoder_input_ids` is provided, the model will create this tensor by shifting the `input_ids` to
the right for denoising pre-training following the paper.
decoder_attention_mask (:obj:`jnp.ndarray` of shape :obj:`(batch_size, target_sequence_length)`, `optional`):
Default behavior: generate a tensor that ignores pad tokens in :obj:`decoder_input_ids`. Causal mask will
decoder_attention_mask (`jnp.ndarray` of shape `(batch_size, target_sequence_length)`, *optional*):
Default behavior: generate a tensor that ignores pad tokens in `decoder_input_ids`. Causal mask will
also be used by default.
If you want to change padding behavior, you should modify to your needs. See diagram 1 in `the paper
<https://arxiv.org/abs/1910.13461>`__ for more information on the default strategy.
position_ids (:obj:`numpy.ndarray` of shape :obj:`(batch_size, sequence_length)`, `optional`):
Indices of positions of each input sequence tokens in the position embeddings. Selected in the range ``[0,
config.max_position_embeddings - 1]``.
decoder_position_ids (:obj:`numpy.ndarray` of shape :obj:`(batch_size, sequence_length)`, `optional`):
If you want to change padding behavior, you should modify to your needs. See diagram 1 in [the paper](https://arxiv.org/abs/1910.13461) for more information on the default strategy.
position_ids (`numpy.ndarray` of shape `(batch_size, sequence_length)`, *optional*):
Indices of positions of each input sequence tokens in the position embeddings. Selected in the range `[0, config.max_position_embeddings - 1]`.
decoder_position_ids (`numpy.ndarray` of shape `(batch_size, sequence_length)`, *optional*):
Indices of positions of each decoder input sequence tokens in the position embeddings. Selected in the
range ``[0, config.max_position_embeddings - 1]``.
output_attentions (:obj:`bool`, `optional`):
Whether or not to return the attentions tensors of all attention layers. See ``attentions`` under returned
range `[0, config.max_position_embeddings - 1]`.
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under returned
tensors for more detail.
output_hidden_states (:obj:`bool`, `optional`):
Whether or not to return the hidden states of all layers. See ``hidden_states`` under returned tensors for
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for
more detail.
return_dict (:obj:`bool`, `optional`):
Whether or not to return a :class:`~transformers.file_utils.ModelOutput` instead of a plain tuple.
return_dict (`bool`, *optional*):
Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple.
"""
{{cookiecutter.uppercase_modelname}}_ENCODE_INPUTS_DOCSTRING = r"""
Args:
input_ids (:obj:`jnp.ndarray` of shape :obj:`(batch_size, sequence_length)`):
input_ids (`jnp.ndarray` of shape `(batch_size, sequence_length)`):
Indices of input sequence tokens in the vocabulary. Padding will be ignored by default should you provide
it.
Indices can be obtained using :class:`~transformers.{{cookiecutter.camelcase_modelname}}Tokenizer`. See
:meth:`transformers.PreTrainedTokenizer.encode` and :meth:`transformers.PreTrainedTokenizer.__call__` for
Indices can be obtained using [`~{{cookiecutter.camelcase_modelname}}Tokenizer`]. See
[`PreTrainedTokenizer.encode`] and [`PreTrainedTokenizer.__call__`] for
details.
`What are input IDs? <../glossary.html#input-ids>`__
attention_mask (:obj:`jnp.ndarray` of shape :obj:`(batch_size, sequence_length)`, `optional`):
Mask to avoid performing attention on padding token indices. Mask values selected in ``[0, 1]``:
[What are input IDs?](../glossary#input-ids)
attention_mask (`jnp.ndarray` of shape `(batch_size, sequence_length)`, *optional*):
Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
`What are attention masks? <../glossary.html#attention-mask>`__
position_ids (:obj:`numpy.ndarray` of shape :obj:`(batch_size, sequence_length)`, `optional`):
Indices of positions of each input sequence tokens in the position embeddings. Selected in the range ``[0,
config.max_position_embeddings - 1]``.
output_attentions (:obj:`bool`, `optional`):
Whether or not to return the attentions tensors of all attention layers. See ``attentions`` under returned
[What are attention masks?](../glossary#attention-mask)
position_ids (`numpy.ndarray` of shape `(batch_size, sequence_length)`, *optional*):
Indices of positions of each input sequence tokens in the position embeddings. Selected in the range `[0, config.max_position_embeddings - 1]`.
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under returned
tensors for more detail.
output_hidden_states (:obj:`bool`, `optional`):
Whether or not to return the hidden states of all layers. See ``hidden_states`` under returned tensors for
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for
more detail.
return_dict (:obj:`bool`, `optional`):
Whether or not to return a :class:`~transformers.file_utils.ModelOutput` instead of a plain tuple.
return_dict (`bool`, *optional*):
Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple.
"""
{{cookiecutter.uppercase_modelname}}_DECODE_INPUTS_DOCSTRING = r"""
Args:
decoder_input_ids (:obj:`jnp.ndarray` of shape :obj:`(batch_size, target_sequence_length)`):
decoder_input_ids (`jnp.ndarray` of shape `(batch_size, target_sequence_length)`):
Indices of decoder input sequence tokens in the vocabulary.
Indices can be obtained using :class:`~transformers.{{cookiecutter.camelcase_modelname}}Tokenizer`. See
:meth:`transformers.PreTrainedTokenizer.encode` and :meth:`transformers.PreTrainedTokenizer.__call__` for
Indices can be obtained using [`~{{cookiecutter.camelcase_modelname}}Tokenizer`]. See
[`PreTrainedTokenizer.encode`] and [`PreTrainedTokenizer.__call__`] for
details.
`What are decoder input IDs? <../glossary.html#decoder-input-ids>`__
[What are decoder input IDs?](../glossary#decoder-input-ids)
For translation and summarization training, :obj:`decoder_input_ids` should be provided. If no
:obj:`decoder_input_ids` is provided, the model will create this tensor by shifting the :obj:`input_ids` to
For translation and summarization training, `decoder_input_ids` should be provided. If no
`decoder_input_ids` is provided, the model will create this tensor by shifting the `input_ids` to
the right for denoising pre-training following the paper.
encoder_outputs (:obj:`tuple(tuple(jnp.ndarray)`):
Tuple consists of (:obj:`last_hidden_state`, `optional`: :obj:`hidden_states`, `optional`:
:obj:`attentions`) :obj:`last_hidden_state` of shape :obj:`(batch_size, sequence_length, hidden_size)`,
`optional`) is a sequence of hidden-states at the output of the last layer of the encoder. Used in the
encoder_outputs (`tuple(tuple(jnp.ndarray)`):
Tuple consists of (`last_hidden_state`, *optional*: `hidden_states`, *optional*:
`attentions`) `last_hidden_state` of shape `(batch_size, sequence_length, hidden_size)`,
*optional*) is a sequence of hidden-states at the output of the last layer of the encoder. Used in the
cross-attention of the decoder.
encoder_attention_mask (:obj:`jnp.ndarray` of shape :obj:`(batch_size, sequence_length)`, `optional`):
Mask to avoid performing attention on padding token indices. Mask values selected in ``[0, 1]``:
encoder_attention_mask (`jnp.ndarray` of shape `(batch_size, sequence_length)`, *optional*):
Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
`What are attention masks? <../glossary.html#attention-mask>`__
decoder_attention_mask (:obj:`jnp.ndarray` of shape :obj:`(batch_size, target_sequence_length)`, `optional`):
Default behavior: generate a tensor that ignores pad tokens in :obj:`decoder_input_ids`. Causal mask will
[What are attention masks?](../glossary#attention-mask)
decoder_attention_mask (`jnp.ndarray` of shape `(batch_size, target_sequence_length)`, *optional*):
Default behavior: generate a tensor that ignores pad tokens in `decoder_input_ids`. Causal mask will
also be used by default.
If you want to change padding behavior, you should modify to your needs. See diagram 1 in `the paper
<https://arxiv.org/abs/1910.13461>`__ for more information on the default strategy.
decoder_position_ids (:obj:`numpy.ndarray` of shape :obj:`(batch_size, sequence_length)`, `optional`):
If you want to change padding behavior, you should modify to your needs. See diagram 1 in [the paper](https://arxiv.org/abs/1910.13461) for more information on the default strategy.
decoder_position_ids (`numpy.ndarray` of shape `(batch_size, sequence_length)`, *optional*):
Indices of positions of each decoder input sequence tokens in the position embeddings. Selected in the
range ``[0, config.max_position_embeddings - 1]``.
past_key_values (:obj:`Dict[str, np.ndarray]`, `optional`, returned by ``init_cache`` or when passing previous ``past_key_values``):
range `[0, config.max_position_embeddings - 1]`.
past_key_values (`Dict[str, np.ndarray]`, *optional*, returned by `init_cache` or when passing previous `past_key_values`):
Dictionary of pre-computed hidden-states (key and values in the attention blocks) that can be used for fast
auto-regressive decoding. Pre-computed key and value hidden-states are of shape `[batch_size, max_length]`.
output_attentions (:obj:`bool`, `optional`):
Whether or not to return the attentions tensors of all attention layers. See ``attentions`` under returned
auto-regressive decoding. Pre-computed key and value hidden-states are of shape *[batch_size, max_length]*.
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under returned
tensors for more detail.
output_hidden_states (:obj:`bool`, `optional`):
Whether or not to return the hidden states of all layers. See ``hidden_states`` under returned tensors for
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for
more detail.
return_dict (:obj:`bool`, `optional`):
Whether or not to return a :class:`~transformers.file_utils.ModelOutput` instead of a plain tuple.
return_dict (`bool`, *optional*):
Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple.
"""
def shift_tokens_right(input_ids: jnp.ndarray, pad_token_id: int, decoder_start_token_id: int) -> jnp.ndarray:
......@@ -2074,15 +2065,14 @@ class Flax{{cookiecutter.camelcase_modelname}}PreTrainedModel(FlaxPreTrainedMode
def init_cache(self, batch_size, max_length, encoder_outputs):
r"""
Args:
batch_size (:obj:`int`):
batch_size (`int`):
batch_size used for fast auto-regressive decoding. Defines the batch size of the initialized cache.
max_length (:obj:`int`):
max_length (`int`):
maximum possible length for auto-regressive decoding. Defines the sequence length of the initialized
cache.
encoder_outputs (:obj:`Union[FlaxBaseModelOutput, tuple(tuple(jnp.ndarray)]`):
``encoder_outputs`` consists of (:obj:`last_hidden_state`, `optional`: :obj:`hidden_states`,
`optional`: :obj:`attentions`). :obj:`last_hidden_state` of shape :obj:`(batch_size, sequence_length,
hidden_size)`, `optional`) is a sequence of hidden-states at the output of the last layer of the
encoder_outputs (`Union[FlaxBaseModelOutput, tuple(tuple(jnp.ndarray)]`):
`encoder_outputs` consists of (`last_hidden_state`, *optional*: `hidden_states`,
*optional*: `attentions`). `last_hidden_state` of shape `(batch_size, sequence_length, hidden_size)`, *optional*) is a sequence of hidden-states at the output of the last layer of the
encoder. Used in the cross-attention of the decoder.
"""
# init input variables to retrieve cache
......
templates/adding_a_new_model/cookiecutter-template-{{cookiecutter.modelname}}/modeling_tf_{{cookiecutter.lowercase_modelname}}.py
View file @ 7af80f66
......@@ -123,7 +123,7 @@ class TF{{cookiecutter.camelcase_modelname}}Embeddings(tf.keras.layers.Layer):
Applies embedding based on inputs tensor.
Returns:
final_embeddings (:obj:`tf.Tensor`): output embedding tensor.
final_embeddings (`tf.Tensor`): output embedding tensor.
"""
assert not (input_ids is None and inputs_embeds is None)
......@@ -826,7 +826,7 @@ class TF{{cookiecutter.camelcase_modelname}}PreTrainedModel(TFPreTrainedModel):
Dummy inputs to build the network.
Returns:
:obj:`Dict[str, tf.Tensor]`: The dummy inputs.
`Dict[str, tf.Tensor]`: The dummy inputs.
"""
dummy = {"input_ids": tf.constant(DUMMY_INPUTS)}
# Add `encoder_hidden_states` to make the cross-attention layers' weights initialized
......@@ -841,91 +841,91 @@ class TF{{cookiecutter.camelcase_modelname}}PreTrainedModel(TFPreTrainedModel):
{{cookiecutter.uppercase_modelname}}_START_DOCSTRING = r"""
This model inherits from :class:`~transformers.TFPreTrainedModel`. Check the superclass documentation for the
This model inherits from [`TFPreTrainedModel`]. Check the superclass documentation for the
generic methods the library implements for all its model (such as downloading or saving, resizing the input
embeddings, pruning heads etc.)
This model is also a `tf.keras.Model <https://www.tensorflow.org/api_docs/python/tf/keras/Model>`__ subclass.
This model is also a [tf.keras.Model](https://www.tensorflow.org/api_docs/python/tf/keras/Model) subclass.
Use it as a regular TF 2.0 Keras Model and refer to the TF 2.0 documentation for all matter related to general
usage and behavior.
.. note::
<Tip>
TF 2.0 models accepts two formats as inputs:
TF 2.0 models accepts two formats as inputs:
- having all inputs as keyword arguments (like PyTorch models), or
- having all inputs as a list, tuple or dict in the first positional arguments.
- having all inputs as keyword arguments (like PyTorch models), or
- having all inputs as a list, tuple or dict in the first positional arguments.
This second option is useful when using :meth:`tf.keras.Model.fit` method which currently requires having
all the tensors in the first argument of the model call function: :obj:`model(inputs)`.
This second option is useful when using [`tf.keras.Model.fit`] method which currently requires having
all the tensors in the first argument of the model call function: `model(inputs)`.
If you choose this second option, there are three possibilities you can use to gather all the input Tensors
in the first positional argument :
If you choose this second option, there are three possibilities you can use to gather all the input Tensors
in the first positional argument :
- a single Tensor with :obj:`input_ids` only and nothing else: :obj:`model(inputs_ids)`
- a list of varying length with one or several input Tensors IN THE ORDER given in the docstring:
:obj:`model([input_ids, attention_mask])` or :obj:`model([input_ids, attention_mask, token_type_ids])`
- a dictionary with one or several input Tensors associated to the input names given in the docstring:
:obj:`model({"input_ids": input_ids, "token_type_ids": token_type_ids})`
- a single Tensor with `input_ids` only and nothing else: `model(inputs_ids)`
- a list of varying length with one or several input Tensors IN THE ORDER given in the docstring:
`model([input_ids, attention_mask])` or `model([input_ids, attention_mask, token_type_ids])`
- a dictionary with one or several input Tensors associated to the input names given in the docstring:
`model({"input_ids": input_ids, "token_type_ids": token_type_ids})`
</Tip>
Args:
config (:class:`~transformers.{{cookiecutter.camelcase_modelname}}Config`): Model configuration class with all the parameters of the model.
config ([`~{{cookiecutter.camelcase_modelname}}Config`]): Model configuration class with all the parameters of the model.
Initializing with a config file does not load the weights associated with the model, only the configuration.
Check out the :meth:`~transformers.PreTrainedModel.from_pretrained` method to load the model weights.
Check out the [`~PreTrainedModel.from_pretrained`] method to load the model weights.
"""
{{cookiecutter.uppercase_modelname}}_INPUTS_DOCSTRING = r"""
Args:
input_ids (:obj:`np.ndarray`, :obj:`tf.Tensor`, :obj:`List[tf.Tensor]` :obj:`Dict[str, tf.Tensor]` or :obj:`Dict[str, np.ndarray]` and each example must have the shape :obj:`({0})`):
input_ids (`np.ndarray`, `tf.Tensor`, `List[tf.Tensor]` ``Dict[str, tf.Tensor]` or `Dict[str, np.ndarray]` and each example must have the shape `({0})`):
Indices of input sequence tokens in the vocabulary.
Indices can be obtained using :class:`~transformers.BertTokenizer`. See
:func:`transformers.PreTrainedTokenizer.__call__` and :func:`transformers.PreTrainedTokenizer.encode` for
Indices can be obtained using [`BertTokenizer`]. See
[`PreTrainedTokenizer.__call__`] and [`PreTrainedTokenizer.encode`] for
details.
`What are input IDs? <../glossary.html#input-ids>`__
attention_mask (:obj:`np.ndarray` or :obj:`tf.Tensor` of shape :obj:`({0})`, `optional`):
Mask to avoid performing attention on padding token indices. Mask values selected in ``[0, 1]``:
[What are input IDs?](../glossary#input-ids)
attention_mask (`np.ndarray` or `tf.Tensor` of shape `({0})`, *optional*):
Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
`What are attention masks? <../glossary.html#attention-mask>`__
token_type_ids (:obj:`np.ndarray` or :obj:`tf.Tensor` of shape :obj:`({0})`, `optional`):
Segment token indices to indicate first and second portions of the inputs. Indices are selected in ``[0,
1]``:
[What are attention masks?](../glossary#attention-mask)
token_type_ids (`np.ndarray` or `tf.Tensor` of shape `({0})`, *optional*):
Segment token indices to indicate first and second portions of the inputs. Indices are selected in `[0, 1]`:
- 0 corresponds to a `sentence A` token,
- 1 corresponds to a `sentence B` token.
- 0 corresponds to a *sentence A* token,
- 1 corresponds to a *sentence B* token.
`What are token type IDs? <../glossary.html#token-type-ids>`__
position_ids (:obj:`np.ndarray` or :obj:`tf.Tensor` of shape :obj:`({0})`, `optional`):
Indices of positions of each input sequence tokens in the position embeddings. Selected in the range ``[0,
config.max_position_embeddings - 1]``.
[What are token type IDs?](../glossary#token-type-ids)
position_ids (`np.ndarray` or `tf.Tensor` of shape `({0})`, *optional*):
Indices of positions of each input sequence tokens in the position embeddings. Selected in the range `[0, config.max_position_embeddings - 1]`.
`What are position IDs? <../glossary.html#position-ids>`__
head_mask (:obj:`np.ndarray` or :obj:`tf.Tensor` of shape :obj:`(num_heads,)` or :obj:`(num_layers, num_heads)`, `optional`):
Mask to nullify selected heads of the self-attention modules. Mask values selected in ``[0, 1]``:
[What are position IDs?](../glossary#position-ids)
head_mask (`np.ndarray` or `tf.Tensor` of shape `(num_heads,)` or `(num_layers, num_heads)`, *optional*):
Mask to nullify selected heads of the self-attention modules. Mask values selected in `[0, 1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
inputs_embeds (:obj:`np.ndarray` or :obj:`tf.Tensor` of shape :obj:`({0}, hidden_size)`, `optional`):
Optionally, instead of passing :obj:`input_ids` you can choose to directly pass an embedded representation.
This is useful if you want more control over how to convert :obj:`input_ids` indices into associated
inputs_embeds (`np.ndarray` or `tf.Tensor` of shape `({0}, hidden_size)`, *optional*):
Optionally, instead of passing `input_ids` you can choose to directly pass an embedded representation.
This is useful if you want more control over how to convert `input_ids` indices into associated
vectors than the model's internal embedding lookup matrix.
output_attentions (:obj:`bool`, `optional`):
Whether or not to return the attentions tensors of all attention layers. See ``attentions`` under returned
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under returned
tensors for more detail. This argument can be used only in eager mode, in graph mode the value in the
config will be used instead.
output_hidden_states (:obj:`bool`, `optional`):
Whether or not to return the hidden states of all layers. See ``hidden_states`` under returned tensors for
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for
more detail. This argument can be used only in eager mode, in graph mode the value in the config will be
used instead.
return_dict (:obj:`bool`, `optional`):
Whether or not to return a :class:`~transformers.file_utils.ModelOutput` instead of a plain tuple. This
return_dict (`bool`, *optional*):
Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple. This
argument can be used in eager mode, in graph mode the value will always be set to True.
training (:obj:`bool`, `optional`, defaults to :obj:`False`):
training (`bool`, *optional*, defaults to `False``):
Whether or not to use the model in training mode (some modules like dropout modules have different
behaviors between training and evaluation).
"""
......@@ -967,24 +967,24 @@ class TF{{cookiecutter.camelcase_modelname}}Model(TF{{cookiecutter.camelcase_mod
**kwargs,
) -> Union[TFBaseModelOutputWithPoolingAndCrossAttentions, Tuple[tf.Tensor]]:
r"""
encoder_hidden_states (:obj:`tf.Tensor` of shape :obj:`(batch_size, sequence_length, hidden_size)`, `optional`):
encoder_hidden_states (`tf.Tensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*):
Sequence of hidden-states at the output of the last layer of the encoder. Used in the cross-attention if
the model is configured as a decoder.
encoder_attention_mask (:obj:`tf.Tensor` of shape :obj:`(batch_size, sequence_length)`, `optional`):
encoder_attention_mask (`tf.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
Mask to avoid performing attention on the padding token indices of the encoder input. This mask is used in
the cross-attention if the model is configured as a decoder. Mask values selected in ``[0, 1]``:
the cross-attention if the model is configured as a decoder. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
past_key_values (:obj:`Tuple[Tuple[tf.Tensor]]` of length :obj:`config.n_layers`)
past_key_values (`Tuple[Tuple[tf.Tensor]]` of length `config.n_layers`)
contains precomputed key and value hidden states of the attention blocks. Can be used to speed up decoding.
If :obj:`past_key_values` are used, the user can optionally input only the last :obj:`decoder_input_ids`
(those that don't have their past key value states given to this model) of shape :obj:`(batch_size, 1)`
instead of all :obj:`decoder_input_ids` of shape :obj:`(batch_size, sequence_length)`.
use_cache (:obj:`bool`, `optional`, defaults to :obj:`True`):
If set to :obj:`True`, :obj:`past_key_values` key value states are returned and can be used to speed up
decoding (see :obj:`past_key_values`). Set to :obj:`False` during training, :obj:`True` during generation
If `past_key_values` are used, the user can optionally input only the last `decoder_input_ids`
(those that don't have their past key value states given to this model) of shape `(batch_size, 1)`
instead of all `decoder_input_ids` of shape `(batch_size, sequence_length)`.
use_cache (`bool`, *optional*, defaults to `True`):
If set to `True`, `past_key_values` key value states are returned and can be used to speed up
decoding (see `past_key_values`). Set to `False` during training, `True` during generation
"""
inputs = input_processing(
func=self.call,
......@@ -1087,10 +1087,9 @@ class TF{{cookiecutter.camelcase_modelname}}ForMaskedLM(TF{{cookiecutter.camelca
**kwargs,
) -> Union[TFMaskedLMOutput, Tuple[tf.Tensor]]:
r"""
labels (:obj:`tf.Tensor` or :obj:`np.ndarray` of shape :obj:`(batch_size, sequence_length)`, `optional`):
Labels for computing the masked language modeling loss. Indices should be in ``[-100, 0, ...,
config.vocab_size]`` (see ``input_ids`` docstring) Tokens with indices set to ``-100`` are ignored
(masked), the loss is only computed for the tokens with labels in ``[0, ..., config.vocab_size]``
labels (`tf.Tensor` or `np.ndarray` of shape `(batch_size, sequence_length)`, *optional*):
Labels for computing the masked language modeling loss. Indices should be in `[-100, 0, ..., config.vocab_size]` (see `input_ids` docstring) Tokens with indices set to `-100` are ignored
(masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]`
"""
inputs = input_processing(
func=self.call,
......@@ -1200,27 +1199,26 @@ class TF{{cookiecutter.camelcase_modelname}}ForCausalLM(TF{{cookiecutter.camelca
**kwargs,
) -> Union[TFCausalLMOutputWithCrossAttentions, Tuple[tf.Tensor]]:
r"""
encoder_hidden_states (:obj:`tf.Tensor` of shape :obj:`(batch_size, sequence_length, hidden_size)`, `optional`):
encoder_hidden_states (`tf.Tensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*):
Sequence of hidden-states at the output of the last layer of the encoder. Used in the cross-attention if
the model is configured as a decoder.
encoder_attention_mask (:obj:`tf.Tensor` of shape :obj:`(batch_size, sequence_length)`, `optional`):
encoder_attention_mask (`tf.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
Mask to avoid performing attention on the padding token indices of the encoder input. This mask is used in
the cross-attention if the model is configured as a decoder. Mask values selected in ``[0, 1]``:
the cross-attention if the model is configured as a decoder. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
past_key_values (:obj:`Tuple[Tuple[tf.Tensor]]` of length :obj:`config.n_layers`)
past_key_values (`Tuple[Tuple[tf.Tensor]]` of length `config.n_layers`)
contains precomputed key and value hidden states of the attention blocks. Can be used to speed up decoding.
If :obj:`past_key_values` are used, the user can optionally input only the last :obj:`decoder_input_ids`
(those that don't have their past key value states given to this model) of shape :obj:`(batch_size, 1)`
instead of all :obj:`decoder_input_ids` of shape :obj:`(batch_size, sequence_length)`.
use_cache (:obj:`bool`, `optional`, defaults to :obj:`True`):
If set to :obj:`True`, :obj:`past_key_values` key value states are returned and can be used to speed up
decoding (see :obj:`past_key_values`). Set to :obj:`False` during training, :obj:`True` during generation
labels (:obj:`tf.Tensor` or :obj:`np.ndarray` of shape :obj:`(batch_size, sequence_length)`, `optional`):
Labels for computing the cross entropy classification loss. Indices should be in ``[0, ...,
config.vocab_size - 1]``.
If `past_key_values` are used, the user can optionally input only the last `decoder_input_ids`
(those that don't have their past key value states given to this model) of shape `(batch_size, 1)`
instead of all `decoder_input_ids` of shape `(batch_size, sequence_length)`.
use_cache (`bool`, *optional*, defaults to `True`):
If set to `True`, `past_key_values` key value states are returned and can be used to speed up
decoding (see `past_key_values`). Set to `False` during training, `True` during generation
labels (`tf.Tensor` or `np.ndarray` of shape `(batch_size, sequence_length)`, *optional*):
Labels for computing the cross entropy classification loss. Indices should be in `[0, ..., config.vocab_size - 1]`.
"""
inputs = input_processing(
func=self.call,
......@@ -1363,10 +1361,9 @@ class TF{{cookiecutter.camelcase_modelname}}ForSequenceClassification(TF{{cookie
**kwargs,
) -> Union[TFSequenceClassifierOutput, Tuple[tf.Tensor]]:
r"""
labels (:obj:`tf.Tensor` or :obj:`np.ndarray` of shape :obj:`(batch_size,)`, `optional`):
Labels for computing the sequence classification/regression loss. Indices should be in :obj:`[0, ...,
config.num_labels - 1]`. If :obj:`config.num_labels == 1` a regression loss is computed (Mean-Square loss),
If :obj:`config.num_labels > 1` a classification loss is computed (Cross-Entropy).
labels (`tf.Tensor` or `np.ndarray` of shape `(batch_size,)`, *optional*):
Labels for computing the sequence classification/regression loss. Indices should be in `[0, ..., config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss),
If `config.num_labels > 1` a classification loss is computed (Cross-Entropy).
"""
inputs = input_processing(
func=self.call,
......@@ -1469,10 +1466,9 @@ class TF{{cookiecutter.camelcase_modelname}}ForMultipleChoice(TF{{cookiecutter.c
**kwargs,
) -> Union[TFMultipleChoiceModelOutput, Tuple[tf.Tensor]]:
r"""
labels (:obj:`tf.Tensor` or :obj:`np.ndarray` of shape :obj:`(batch_size,)`, `optional`):
Labels for computing the multiple choice classification loss. Indices should be in ``[0, ...,
num_choices]`` where :obj:`num_choices` is the size of the second dimension of the input tensors. (See
:obj:`input_ids` above)
labels (`tf.Tensor` or `np.ndarray` of shape `(batch_size,)`, *optional*):
Labels for computing the multiple choice classification loss. Indices should be in `[0, ..., num_choices]` where `num_choices` is the size of the second dimension of the input tensors. (See
`input_ids` above)
"""
inputs = input_processing(
func=self.call,
......@@ -1612,9 +1608,8 @@ class TF{{cookiecutter.camelcase_modelname}}ForTokenClassification(TF{{cookiecut
**kwargs,
) -> Union[TFTokenClassifierOutput, Tuple[tf.Tensor]]:
r"""
labels (:obj:`tf.Tensor` or :obj:`np.ndarray` of shape :obj:`(batch_size, sequence_length)`, `optional`):
Labels for computing the token classification loss. Indices should be in ``[0, ..., config.num_labels -
1]``.
labels (`tf.Tensor` or `np.ndarray` of shape `(batch_size, sequence_length)`, *optional*):
Labels for computing the token classification loss. Indices should be in `[0, ..., config.num_labels - 1]`.
"""
inputs = input_processing(
func=self.call,
......@@ -1709,13 +1704,13 @@ class TF{{cookiecutter.camelcase_modelname}}ForQuestionAnswering(TF{{cookiecutte
**kwargs,
) -> Union[TFQuestionAnsweringModelOutput, Tuple[tf.Tensor]]:
r"""
start_positions (:obj:`tf.Tensor` or :obj:`np.ndarray` of shape :obj:`(batch_size,)`, `optional`):
start_positions (`tf.Tensor` or `np.ndarray` of shape `(batch_size,)`, *optional*):
Labels for position (index) of the start of the labelled span for computing the token classification loss.
Positions are clamped to the length of the sequence (:obj:`sequence_length`). Position outside of the
Positions are clamped to the length of the sequence (`sequence_length`). Position outside of the
sequence are not taken into account for computing the loss.
end_positions (:obj:`tf.Tensor` or :obj:`np.ndarray` of shape :obj:`(batch_size,)`, `optional`):
end_positions (`tf.Tensor` or `np.ndarray` of shape `(batch_size,)`, *optional*):
Labels for position (index) of the end of the labelled span for computing the token classification loss.
Positions are clamped to the length of the sequence (:obj:`sequence_length`). Position outside of the
Positions are clamped to the length of the sequence (`sequence_length`). Position outside of the
sequence are not taken into account for computing the loss.
"""
inputs = input_processing(
......@@ -2057,11 +2052,11 @@ class TF{{cookiecutter.camelcase_modelname}}EncoderLayer(tf.keras.layers.Layer):
def call(self, hidden_states: tf.Tensor, attention_mask: tf.Tensor, layer_head_mask: tf.Tensor, training=False):
"""
Args:
hidden_states (:obj:`tf.Tensor`): input to the layer of shape `(seq_len, batch, embed_dim)`
attention_mask (:obj:`tf.Tensor`): attention mask of size
`(batch, 1, tgt_len, src_len)` where padding elements are indicated by very large negative values.
layer_head_mask (:obj:`tf.Tensor`): mask for attention heads in a given layer of size
`(encoder_attention_heads,)`
hidden_states (`tf.Tensor`): input to the layer of shape *(seq_len, batch, embed_dim)*
attention_mask (`tf.Tensor`): attention mask of size
*(batch, 1, tgt_len, src_len)* where padding elements are indicated by very large negative values.
layer_head_mask (`tf.Tensor`): mask for attention heads in a given layer of size
*(encoder_attention_heads,)*
"""
residual = hidden_states
hidden_states, self_attn_weights, _ = self.self_attn(
......@@ -2133,17 +2128,17 @@ class TF{{cookiecutter.camelcase_modelname}}DecoderLayer(tf.keras.layers.Layer):
) -> Tuple[tf.Tensor, tf.Tensor, Tuple[Tuple[tf.Tensor]]]:
"""
Args:
hidden_states (:obj:`tf.Tensor`): input to the layer of shape `(seq_len, batch, embed_dim)`
attention_mask (:obj:`tf.Tensor`): attention mask of size
`(batch, 1, tgt_len, src_len)` where padding elements are indicated by very large negative values.
encoder_hidden_states (:obj:`tf.Tensor`): cross attention input to the layer of shape `(seq_len, batch, embed_dim)`
encoder_attention_mask (:obj:`tf.Tensor`): encoder attention mask of size
`(batch, 1, tgt_len, src_len)` where padding elements are indicated by very large negative values.
layer_head_mask (:obj:`tf.Tensor`): mask for attention heads in a given layer of size
`(decoder_attention_heads,)`
cross_attn_layer_head_mask (:obj:`tf.Tensor`): mask for heads of the cross-attention module.
`(decoder_attention_heads,)`
past_key_value (:obj:`Tuple(tf.Tensor)`): cached past key and value projection states
hidden_states (`tf.Tensor`): input to the layer of shape *(seq_len, batch, embed_dim)*
attention_mask (`tf.Tensor`): attention mask of size
*(batch, 1, tgt_len, src_len)* where padding elements are indicated by very large negative values.
encoder_hidden_states (`tf.Tensor`): cross attention input to the layer of shape *(seq_len, batch, embed_dim)*
encoder_attention_mask (`tf.Tensor`): encoder attention mask of size
*(batch, 1, tgt_len, src_len)* where padding elements are indicated by very large negative values.
layer_head_mask (`tf.Tensor`): mask for attention heads in a given layer of size
*(decoder_attention_heads,)*
cross_attn_layer_head_mask (`tf.Tensor`): mask for heads of the cross-attention module.
*(decoder_attention_heads,)*
past_key_value (`Tuple(tf.Tensor)`): cached past key and value projection states
"""
residual = hidden_states
......@@ -2234,116 +2229,118 @@ class TF{{cookiecutter.camelcase_modelname}}PreTrainedModel(TFPreTrainedModel):
{{cookiecutter.uppercase_modelname}}_START_DOCSTRING = r"""
This model inherits from :class:`~transformers.TFPreTrainedModel`. Check the superclass documentation for the
This model inherits from [`TFPreTrainedModel`]. Check the superclass documentation for the
generic methods the library implements for all its model (such as downloading or saving, resizing the input
embeddings, pruning heads etc.)
This model is also a `tf.keras.Model <https://www.tensorflow.org/api_docs/python/tf/keras/Model>`__ subclass. Use
This model is also a [tf.keras.Model](https://www.tensorflow.org/api_docs/python/tf/keras/Model) subclass. Use
it as a regular TF 2.0 Keras Model and refer to the TF 2.0 documentation for all matter related to general usage
and behavior.
.. note::
<Tip>
TF 2.0 models accepts two formats as inputs:
TF 2.0 models accepts two formats as inputs:
- having all inputs as keyword arguments (like PyTorch models), or
- having all inputs as a list, tuple or dict in the first positional arguments.
- having all inputs as keyword arguments (like PyTorch models), or
- having all inputs as a list, tuple or dict in the first positional arguments.
This second option is useful when using [`tf.keras.Model.fit`] method which currently requires having all
the tensors in the first argument of the model call function: `model(inputs)`.
This second option is useful when using :meth:`tf.keras.Model.fit` method which currently requires having all
the tensors in the first argument of the model call function: :obj:`model(inputs)`.
If you choose this second option, there are three possibilities you can use to gather all the input Tensors in
the first positional argument :
If you choose this second option, there are three possibilities you can use to gather all the input Tensors in
the first positional argument :
- a single Tensor with `input_ids` only and nothing else: `model(input_ids)`
- a list of varying length with one or several input Tensors IN THE ORDER given in the docstring:
`model([input_ids, attention_mask])` or `model([input_ids, attention_mask, token_type_ids])`
- a dictionary with one or several input Tensors associated to the input names given in the docstring:
`model({"input_ids": input_ids, "token_type_ids": token_type_ids})`
- a single Tensor with :obj:`input_ids` only and nothing else: :obj:`model(input_ids)`
- a list of varying length with one or several input Tensors IN THE ORDER given in the docstring:
:obj:`model([input_ids, attention_mask])` or :obj:`model([input_ids, attention_mask, token_type_ids])`
- a dictionary with one or several input Tensors associated to the input names given in the docstring:
:obj:`model({"input_ids": input_ids, "token_type_ids": token_type_ids})`
</Tip>
Args:
config (:class:`~transformers.{{cookiecutter.camelcase_modelname}}Config`): Model configuration class with all the parameters of the model.
config ([`~{{cookiecutter.camelcase_modelname}}Config`]): Model configuration class with all the parameters of the model.
Initializing with a config file does not load the weights associated with the model, only the
configuration. Check out the :meth:`~transformers.TFPreTrainedModel.from_pretrained` method to load the
configuration. Check out the [`~TFPreTrainedModel.from_pretrained`] method to load the
model weights.
"""
{{cookiecutter.uppercase_modelname}}_INPUTS_DOCSTRING = r"""
Args:
input_ids (:obj:`tf.Tensor` of shape :obj:`({0})`):
input_ids (`tf.Tensor` of shape `({0})`):
Indices of input sequence tokens in the vocabulary.
Indices can be obtained using :class:`~transformers.{{cookiecutter.camelcase_modelname}}Tokenizer`. See
:meth:`transformers.PreTrainedTokenizer.encode` and :meth:`transformers.PreTrainedTokenizer.__call__` for
Indices can be obtained using [`~{{cookiecutter.camelcase_modelname}}Tokenizer`]. See
[`PreTrainedTokenizer.encode`] and [`PreTrainedTokenizer.__call__`] for
details.
`What are input IDs? <../glossary.html#input-ids>`__
attention_mask (:obj:`tf.Tensor` of shape :obj:`({0})`, `optional`):
Mask to avoid performing attention on padding token indices. Mask values selected in ``[0, 1]``:
[What are input IDs?](../glossary#input-ids)
attention_mask (`tf.Tensor` of shape `({0})`, *optional*):
Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
`What are attention masks? <../glossary.html#attention-mask>`__
decoder_input_ids (:obj:`tf.Tensor` of shape :obj:`(batch_size, target_sequence_length)`, `optional`):
[What are attention masks?](../glossary#attention-mask)
decoder_input_ids (`tf.Tensor` of shape `(batch_size, target_sequence_length)`, *optional*):
Indices of decoder input sequence tokens in the vocabulary.
Indices can be obtained using :class:`~transformers.{{cookiecutter.camelcase_modelname}}Tokenizer`. See
:meth:`transformers.PreTrainedTokenizer.encode` and :meth:`transformers.PreTrainedTokenizer.__call__` for
Indices can be obtained using [`~{{cookiecutter.camelcase_modelname}}Tokenizer`]. See
[`PreTrainedTokenizer.encode`] and [`PreTrainedTokenizer.__call__`] for
details.
`What are input IDs? <../glossary.html#input-ids>`__
[What are input IDs?](../glossary#input-ids)
{{cookiecutter.camelcase_modelname}} uses the :obj:`eos_token_id` as the starting token for
:obj:`decoder_input_ids` generation. If :obj:`past_key_values` is used, optionally only the last
:obj:`decoder_input_ids` have to be input (see :obj:`past_key_values`).
{{cookiecutter.camelcase_modelname}} uses the `eos_token_id` as the starting token for
`decoder_input_ids` generation. If `past_key_values` is used, optionally only the last
`decoder_input_ids` have to be input (see `past_key_values`).
For translation and summarization training, :obj:`decoder_input_ids` should be provided. If no
:obj:`decoder_input_ids` is provided, the model will create this tensor by shifting the :obj:`input_ids` to
For translation and summarization training, `decoder_input_ids` should be provided. If no
`decoder_input_ids` is provided, the model will create this tensor by shifting the `input_ids` to
the right for denoising pre-training following the paper.
decoder_attention_mask (:obj:`tf.Tensor` of shape :obj:`(batch_size, target_sequence_length)`, `optional`):
decoder_attention_mask (`tf.Tensor` of shape `(batch_size, target_sequence_length)`, *optional*):
will be made by default and ignore pad tokens. It is not recommended to set this for most use cases.
head_mask (:obj:`tf.Tensor` of shape :obj:`(encoder_layers, encoder_attention_heads)`, `optional`):
Mask to nullify selected heads of the attention modules in the encoder. Mask values selected in ``[0, 1]``:
head_mask (`tf.Tensor` of shape `(encoder_layers, encoder_attention_heads)`, *optional*):
Mask to nullify selected heads of the attention modules in the encoder. Mask values selected in `[0, 1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
decoder_head_mask (:obj:`tf.Tensor` of shape :obj:`(decoder_layers, decoder_attention_heads)`, `optional`):
Mask to nullify selected heads of the attention modules in the decoder. Mask values selected in ``[0, 1]``:
decoder_head_mask (`tf.Tensor` of shape `(decoder_layers, decoder_attention_heads)`, *optional*):
Mask to nullify selected heads of the attention modules in the decoder. Mask values selected in `[0, 1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
cross_attn_head_mask (:obj:`tf.Tensor` of shape :obj:`(decoder_layers, decoder_attention_heads)`, `optional`):
Mask to nullify selected heads of the cross-attention modules. Mask values selected in ``[0, 1]``:
cross_attn_head_mask (`tf.Tensor` of shape `(decoder_layers, decoder_attention_heads)`, *optional*):
Mask to nullify selected heads of the cross-attention modules. Mask values selected in `[0, 1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
encoder_outputs (:obj:`tf.FloatTensor`, `optional`):
encoder_outputs (`tf.FloatTensor`, *optional*):
hidden states at the output of the last layer of the encoder. Used in the cross-attention of the decoder.
of shape :obj:`(batch_size, sequence_length, hidden_size)` is a sequence of
past_key_values (:obj:`Tuple[Tuple[tf.Tensor]]` of length :obj:`config.n_layers`)
of shape `(batch_size, sequence_length, hidden_size)` is a sequence of
past_key_values (`Tuple[Tuple[tf.Tensor]]` of length `config.n_layers`)
contains precomputed key and value hidden states of the attention blocks. Can be used to speed up decoding.
If :obj:`past_key_values` are used, the user can optionally input only the last :obj:`decoder_input_ids`
(those that don't have their past key value states given to this model) of shape :obj:`(batch_size, 1)`
instead of all :obj:`decoder_input_ids` of shape :obj:`(batch_size, sequence_length)`.
use_cache (:obj:`bool`, `optional`, defaults to :obj:`True`):
If set to :obj:`True`, :obj:`past_key_values` key value states are returned and can be used to speed up
decoding (see :obj:`past_key_values`). Set to :obj:`False` during training, :obj:`True` during generation
output_attentions (:obj:`bool`, `optional`):
Whether or not to return the attentions tensors of all attention layers. See ``attentions`` under returned
If `past_key_values` are used, the user can optionally input only the last `decoder_input_ids`
(those that don't have their past key value states given to this model) of shape `(batch_size, 1)`
instead of all `decoder_input_ids` of shape `(batch_size, sequence_length)`.
use_cache (`bool`, *optional*, defaults to `True`):
If set to `True`, `past_key_values` key value states are returned and can be used to speed up
decoding (see `past_key_values`). Set to `False` during training, `True` during generation
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under returned
tensors for more detail. This argument can be used only in eager mode, in graph mode the value in the
config will be used instead.
output_hidden_states (:obj:`bool`, `optional`):
Whether or not to return the hidden states of all layers. See ``hidden_states`` under returned tensors for
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for
more detail. This argument can be used only in eager mode, in graph mode the value in the config will be
used instead.
return_dict (:obj:`bool`, `optional`):
Whether or not to return a :class:`~transformers.file_utils.ModelOutput` instead of a plain tuple. This
return_dict (`bool`, *optional*):
Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple. This
argument can be used in eager mode, in graph mode the value will always be set to True.
training (:obj:`bool`, `optional`, defaults to :obj:`False`):
training (`bool`, *optional*, defaults to `False`):
Whether or not to use the model in training mode (some modules like dropout modules have different
behaviors between training and evaluation).
"""
......@@ -2354,7 +2351,7 @@ class TF{{cookiecutter.camelcase_modelname}}Encoder(tf.keras.layers.Layer):
config_class = {{cookiecutter.camelcase_modelname}}Config
"""
Transformer encoder consisting of *config.encoder_layers* self attention layers. Each layer is a
:class:`TF{{cookiecutter.camelcase_modelname}}EncoderLayer`.
[`TF{{cookiecutter.camelcase_modelname}}EncoderLayer`].
Args:
config: {{cookiecutter.camelcase_modelname}}Config
......@@ -2398,44 +2395,43 @@ class TF{{cookiecutter.camelcase_modelname}}Encoder(tf.keras.layers.Layer):
):
"""
Args:
input_ids (:obj:`tf.Tensor` of shape :obj:`(batch_size, sequence_length)`):
input_ids (`tf.Tensor` of shape `(batch_size, sequence_length)`):
Indices of input sequence tokens in the vocabulary. Padding will be ignored by default should you
provide it.
Indices can be obtained using :class:`~transformers.{{cookiecutter.camelcase_modelname}}Tokenizer`. See
:meth:`transformers.PreTrainedTokenizer.encode` and :meth:`transformers.PreTrainedTokenizer.__call__`
Indices can be obtained using [`~{{cookiecutter.camelcase_modelname}}Tokenizer`]. See
[`PreTrainedTokenizer.encode`] and [`PreTrainedTokenizer.__call__`]
for details.
`What are input IDs? <../glossary.html#input-ids>`__
attention_mask (:obj:`tf.Tensor` of shape :obj:`(batch_size, sequence_length)`, `optional`):
Mask to avoid performing attention on padding token indices. Mask values selected in ``[0, 1]``:
[What are input IDs?](../glossary#input-ids)
attention_mask (`tf.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
`What are attention masks? <../glossary.html#attention-mask>`__
head_mask (:obj:`tf.Tensor` of shape :obj:`(encoder_layers, encoder_attention_heads)`, `optional):
Mask to nullify selected heads of the attention modules. Mask values selected in ``[0, 1]``:
[What are attention masks?](../glossary#attention-mask)
head_mask (`tf.Tensor` of shape `(encoder_layers, encoder_attention_heads)`, `optional): Mask to nullify selected heads of the attention modules. Mask values selected in `[0, 1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
inputs_embeds (:obj:`tf.Tensor` of shape :obj:`(batch_size, sequence_length, hidden_size)`, `optional`):
Optionally, instead of passing :obj:`input_ids` you can choose to directly pass an embedded
representation. This is useful if you want more control over how to convert :obj:`input_ids` indices
inputs_embeds (`tf.Tensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*):
Optionally, instead of passing `input_ids` you can choose to directly pass an embedded
representation. This is useful if you want more control over how to convert `input_ids` indices
into associated vectors than the model's internal embedding lookup matrix.
output_attentions (:obj:`bool`, `optional`):
Whether or not to return the attentions tensors of all attention layers. See ``attentions`` under
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under
returned tensors for more detail. This argument can be used only in eager mode, in graph mode the value
in the config will be used instead.
output_hidden_states (:obj:`bool`, `optional`):
Whether or not to return the hidden states of all layers. See ``hidden_states`` under returned tensors
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors
for more detail. This argument can be used only in eager mode, in graph mode the value in the config
will be used instead.
return_dict (:obj:`bool`, `optional`):
Whether or not to return a :class:`~transformers.file_utils.ModelOutput` instead of a plain tuple. This
return_dict (`bool`, *optional*):
Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple. This
argument can be used in eager mode, in graph mode the value will always be set to True.
training (:obj:`bool`, `optional`, defaults to :obj:`False`):
training (`bool`, *optional*, defaults to `False`):
Whether or not to use the model in training mode (some modules like dropout modules have different
behaviors between training and evaluation).
"""
......@@ -2521,7 +2517,7 @@ class TF{{cookiecutter.camelcase_modelname}}Encoder(tf.keras.layers.Layer):
class TF{{cookiecutter.camelcase_modelname}}Decoder(tf.keras.layers.Layer):
config_class = {{cookiecutter.camelcase_modelname}}Config
"""
Transformer decoder consisting of *config.decoder_layers* layers. Each layer is a :class:`TF{{cookiecutter.camelcase_modelname}}DecoderLayer`
Transformer decoder consisting of *config.decoder_layers* layers. Each layer is a [`TF{{cookiecutter.camelcase_modelname}}DecoderLayer`]
Args:
config: {{cookiecutter.camelcase_modelname}}Config
......@@ -2570,69 +2566,66 @@ class TF{{cookiecutter.camelcase_modelname}}Decoder(tf.keras.layers.Layer):
):
r"""
Args:
input_ids (:obj:`tf.Tensor` of shape :obj:`(batch_size, sequence_length)`):
input_ids (`tf.Tensor` of shape `(batch_size, sequence_length)`):
Indices of input sequence tokens in the vocabulary. Padding will be ignored by default should you
provide it.
Indices can be obtained using :class:`~transformers.{{cookiecutter.camelcase_modelname}}Tokenizer`. See
:meth:`transformers.PreTrainedTokenizer.encode` and :meth:`transformers.PreTrainedTokenizer.__call__`
Indices can be obtained using [`~{{cookiecutter.camelcase_modelname}}Tokenizer`]. See
[`PreTrainedTokenizer.encode`] and [`PreTrainedTokenizer.__call__`]
for details.
`What are input IDs? <../glossary.html#input-ids>`__
attention_mask (:obj:`tf.Tensor` of shape :obj:`(batch_size, sequence_length)`, `optional`):
Mask to avoid performing attention on padding token indices. Mask values selected in ``[0, 1]``:
[What are input IDs?](../glossary#input-ids)
attention_mask (`tf.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
`What are attention masks? <../glossary.html#attention-mask>`__
encoder_hidden_states (:obj:`tf.Tensor` of shape :obj:`(batch_size, encoder_sequence_length, hidden_size)`, `optional`):
[What are attention masks?](../glossary#attention-mask)
encoder_hidden_states (`tf.Tensor` of shape `(batch_size, encoder_sequence_length, hidden_size)`, *optional*):
Sequence of hidden-states at the output of the last layer of the encoder. Used in the cross-attention
of the decoder.
encoder_attention_mask (:obj:`tf.Tensor` of shape :obj:`(batch_size, encoder_sequence_length)`, `optional`):
encoder_attention_mask (`tf.Tensor` of shape `(batch_size, encoder_sequence_length)`, *optional*):
Mask to avoid performing cross-attention on padding tokens indices of encoder input_ids. Mask values
selected in ``[0, 1]``:
selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
`What are attention masks? <../glossary.html#attention-mask>`__
head_mask (:obj:`tf.Tensor` of shape :obj:`(decoder_layers, decoder_attention_heads)`, `optional`):
Mask to nullify selected heads of the attention modules. Mask values selected in ``[0, 1]``:
[What are attention masks?](../glossary#attention-mask)
head_mask (`tf.Tensor` of shape `(decoder_layers, decoder_attention_heads)`, *optional*):
Mask to nullify selected heads of the attention modules. Mask values selected in `[0, 1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
cross_attn_head_mask (:obj:`tf.Tensor` of shape :obj:`(decoder_layers, decoder_attention_heads)`, `optional`):
Mask to nullify selected heads of the cross-attention modules. Mask values selected in ``[0, 1]``:
cross_attn_head_mask (`tf.Tensor` of shape `(decoder_layers, decoder_attention_heads)`, *optional*):
Mask to nullify selected heads of the cross-attention modules. Mask values selected in `[0, 1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
past_key_values (:obj:`Tuple[Tuple[tf.Tensor]]` of length :obj:`config.n_layers` with each tuple having 2 tuples each of which has 2 tensors of shape :obj:`(batch_size, num_heads, sequence_length - 1, embed_size_per_head)`):
past_key_values (`Tuple[Tuple[tf.Tensor]]` of length `config.n_layers` with each tuple having 2 tuples each of which has 2 tensors of shape `(batch_size, num_heads, sequence_length - 1, embed_size_per_head)`):
Contains precomputed key and value hidden-states of the attention blocks. Can be used to speed up
decoding.
If :obj:`past_key_values` are used, the user can optionally input only the last
:obj:`decoder_input_ids` (those that don't have their past key value states given to this model) of
shape :obj:`(batch_size, 1)` instead of all :obj:`decoder_input_ids`` of shape :obj:`(batch_size,
sequence_length)`.
inputs_embeds (:obj:`tf.Tensor` of shape :obj:`(batch_size, sequence_length, hidden_size)`, `optional`):
Optionally, instead of passing :obj:`input_ids` you can choose to directly pass an embedded
representation. This is useful if you want more control over how to convert :obj:`input_ids` indices
If `past_key_values` are used, the user can optionally input only the last
`decoder_input_ids` (those that don't have their past key value states given to this model) of
shape `(batch_size, 1)` instead of all ``decoder_input_ids``` of shape `(batch_size,
sequence_length)`. inputs_embeds (`tf.Tensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*): Optionally, instead of passing `input_ids` you can choose to directly pass an embedded representation. This is useful if you want more control over how to convert `input_ids` indices
into associated vectors than the model's internal embedding lookup matrix.
output_attentions (:obj:`bool`, `optional`):
Whether or not to return the attentions tensors of all attention layers. See ``attentions`` under
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under
returned tensors for more detail. This argument can be used only in eager mode, in graph mode the value
in the config will be used instead.
output_hidden_states (:obj:`bool`, `optional`):
Whether or not to return the hidden states of all layers. See ``hidden_states`` under returned tensors
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors
for more detail. This argument can be used only in eager mode, in graph mode the value in the config
will be used instead.
return_dict (:obj:`bool`, `optional`):
Whether or not to return a :class:`~transformers.file_utils.ModelOutput` instead of a plain tuple. This
return_dict (`bool`, *optional*):
Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple. This
argument can be used in eager mode, in graph mode the value will always be set to True.
training (:obj:`bool`, `optional`, defaults to :obj:`False`):
training (`bool`, *optional*, defaults to `False`):
Whether or not to use the model in training mode (some modules like dropout modules have different
behaviors between training and evaluation).
"""
......
templates/adding_a_new_model/cookiecutter-template-{{cookiecutter.modelname}}/modeling_{{cookiecutter.lowercase_modelname}}.py
View file @ 7af80f66
......@@ -689,64 +689,63 @@ class {{cookiecutter.camelcase_modelname}}PreTrainedModel(PreTrainedModel):
{{cookiecutter.uppercase_modelname}}_START_DOCSTRING = r"""
This model is a PyTorch `torch.nn.Module <https://pytorch.org/docs/stable/nn.html#torch.nn.Module>`_ sub-class.
This model is a PyTorch [torch.nn.Module](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) sub-class.
Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general
usage and behavior.
Parameters:
config (:class:`~transformers.{{cookiecutter.camelcase_modelname}}Config`): Model configuration class with all the parameters of the model.
config ([`~{{cookiecutter.camelcase_modelname}}Config`]): Model configuration class with all the parameters of the model.
Initializing with a config file does not load the weights associated with the model, only the configuration.
Check out the :meth:`~transformers.PreTrainedModel.from_pretrained` method to load the model weights.
Check out the [`~PreTrainedModel.from_pretrained`] method to load the model weights.
"""
{{cookiecutter.uppercase_modelname}}_INPUTS_DOCSTRING = r"""
Args:
input_ids (:obj:`torch.LongTensor` of shape :obj:`({0})`):
input_ids (`torch.LongTensor` of shape `({0})`):
Indices of input sequence tokens in the vocabulary.
Indices can be obtained using :class:`transformers.{{cookiecutter.camelcase_modelname}}Tokenizer`.
See :func:`transformers.PreTrainedTokenizer.encode` and
:func:`transformers.PreTrainedTokenizer.__call__` for details.
Indices can be obtained using [`{{cookiecutter.camelcase_modelname}}Tokenizer`].
See [`PreTrainedTokenizer.encode`] and
[`PreTrainedTokenizer.__call__`] for details.
`What are input IDs? <../glossary.html#input-ids>`__
attention_mask (:obj:`torch.FloatTensor` of shape :obj:`({0})`, `optional`):
Mask to avoid performing attention on padding token indices. Mask values selected in ``[0, 1]``:
[What are input IDs?](../glossary#input-ids)
attention_mask (`torch.FloatTensor` of shape `({0})`, *optional*):
Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
`What are attention masks? <../glossary.html#attention-mask>`__
token_type_ids (:obj:`torch.LongTensor` of shape :obj:`({0})`, `optional`):
Segment token indices to indicate first and second portions of the inputs. Indices are selected in ``[0,
1]``:
[What are attention masks?](../glossary#attention-mask)
token_type_ids (`torch.LongTensor` of shape `({0})`, *optional*):
Segment token indices to indicate first and second portions of the inputs. Indices are selected in `[0, 1]`:
- 0 corresponds to a `sentence A` token,
- 1 corresponds to a `sentence B` token.
- 0 corresponds to a *sentence A* token,
- 1 corresponds to a *sentence B* token.
`What are token type IDs? <../glossary.html#token-type-ids>`_
position_ids (:obj:`torch.LongTensor` of shape :obj:`({0})`, `optional`):
[What are token type IDs?](../glossary#token-type-ids)
position_ids (`torch.LongTensor` of shape `({0})`, *optional*):
Indices of positions of each input sequence tokens in the position embeddings.
Selected in the range ``[0, config.max_position_embeddings - 1]``.
Selected in the range `[0, config.max_position_embeddings - 1]`.
`What are position IDs? <../glossary.html#position-ids>`_
head_mask (:obj:`torch.FloatTensor` of shape :obj:`(num_heads,)` or :obj:`(num_layers, num_heads)`, `optional`):
Mask to nullify selected heads of the self-attention modules. Mask values selected in ``[0, 1]``:
[What are position IDs?](../glossary#position-ids)
head_mask (`torch.FloatTensor` of shape `(num_heads,)` or `(num_layers, num_heads)`, *optional*):
Mask to nullify selected heads of the self-attention modules. Mask values selected in `[0, 1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
inputs_embeds (:obj:`torch.FloatTensor` of shape :obj:`({0}, hidden_size)`, `optional`):
Optionally, instead of passing :obj:`input_ids` you can choose to directly pass an embedded representation.
This is useful if you want more control over how to convert `input_ids` indices into associated vectors
inputs_embeds (`torch.FloatTensor` of shape `({0}, hidden_size)`, *optional*):
Optionally, instead of passing `input_ids` you can choose to directly pass an embedded representation.
This is useful if you want more control over how to convert *input_ids* indices into associated vectors
than the model's internal embedding lookup matrix.
output_attentions (:obj:`bool`, `optional`):
Whether or not to return the attentions tensors of all attention layers. See ``attentions`` under returned
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under returned
tensors for more detail.
output_hidden_states (:obj:`bool`, `optional`):
Whether or not to return the hidden states of all layers. See ``hidden_states`` under returned tensors for
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for
more detail.
return_dict (:obj:`bool`, `optional`):
Whether or not to return a :class:`~transformers.file_utils.ModelOutput` instead of a plain tuple.
return_dict (`bool`, *optional*):
Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple.
"""
......@@ -759,15 +758,15 @@ class {{cookiecutter.camelcase_modelname}}Model({{cookiecutter.camelcase_modelna
The model can behave as an encoder (with only self-attention) as well
as a decoder, in which case a layer of cross-attention is added between
the self-attention layers, following the architecture described in `Attention is
all you need <https://arxiv.org/abs/1706.03762>`__ by Ashish Vaswani,
the self-attention layers, following the architecture described in [Attention is
all you need](https://arxiv.org/abs/1706.03762) by Ashish Vaswani,
Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N. Gomez, Lukasz Kaiser and Illia Polosukhin.
To behave as an decoder the model needs to be initialized with the
:obj:`is_decoder` argument of the configuration set to :obj:`True`.
To be used in a Seq2Seq model, the model needs to initialized with both :obj:`is_decoder`
argument and :obj:`add_cross_attention` set to :obj:`True`; an
:obj:`encoder_hidden_states` is then expected as an input to the forward pass.
`is_decoder` argument of the configuration set to `True`.
To be used in a Seq2Seq model, the model needs to initialized with both `is_decoder`
argument and `add_cross_attention` set to `True`; an
`encoder_hidden_states` is then expected as an input to the forward pass.
"""
def __init__(self, config):
......@@ -818,24 +817,24 @@ class {{cookiecutter.camelcase_modelname}}Model({{cookiecutter.camelcase_modelna
return_dict=None,
):
r"""
encoder_hidden_states (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length, hidden_size)`, `optional`):
encoder_hidden_states (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*):
Sequence of hidden-states at the output of the last layer of the encoder. Used in the cross-attention
if the model is configured as a decoder.
encoder_attention_mask (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`):
encoder_attention_mask (`torch.FloatTensor` of shape `(batch_size, sequence_length)`, *optional*):
Mask to avoid performing attention on the padding token indices of the encoder input. This mask
is used in the cross-attention if the model is configured as a decoder.
Mask values selected in ``[0, 1]``:
Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
past_key_values (:obj:`tuple(tuple(torch.FloatTensor))` of length :obj:`config.n_layers` with each tuple having 4 tensors of shape :obj:`(batch_size, num_heads, sequence_length - 1, embed_size_per_head)`):
past_key_values (`tuple(tuple(torch.FloatTensor))` of length `config.n_layers` with each tuple having 4 tensors of shape `(batch_size, num_heads, sequence_length - 1, embed_size_per_head)`):
Contains precomputed key and value hidden states of the attention blocks. Can be used to speed up decoding.
If :obj:`past_key_values` are used, the user can optionally input only the last :obj:`decoder_input_ids`
(those that don't have their past key value states given to this model) of shape :obj:`(batch_size, 1)`
instead of all :obj:`decoder_input_ids` of shape :obj:`(batch_size, sequence_length)`.
use_cache (:obj:`bool`, `optional`):
If set to :obj:`True`, :obj:`past_key_values` key value states are returned and can be used to speed up
decoding (see :obj:`past_key_values`).
If `past_key_values` are used, the user can optionally input only the last `decoder_input_ids`
(those that don't have their past key value states given to this model) of shape `(batch_size, 1)`
instead of all `decoder_input_ids` of shape `(batch_size, sequence_length)`.
use_cache (`bool`, *optional*):
If set to `True`, `past_key_values` key value states are returned and can be used to speed up
decoding (see `past_key_values`).
"""
output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
output_hidden_states = (
......@@ -976,11 +975,11 @@ class {{cookiecutter.camelcase_modelname}}ForMaskedLM({{cookiecutter.camelcase_m
return_dict=None,
):
r"""
labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`):
labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
Labels for computing the masked language modeling loss.
Indices should be in ``[-100, 0, ..., config.vocab_size]`` (see ``input_ids`` docstring)
Tokens with indices set to ``-100`` are ignored (masked), the loss is only computed for the tokens with labels
in ``[0, ..., config.vocab_size]``.
Indices should be in `[-100, 0, ..., config.vocab_size]` (see `input_ids` docstring)
Tokens with indices set to `-100` are ignored (masked), the loss is only computed for the tokens with labels
in `[0, ..., config.vocab_size]`.
"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
......@@ -1078,54 +1077,56 @@ class {{cookiecutter.camelcase_modelname}}ForCausalLM({{cookiecutter.camelcase_m
return_dict=None,
):
r"""
encoder_hidden_states (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length, hidden_size)`, `optional`):
encoder_hidden_states (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*):
Sequence of hidden-states at the output of the last layer of the encoder. Used in the cross-attention if
the model is configured as a decoder.
encoder_attention_mask (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`):
encoder_attention_mask (`torch.FloatTensor` of shape `(batch_size, sequence_length)`, *optional*):
Mask to avoid performing attention on the padding token indices of the encoder input. This mask is used in
the cross-attention if the model is configured as a decoder. Mask values selected in ``[0, 1]``:
the cross-attention if the model is configured as a decoder. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
past_key_values (:obj:`tuple(tuple(torch.FloatTensor))`, `optional`, returned when ``use_cache=True`` is passed or when ``config.use_cache=True``):
Tuple of :obj:`tuple(torch.FloatTensor)` of length :obj:`config.n_layers`, with each tuple having 2
tensors of shape :obj:`(batch_size, num_heads, sequence_length, embed_size_per_head)`) and 2 additional
tensors of shape :obj:`(batch_size, num_heads, encoder_sequence_length, embed_size_per_head)`. The two
past_key_values (`tuple(tuple(torch.FloatTensor))`, *optional*, returned when `use_cache=True` is passed or when `config.use_cache=True`):
Tuple of `tuple(torch.FloatTensor)` of length `config.n_layers`, with each tuple having 2
tensors of shape `(batch_size, num_heads, sequence_length, embed_size_per_head)`) and 2 additional
tensors of shape `(batch_size, num_heads, encoder_sequence_length, embed_size_per_head)`. The two
additional tensors are only required when the model is used as a decoder in a Sequence to Sequence
model.
Contains pre-computed hidden-states (key and values in the self-attention blocks and in the
cross-attention blocks) that can be used (see :obj:`past_key_values` input) to speed up sequential
cross-attention blocks) that can be used (see `past_key_values` input) to speed up sequential
decoding.
If :obj:`past_key_values` are used, the user can optionally input only the last ``decoder_input_ids``
(those that don't have their past key value states given to this model) of shape :obj:`(batch_size, 1)`
instead of all ``decoder_input_ids`` of shape :obj:`(batch_size, sequence_length)`.
labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`):
If `past_key_values` are used, the user can optionally input only the last `decoder_input_ids`
(those that don't have their past key value states given to this model) of shape `(batch_size, 1)`
instead of all `decoder_input_ids` of shape `(batch_size, sequence_length)`.
labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
Labels for computing the left-to-right language modeling loss (next word prediction). Indices should be in
``[-100, 0, ..., config.vocab_size]`` (see ``input_ids`` docstring) Tokens with indices set to ``-100`` are
ignored (masked), the loss is only computed for the tokens with labels n ``[0, ..., config.vocab_size]``.
use_cache (:obj:`bool`, `optional`):
If set to :obj:`True`, :obj:`past_key_values` key value states are returned and can be used to speed up
decoding (see :obj:`past_key_values`).
`[-100, 0, ..., config.vocab_size]` (see `input_ids` docstring) Tokens with indices set to `-100` are
ignored (masked), the loss is only computed for the tokens with labels n `[0, ..., config.vocab_size]`.
use_cache (`bool`, *optional*):
If set to `True`, `past_key_values` key value states are returned and can be used to speed up
decoding (see `past_key_values`).
Returns:
Example::
Example:
>>> from transformers import {{cookiecutter.camelcase_modelname}}Tokenizer, {{cookiecutter.camelcase_modelname}}ForCausalLM, {{cookiecutter.camelcase_modelname}}Config
>>> import torch
```python
>>> from transformers import {{cookiecutter.camelcase_modelname}}Tokenizer, {{cookiecutter.camelcase_modelname}}ForCausalLM, {{cookiecutter.camelcase_modelname}}Config
>>> import torch
>>> tokenizer = {{cookiecutter.camelcase_modelname}}Tokenizer.from_pretrained('{{cookiecutter.checkpoint_identifier}}')
>>> config = {{cookiecutter.camelcase_modelname}}Config.from_pretrained("{{cookiecutter.checkpoint_identifier}}")
>>> config.is_decoder = True
>>> model = {{cookiecutter.camelcase_modelname}}ForCausalLM.from_pretrained('{{cookiecutter.checkpoint_identifier}}', config=config)
>>> tokenizer = {{cookiecutter.camelcase_modelname}}Tokenizer.from_pretrained('{{cookiecutter.checkpoint_identifier}}')
>>> config = {{cookiecutter.camelcase_modelname}}Config.from_pretrained("{{cookiecutter.checkpoint_identifier}}")
>>> config.is_decoder = True
>>> model = {{cookiecutter.camelcase_modelname}}ForCausalLM.from_pretrained('{{cookiecutter.checkpoint_identifier}}', config=config)
>>> inputs = tokenizer("Hello, my dog is cute", return_tensors="pt")
>>> outputs = model(**inputs)
>>> inputs = tokenizer("Hello, my dog is cute", return_tensors="pt")
>>> outputs = model(**inputs)
>>> prediction_logits = outputs.logits
"""
>>> prediction_logits = outputs.logits
```
"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
outputs = self.{{cookiecutter.lowercase_modelname}}(
......@@ -1244,11 +1245,11 @@ class {{cookiecutter.camelcase_modelname}}ForSequenceClassification({{cookiecutt
return_dict=None,
):
r"""
labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`):
labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
Labels for computing the sequence classification/regression loss.
Indices should be in :obj:`[0, ..., config.num_labels - 1]`.
If :obj:`config.num_labels == 1` a regression loss is computed (Mean-Square loss),
If :obj:`config.num_labels > 1` a classification loss is computed (Cross-Entropy).
Indices should be in `[0, ..., config.num_labels - 1]`.
If `config.num_labels == 1` a regression loss is computed (Mean-Square loss),
If `config.num_labels > 1` a classification loss is computed (Cross-Entropy).
"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
......@@ -1337,10 +1338,10 @@ class {{cookiecutter.camelcase_modelname}}ForMultipleChoice({{cookiecutter.camel
return_dict=None,
):
r"""
labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`):
labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
Labels for computing the multiple choice classification loss.
Indices should be in ``[0, ..., num_choices-1]`` where :obj:`num_choices` is the size of the second dimension
of the input tensors. (See :obj:`input_ids` above)
Indices should be in `[0, ..., num_choices-1]` where `num_choices` is the size of the second dimension
of the input tensors. (See `input_ids` above)
"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
num_choices = input_ids.shape[1] if input_ids is not None else inputs_embeds.shape[1]
......@@ -1428,9 +1429,9 @@ class {{cookiecutter.camelcase_modelname}}ForTokenClassification({{cookiecutter.
return_dict=None,
):
r"""
labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`):
labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
Labels for computing the token classification loss.
Indices should be in ``[0, ..., config.num_labels - 1]``.
Indices should be in `[0, ..., config.num_labels - 1]`.
"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
......@@ -1517,13 +1518,13 @@ class {{cookiecutter.camelcase_modelname}}ForQuestionAnswering({{cookiecutter.ca
return_dict=None,
):
r"""
start_positions (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`):
start_positions (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
Labels for position (index) of the start of the labelled span for computing the token classification loss.
Positions are clamped to the length of the sequence (:obj:`sequence_length`).
Positions are clamped to the length of the sequence (`sequence_length`).
Position outside of the sequence are not taken into account for computing the loss.
end_positions (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`):
end_positions (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
Labels for position (index) of the end of the labelled span for computing the token classification loss.
Positions are clamped to the length of the sequence (:obj:`sequence_length`).
Positions are clamped to the length of the sequence (`sequence_length`).
Position outside of the sequence are not taken into account for computing the loss.
"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
......@@ -1846,13 +1847,13 @@ class {{cookiecutter.camelcase_modelname}}EncoderLayer(nn.Module):
):
"""
Args:
hidden_states (:obj:`torch.FloatTensor`): input to the layer of shape `(seq_len, batch, embed_dim)`
attention_mask (:obj:`torch.FloatTensor`): attention mask of size
`(batch, 1, tgt_len, src_len)` where padding elements are indicated by very large negative values.
layer_head_mask (:obj:`torch.FloatTensor`): mask for attention heads in a given layer of size
`(config.encoder_attention_heads,)`.
output_attentions (:obj:`bool`, `optional`):
Whether or not to return the attentions tensors of all attention layers. See ``attentions`` under
hidden_states (`torch.FloatTensor`): input to the layer of shape *(seq_len, batch, embed_dim)*
attention_mask (`torch.FloatTensor`): attention mask of size
*(batch, 1, tgt_len, src_len)* where padding elements are indicated by very large negative values.
layer_head_mask (`torch.FloatTensor`): mask for attention heads in a given layer of size
*(config.encoder_attention_heads,)*.
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under
returned tensors for more detail.
"""
residual = hidden_states
......@@ -1927,19 +1928,19 @@ class {{cookiecutter.camelcase_modelname}}DecoderLayer(nn.Module):
):
"""
Args:
hidden_states (:obj:`torch.FloatTensor`): input to the layer of shape `(seq_len, batch, embed_dim)`
attention_mask (:obj:`torch.FloatTensor`): attention mask of size
`(batch, 1, tgt_len, src_len)` where padding elements are indicated by very large negative values.
encoder_hidden_states (:obj:`torch.FloatTensor`): cross attention input to the layer of shape `(seq_len, batch, embed_dim)`
encoder_attention_mask (:obj:`torch.FloatTensor`): encoder attention mask of size
`(batch, 1, tgt_len, src_len)` where padding elements are indicated by very large negative values.
layer_head_mask (:obj:`torch.FloatTensor`): mask for attention heads in a given layer of size
`(encoder_attention_heads,)`.
cross_layer_head_mask (:obj:`torch.FloatTensor`): mask for cross-attention heads in a given layer of
size `(decoder_attention_heads,)`.
past_key_value (:obj:`Tuple(torch.FloatTensor)`): cached past key and value projection states
output_attentions (:obj:`bool`, `optional`):
Whether or not to return the attentions tensors of all attention layers. See ``attentions`` under
hidden_states (`torch.FloatTensor`): input to the layer of shape *(seq_len, batch, embed_dim)*
attention_mask (`torch.FloatTensor`): attention mask of size
*(batch, 1, tgt_len, src_len)* where padding elements are indicated by very large negative values.
encoder_hidden_states (`torch.FloatTensor`): cross attention input to the layer of shape *(seq_len, batch, embed_dim)*
encoder_attention_mask (`torch.FloatTensor`): encoder attention mask of size
*(batch, 1, tgt_len, src_len)* where padding elements are indicated by very large negative values.
layer_head_mask (`torch.FloatTensor`): mask for attention heads in a given layer of size
*(encoder_attention_heads,)*.
cross_layer_head_mask (`torch.FloatTensor`): mask for cross-attention heads in a given layer of
size *(decoder_attention_heads,)*.
past_key_value (`Tuple(torch.FloatTensor)`): cached past key and value projection states
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under
returned tensors for more detail.
"""
residual = hidden_states
......@@ -2049,19 +2050,19 @@ class {{cookiecutter.camelcase_modelname}}PreTrainedModel(PreTrainedModel):
{{cookiecutter.uppercase_modelname}}_START_DOCSTRING = r"""
This model inherits from :class:`~transformers.PreTrainedModel`. Check the superclass documentation for the generic
This model inherits from [`PreTrainedModel`]. Check the superclass documentation for the generic
methods the library implements for all its model (such as downloading or saving, resizing the input embeddings,
pruning heads etc.)
This model is also a PyTorch `torch.nn.Module <https://pytorch.org/docs/stable/nn.html#torch.nn.Module>`__
This model is also a PyTorch [torch.nn.Module](https://pytorch.org/docs/stable/nn.html#torch.nn.Module)
subclass. Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to
general usage and behavior.
Parameters:
config (:class:`~transformers.{{cookiecutter.camelcase_modelname}}Config`):
config ([`~{{cookiecutter.camelcase_modelname}}Config`]):
Model configuration class with all the parameters of the model.
Initializing with a config file does not load the weights associated with the model, only the
configuration. Check out the :meth:`~transformers.PreTrainedModel.from_pretrained` method to load the model
configuration. Check out the [`~PreTrainedModel.from_pretrained`] method to load the model
weights.
"""
......@@ -2083,125 +2084,122 @@ class {{cookiecutter.camelcase_modelname}}PreTrainedModel(PreTrainedModel):
{{cookiecutter.uppercase_modelname}}_INPUTS_DOCSTRING = r"""
Args:
input_ids (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`):
input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`):
Indices of input sequence tokens in the vocabulary. Padding will be ignored by default should you provide
it.
Indices can be obtained using :class:`~transformers.{{cookiecutter.camelcase_modelname}}Tokenizer`. See
:meth:`transformers.PreTrainedTokenizer.encode` and :meth:`transformers.PreTrainedTokenizer.__call__` for
Indices can be obtained using [`~{{cookiecutter.camelcase_modelname}}Tokenizer`]. See
[`PreTrainedTokenizer.encode`] and [`PreTrainedTokenizer.__call__`] for
details.
`What are input IDs? <../glossary.html#input-ids>`__
attention_mask (:obj:`torch.Tensor` of shape :obj:`(batch_size, sequence_length)`, `optional`):
Mask to avoid performing attention on padding token indices. Mask values selected in ``[0, 1]``:
[What are input IDs?](../glossary#input-ids)
attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
`What are attention masks? <../glossary.html#attention-mask>`__
decoder_input_ids (:obj:`torch.LongTensor` of shape :obj:`(batch_size, target_sequence_length)`, `optional`):
[What are attention masks?](../glossary#attention-mask)
decoder_input_ids (`torch.LongTensor` of shape `(batch_size, target_sequence_length)`, *optional*):
Provide for translation and summarization training. By default, the model will create this tensor by
shifting the :obj:`input_ids` to the right, following the paper.
decoder_attention_mask (:obj:`torch.LongTensor` of shape :obj:`(batch_size, target_sequence_length)`, `optional`):
Default behavior: generate a tensor that ignores pad tokens in :obj:`decoder_input_ids`. Causal mask will
shifting the `input_ids` to the right, following the paper.
decoder_attention_mask (`torch.LongTensor` of shape `(batch_size, target_sequence_length)`, *optional*):
Default behavior: generate a tensor that ignores pad tokens in `decoder_input_ids`. Causal mask will
also be used by default.
If you want to change padding behavior, you should read :func:`modeling_{{cookiecutter.lowercase_modelname}}._prepare_decoder_inputs` and
modify to your needs. See diagram 1 in `the paper <https://arxiv.org/abs/1910.13461>`__ for more
If you want to change padding behavior, you should read [`modeling_{{cookiecutter.lowercase_modelname}}._prepare_decoder_inputs`] and
modify to your needs. See diagram 1 in [the paper](https://arxiv.org/abs/1910.13461) for more
information on the default strategy.
head_mask (:obj:`torch.Tensor` of shape :obj:`(encoder_layers, encoder_attention_heads)`, `optional`):
Mask to nullify selected heads of the attention modules in the encoder. Mask values selected in ``[0, 1]``:
head_mask (`torch.Tensor` of shape `(encoder_layers, encoder_attention_heads)`, *optional*):
Mask to nullify selected heads of the attention modules in the encoder. Mask values selected in `[0, 1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
decoder_head_mask (:obj:`torch.Tensor` of shape :obj:`(decoder_layers, decoder_attention_heads)`, `optional`):
Mask to nullify selected heads of the attention modules in the decoder. Mask values selected in ``[0, 1]``:
decoder_head_mask (`torch.Tensor` of shape `(decoder_layers, decoder_attention_heads)`, *optional*):
Mask to nullify selected heads of the attention modules in the decoder. Mask values selected in `[0, 1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
cross_attn_head_mask (:obj:`torch.Tensor` of shape :obj:`(decoder_layers, decoder_attention_heads)`, `optional`):
Mask to nullify selected heads of the cross-attention modules. Mask values selected in ``[0, 1]``:
cross_attn_head_mask (`torch.Tensor` of shape `(decoder_layers, decoder_attention_heads)`, *optional*):
Mask to nullify selected heads of the cross-attention modules. Mask values selected in `[0, 1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
encoder_outputs (:obj:`tuple(tuple(torch.FloatTensor)`, `optional`):
Tuple consists of (:obj:`last_hidden_state`, `optional`: :obj:`hidden_states`, `optional`:
:obj:`attentions`) :obj:`last_hidden_state` of shape :obj:`(batch_size, sequence_length, hidden_size)`,
`optional`) is a sequence of hidden-states at the output of the last layer of the encoder. Used in the
encoder_outputs (`tuple(tuple(torch.FloatTensor)`, *optional*):
Tuple consists of (`last_hidden_state`, *optional*: `hidden_states`, *optional*:
`attentions`) `last_hidden_state` of shape `(batch_size, sequence_length, hidden_size)`,
*optional*) is a sequence of hidden-states at the output of the last layer of the encoder. Used in the
cross-attention of the decoder.
past_key_values (:obj:`tuple(tuple(torch.FloatTensor))`, `optional`, returned when ``use_cache=True`` is passed or when ``config.use_cache=True``):
Tuple of :obj:`tuple(torch.FloatTensor)` of length :obj:`config.n_layers`, with each tuple having 2 tensors
of shape :obj:`(batch_size, num_heads, sequence_length, embed_size_per_head)`) and 2 additional tensors of
shape :obj:`(batch_size, num_heads, encoder_sequence_length, embed_size_per_head)`.
past_key_values (`tuple(tuple(torch.FloatTensor))`, *optional*, returned when `use_cache=True` is passed or when `config.use_cache=True`):
Tuple of `tuple(torch.FloatTensor)` of length `config.n_layers`, with each tuple having 2 tensors
of shape `(batch_size, num_heads, sequence_length, embed_size_per_head)`) and 2 additional tensors of
shape `(batch_size, num_heads, encoder_sequence_length, embed_size_per_head)`.
Contains pre-computed hidden-states (key and values in the self-attention blocks and in the cross-attention
blocks) that can be used (see :obj:`past_key_values` input) to speed up sequential decoding.
If :obj:`past_key_values` are used, the user can optionally input only the last :obj:`decoder_input_ids`
(those that don't have their past key value states given to this model) of shape :obj:`(batch_size, 1)`
instead of all :obj:`decoder_input_ids`` of shape :obj:`(batch_size, sequence_length)`.
inputs_embeds (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length, hidden_size)`, `optional`):
Optionally, instead of passing :obj:`input_ids` you can choose to directly pass an embedded representation.
This is useful if you want more control over how to convert :obj:`input_ids` indices into associated
blocks) that can be used (see `past_key_values` input) to speed up sequential decoding.
If `past_key_values` are used, the user can optionally input only the last `decoder_input_ids`
(those that don't have their past key value states given to this model) of shape `(batch_size, 1)`
instead of all ``decoder_input_ids``` of shape `(batch_size, sequence_length)`. inputs_embeds (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*): Optionally, instead of passing `input_ids` you can choose to directly pass an embedded representation. This is useful if you want more control over how to convert `input_ids` indices into associated
vectors than the model's internal embedding lookup matrix.
decoder_inputs_embeds (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, target_sequence_length, hidden_size)`, `optional`):
Optionally, instead of passing :obj:`decoder_input_ids` you can choose to directly pass an embedded
representation. If :obj:`past_key_values` is used, optionally only the last :obj:`decoder_inputs_embeds`
have to be input (see :obj:`past_key_values`). This is useful if you want more control over how to convert
:obj:`decoder_input_ids` indices into associated vectors than the model's internal embedding lookup matrix.
If :obj:`decoder_input_ids` and :obj:`decoder_inputs_embeds` are both unset, :obj:`decoder_inputs_embeds`
takes the value of :obj:`inputs_embeds`.
use_cache (:obj:`bool`, `optional`):
If set to :obj:`True`, :obj:`past_key_values` key value states are returned and can be used to speed up
decoding (see :obj:`past_key_values`).
output_attentions (:obj:`bool`, `optional`):
Whether or not to return the attentions tensors of all attention layers. See ``attentions`` under returned
decoder_inputs_embeds (`torch.FloatTensor` of shape `(batch_size, target_sequence_length, hidden_size)`, *optional*):
Optionally, instead of passing `decoder_input_ids` you can choose to directly pass an embedded
representation. If `past_key_values` is used, optionally only the last `decoder_inputs_embeds`
have to be input (see `past_key_values`). This is useful if you want more control over how to convert
`decoder_input_ids` indices into associated vectors than the model's internal embedding lookup matrix.
If `decoder_input_ids` and `decoder_inputs_embeds` are both unset, `decoder_inputs_embeds`
takes the value of `inputs_embeds`.
use_cache (`bool`, *optional*):
If set to `True`, `past_key_values` key value states are returned and can be used to speed up
decoding (see `past_key_values`).
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under returned
tensors for more detail.
output_hidden_states (:obj:`bool`, `optional`):
Whether or not to return the hidden states of all layers. See ``hidden_states`` under returned tensors for
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for
more detail.
return_dict (:obj:`bool`, `optional`):
Whether or not to return a :class:`~transformers.file_utils.ModelOutput` instead of a plain tuple.
return_dict (`bool`, *optional*):
Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple.
"""
{{cookiecutter.uppercase_modelname}}_STANDALONE_INPUTS_DOCSTRING = r"""
Args:
input_ids (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`):
input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`):
Indices of input sequence tokens in the vocabulary. Padding will be ignored by default should you provide
it.
Indices can be obtained using :class:`~transformers.ProphetNetTokenizer`. See
:meth:`transformers.PreTrainedTokenizer.encode` and :meth:`transformers.PreTrainedTokenizer.__call__` for
Indices can be obtained using [`ProphetNetTokenizer`]. See
[`PreTrainedTokenizer.encode`] and [`PreTrainedTokenizer.__call__`] for
details.
`What are input IDs? <../glossary.html#input-ids>`__
attention_mask (:obj:`torch.Tensor` of shape :obj:`(batch_size, sequence_length)`, `optional`):
Mask to avoid performing attention on padding token indices. Mask values selected in ``[0, 1]``:
[What are input IDs?](../glossary#input-ids)
attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
`What are attention masks? <../glossary.html#attention-mask>`__
output_attentions (:obj:`bool`, `optional`):
Whether or not to return the attentions tensors of all attention layers. See ``attentions`` under returned
[What are attention masks?](../glossary#attention-mask)
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under returned
tensors for more detail.
output_hidden_states (:obj:`bool`, `optional`):
Whether or not to return the hidden states of all layers. See ``hidden_states`` under returned tensors for
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for
more detail.
return_dict (:obj:`bool`, `optional`):
Whether or not to return a :class:`~transformers.file_utils.ModelOutput` instead of a plain tuple.
return_dict (`bool`, *optional*):
Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple.
"""
class {{cookiecutter.camelcase_modelname}}Encoder({{cookiecutter.camelcase_modelname}}PreTrainedModel):
"""
Transformer encoder consisting of *config.encoder_layers* self attention layers. Each layer is a
:class:`{{cookiecutter.camelcase_modelname}}EncoderLayer`.
[`{{cookiecutter.camelcase_modelname}}EncoderLayer`].
Args:
config: {{cookiecutter.camelcase_modelname}}Config
......@@ -2247,40 +2245,40 @@ class {{cookiecutter.camelcase_modelname}}Encoder({{cookiecutter.camelcase_model
):
r"""
Args:
input_ids (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`):
input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`):
Indices of input sequence tokens in the vocabulary. Padding will be ignored by default should you
provide it.
Indices can be obtained using :class:`~transformers.{{cookiecutter.camelcase_modelname}}Tokenizer`. See
:meth:`transformers.PreTrainedTokenizer.encode` and :meth:`transformers.PreTrainedTokenizer.__call__`
Indices can be obtained using [`~{{cookiecutter.camelcase_modelname}}Tokenizer`]. See
[`PreTrainedTokenizer.encode`] and [`PreTrainedTokenizer.__call__`]
for details.
`What are input IDs? <../glossary.html#input-ids>`__
attention_mask (:obj:`torch.Tensor` of shape :obj:`(batch_size, sequence_length)`, `optional`):
Mask to avoid performing attention on padding token indices. Mask values selected in ``[0, 1]``:
[What are input IDs?](../glossary#input-ids)
attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
`What are attention masks? <../glossary.html#attention-mask>`__
head_mask (:obj:`torch.Tensor` of shape :obj:`(encoder_layers, encoder_attention_heads)`, `optional`):
Mask to nullify selected heads of the attention modules. Mask values selected in ``[0, 1]``:
[What are attention masks?](../glossary#attention-mask)
head_mask (`torch.Tensor` of shape `(encoder_layers, encoder_attention_heads)`, *optional*):
Mask to nullify selected heads of the attention modules. Mask values selected in `[0, 1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
inputs_embeds (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length, hidden_size)`, `optional`):
Optionally, instead of passing :obj:`input_ids` you can choose to directly pass an embedded
representation. This is useful if you want more control over how to convert :obj:`input_ids` indices
inputs_embeds (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*):
Optionally, instead of passing `input_ids` you can choose to directly pass an embedded
representation. This is useful if you want more control over how to convert `input_ids` indices
into associated vectors than the model's internal embedding lookup matrix.
output_attentions (:obj:`bool`, `optional`):
Whether or not to return the attentions tensors of all attention layers. See ``attentions`` under
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under
returned tensors for more detail.
output_hidden_states (:obj:`bool`, `optional`):
Whether or not to return the hidden states of all layers. See ``hidden_states`` under returned tensors
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors
for more detail.
return_dict (:obj:`bool`, `optional`):
Whether or not to return a :class:`~transformers.file_utils.ModelOutput` instead of a plain tuple.
return_dict (`bool`, *optional*):
Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple.
"""
output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
output_hidden_states = (
......@@ -2369,7 +2367,7 @@ class {{cookiecutter.camelcase_modelname}}Encoder({{cookiecutter.camelcase_model
class {{cookiecutter.camelcase_modelname}}Decoder({{cookiecutter.camelcase_modelname}}PreTrainedModel):
"""
Transformer decoder consisting of *config.decoder_layers* layers. Each layer is a :class:`{{cookiecutter.camelcase_modelname}}DecoderLayer`
Transformer decoder consisting of *config.decoder_layers* layers. Each layer is a [`{{cookiecutter.camelcase_modelname}}DecoderLayer`]
Args:
config: {{cookiecutter.camelcase_modelname}}Config
......@@ -2442,70 +2440,67 @@ class {{cookiecutter.camelcase_modelname}}Decoder({{cookiecutter.camelcase_model
):
r"""
Args:
input_ids (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`):
input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`):
Indices of input sequence tokens in the vocabulary. Padding will be ignored by default should you
provide it.
Indices can be obtained using :class:`~transformers.{{cookiecutter.camelcase_modelname}}Tokenizer`. See
:meth:`transformers.PreTrainedTokenizer.encode` and :meth:`transformers.PreTrainedTokenizer.__call__`
Indices can be obtained using [`~{{cookiecutter.camelcase_modelname}}Tokenizer`]. See
[`PreTrainedTokenizer.encode`] and [`PreTrainedTokenizer.__call__`]
for details.
`What are input IDs? <../glossary.html#input-ids>`__
attention_mask (:obj:`torch.Tensor` of shape :obj:`(batch_size, sequence_length)`, `optional`):
Mask to avoid performing attention on padding token indices. Mask values selected in ``[0, 1]``:
[What are input IDs?](../glossary#input-ids)
attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
`What are attention masks? <../glossary.html#attention-mask>`__
encoder_hidden_states (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, encoder_sequence_length, hidden_size)`, `optional`):
[What are attention masks?](../glossary#attention-mask)
encoder_hidden_states (`torch.FloatTensor` of shape `(batch_size, encoder_sequence_length, hidden_size)`, *optional*):
Sequence of hidden-states at the output of the last layer of the encoder. Used in the cross-attention
of the decoder.
encoder_attention_mask (:obj:`torch.LongTensor` of shape :obj:`(batch_size, encoder_sequence_length)`, `optional`):
encoder_attention_mask (`torch.LongTensor` of shape `(batch_size, encoder_sequence_length)`, *optional*):
Mask to avoid performing cross-attention on padding tokens indices of encoder input_ids. Mask values
selected in ``[0, 1]``:
selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
`What are attention masks? <../glossary.html#attention-mask>`__
head_mask (:obj:`torch.Tensor` of shape :obj:`(decoder_layers, decoder_attention_heads)`, `optional`):
Mask to nullify selected heads of the attention modules. Mask values selected in ``[0, 1]``:
[What are attention masks?](../glossary#attention-mask)
head_mask (`torch.Tensor` of shape `(decoder_layers, decoder_attention_heads)`, *optional*):
Mask to nullify selected heads of the attention modules. Mask values selected in `[0, 1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
cross_attn_head_mask (:obj:`torch.Tensor` of shape :obj:`(decoder_layers, decoder_attention_heads)`, `optional`):
Mask to nullify selected heads of the cross-attention modules. Mask values selected in ``[0, 1]``:
cross_attn_head_mask (`torch.Tensor` of shape `(decoder_layers, decoder_attention_heads)`, *optional*):
Mask to nullify selected heads of the cross-attention modules. Mask values selected in `[0, 1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
past_key_values (:obj:`tuple(tuple(torch.FloatTensor))`, `optional`, returned when ``use_cache=True`` is passed or when ``config.use_cache=True``):
Tuple of :obj:`tuple(torch.FloatTensor)` of length :obj:`config.n_layers`, with each tuple having 2
tensors of shape :obj:`(batch_size, num_heads, sequence_length, embed_size_per_head)`) and 2 additional
tensors of shape :obj:`(batch_size, num_heads, encoder_sequence_length, embed_size_per_head)`.
past_key_values (`tuple(tuple(torch.FloatTensor))`, *optional*, returned when `use_cache=True` is passed or when `config.use_cache=True`):
Tuple of `tuple(torch.FloatTensor)` of length `config.n_layers`, with each tuple having 2
tensors of shape `(batch_size, num_heads, sequence_length, embed_size_per_head)`) and 2 additional
tensors of shape `(batch_size, num_heads, encoder_sequence_length, embed_size_per_head)`.
Contains pre-computed hidden-states (key and values in the self-attention blocks and in the
cross-attention blocks) that can be used (see :obj:`past_key_values` input) to speed up sequential
cross-attention blocks) that can be used (see `past_key_values` input) to speed up sequential
decoding.
If :obj:`past_key_values` are used, the user can optionally input only the last
:obj:`decoder_input_ids` (those that don't have their past key value states given to this model) of
shape :obj:`(batch_size, 1)` instead of all :obj:`decoder_input_ids`` of shape :obj:`(batch_size,
sequence_length)`.
inputs_embeds (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length, hidden_size)`, `optional`):
Optionally, instead of passing :obj:`input_ids` you can choose to directly pass an embedded
representation. This is useful if you want more control over how to convert :obj:`input_ids` indices
If `past_key_values` are used, the user can optionally input only the last
`decoder_input_ids` (those that don't have their past key value states given to this model) of
shape `(batch_size, 1)` instead of all ``decoder_input_ids``` of shape `(batch_size,
sequence_length)`. inputs_embeds (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*): Optionally, instead of passing `input_ids` you can choose to directly pass an embedded representation. This is useful if you want more control over how to convert `input_ids` indices
into associated vectors than the model's internal embedding lookup matrix.
output_attentions (:obj:`bool`, `optional`):
Whether or not to return the attentions tensors of all attention layers. See ``attentions`` under
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under
returned tensors for more detail.
output_hidden_states (:obj:`bool`, `optional`):
Whether or not to return the hidden states of all layers. See ``hidden_states`` under returned tensors
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors
for more detail.
return_dict (:obj:`bool`, `optional`):
Whether or not to return a :class:`~transformers.file_utils.ModelOutput` instead of a plain tuple.
return_dict (`bool`, *optional*):
Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple.
"""
output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
output_hidden_states = (
......@@ -2817,29 +2812,30 @@ class {{cookiecutter.camelcase_modelname}}ForConditionalGeneration({{cookiecutte
return_dict=None,
):
r"""
labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`):
Labels for computing the masked language modeling loss. Indices should either be in ``[0, ...,
config.vocab_size]`` or -100 (see ``input_ids`` docstring). Tokens with indices set to ``-100`` are ignored
(masked), the loss is only computed for the tokens with labels in ``[0, ..., config.vocab_size]``.
labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
Labels for computing the masked language modeling loss. Indices should either be in `[0, ..., config.vocab_size]` or -100 (see `input_ids` docstring). Tokens with indices set to `-100` are ignored
(masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]`.
Returns:
Conditional generation example::
Conditional generation example:
>>> from transformers import {{cookiecutter.camelcase_modelname}}Tokenizer, {{cookiecutter.camelcase_modelname}}ForConditionalGeneration
>>> tokenizer = {{cookiecutter.camelcase_modelname}}Tokenizer.from_pretrained('{{cookiecutter.checkpoint_identifier}}')
>>> TXT = "My friends are <mask> but they eat too many carbs."
```python
>>> from transformers import {{cookiecutter.camelcase_modelname}}Tokenizer, {{cookiecutter.camelcase_modelname}}ForConditionalGeneration
>>> tokenizer = {{cookiecutter.camelcase_modelname}}Tokenizer.from_pretrained('{{cookiecutter.checkpoint_identifier}}')
>>> TXT = "My friends are <mask> but they eat too many carbs."
>>> model = {{cookiecutter.camelcase_modelname}}ForConditionalGeneration.from_pretrained('{{cookiecutter.checkpoint_identifier}}')
>>> input_ids = tokenizer([TXT], return_tensors='pt')['input_ids']
>>> logits = model(input_ids).logits
>>> model = {{cookiecutter.camelcase_modelname}}ForConditionalGeneration.from_pretrained('{{cookiecutter.checkpoint_identifier}}')
>>> input_ids = tokenizer([TXT], return_tensors='pt')['input_ids']
>>> logits = model(input_ids).logits
>>> masked_index = (input_ids[0] == tokenizer.mask_token_id).nonzero().item()
>>> probs = logits[0, masked_index].softmax(dim=0)
>>> values, predictions = probs.topk(5)
>>> masked_index = (input_ids[0] == tokenizer.mask_token_id).nonzero().item()
>>> probs = logits[0, masked_index].softmax(dim=0)
>>> values, predictions = probs.topk(5)
>>> tokenizer.decode(predictions).split()
"""
>>> tokenizer.decode(predictions).split()
```
"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
if labels is not None:
......@@ -2965,9 +2961,8 @@ class {{cookiecutter.camelcase_modelname}}ForSequenceClassification({{cookiecutt
return_dict=None,
):
r"""
labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`):
Labels for computing the sequence classification/regression loss. Indices should be in :obj:`[0, ...,
config.num_labels - 1]`. If :obj:`config.num_labels > 1` a classification loss is computed (Cross-Entropy).
labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
Labels for computing the sequence classification/regression loss. Indices should be in `[0, ..., config.num_labels - 1]`. If `config.num_labels > 1` a classification loss is computed (Cross-Entropy).
"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
if labels is not None:
......@@ -3084,13 +3079,13 @@ class {{cookiecutter.camelcase_modelname}}ForQuestionAnswering({{cookiecutter.ca
return_dict=None,
):
r"""
start_positions (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`):
start_positions (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
Labels for position (index) of the start of the labelled span for computing the token classification loss.
Positions are clamped to the length of the sequence (`sequence_length`). Position outside of the sequence
Positions are clamped to the length of the sequence (*sequence_length*). Position outside of the sequence
are not taken into account for computing the loss.
end_positions (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`):
end_positions (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
Labels for position (index) of the end of the labelled span for computing the token classification loss.
Positions are clamped to the length of the sequence (`sequence_length`). Position outside of the sequence
Positions are clamped to the length of the sequence (*sequence_length*). Position outside of the sequence
are not taken into account for computing the loss.
"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
......@@ -3159,7 +3154,7 @@ class {{cookiecutter.camelcase_modelname}}ForQuestionAnswering({{cookiecutter.ca
class {{cookiecutter.camelcase_modelname}}DecoderWrapper({{cookiecutter.camelcase_modelname}}PreTrainedModel):
"""
This wrapper class is a helper class to correctly load pretrained checkpoints when the causal language model is
used in combination with the :class:`~transformers.EncoderDecoderModel` framework.
used in combination with the [`EncoderDecoderModel`] framework.
"""
def __init__(self, config):
......@@ -3221,81 +3216,81 @@ class {{cookiecutter.camelcase_modelname}}ForCausalLM({{cookiecutter.camelcase_m
):
r"""
Args:
input_ids (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`):
input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`):
Indices of input sequence tokens in the vocabulary. Padding will be ignored by default should you
provide it.
Indices can be obtained using :class:`~transformers.{{cookiecutter.camelcase_modelname}}Tokenizer`. See
:meth:`transformers.PreTrainedTokenizer.encode` and :meth:`transformers.PreTrainedTokenizer.__call__`
Indices can be obtained using [`~{{cookiecutter.camelcase_modelname}}Tokenizer`]. See
[`PreTrainedTokenizer.encode`] and [`PreTrainedTokenizer.__call__`]
for details.
`What are input IDs? <../glossary.html#input-ids>`__
attention_mask (:obj:`torch.Tensor` of shape :obj:`(batch_size, sequence_length)`, `optional`):
Mask to avoid performing attention on padding token indices. Mask values selected in ``[0, 1]``:
[What are input IDs?](../glossary#input-ids)
attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
`What are attention masks? <../glossary.html#attention-mask>`__
encoder_hidden_states (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length, hidden_size)`, `optional`):
[What are attention masks?](../glossary#attention-mask)
encoder_hidden_states (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*):
Sequence of hidden-states at the output of the last layer of the encoder. Used in the cross-attention
if the model is configured as a decoder.
encoder_attention_mask (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`):
encoder_attention_mask (`torch.FloatTensor` of shape `(batch_size, sequence_length)`, *optional*):
Mask to avoid performing attention on the padding token indices of the encoder input. This mask is used
in the cross-attention if the model is configured as a decoder. Mask values selected in ``[0, 1]``:
head_mask (:obj:`torch.Tensor` of shape :obj:`(decoder_layers, decoder_attention_heads)`, `optional`):
Mask to nullify selected heads of the attention modules. Mask values selected in ``[0, 1]``:
in the cross-attention if the model is configured as a decoder. Mask values selected in `[0, 1]`:
head_mask (`torch.Tensor` of shape `(decoder_layers, decoder_attention_heads)`, *optional*):
Mask to nullify selected heads of the attention modules. Mask values selected in `[0, 1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
cross_attn_head_mask (:obj:`torch.Tensor` of shape :obj:`(decoder_layers, decoder_attention_heads)`, `optional`):
Mask to nullify selected heads of the cross-attention modules. Mask values selected in ``[0, 1]``:
cross_attn_head_mask (`torch.Tensor` of shape `(decoder_layers, decoder_attention_heads)`, *optional*):
Mask to nullify selected heads of the cross-attention modules. Mask values selected in `[0, 1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
past_key_values (:obj:`tuple(tuple(torch.FloatTensor))` of length :obj:`config.n_layers` with each tuple having 4 tensors of shape :obj:`(batch_size, num_heads, sequence_length - 1, embed_size_per_head)`):
past_key_values (`tuple(tuple(torch.FloatTensor))` of length `config.n_layers` with each tuple having 4 tensors of shape `(batch_size, num_heads, sequence_length - 1, embed_size_per_head)`):
Contains precomputed key and value hidden-states of the attention blocks. Can be used to speed up
decoding.
If :obj:`past_key_values` are used, the user can optionally input only the last ``decoder_input_ids``
(those that don't have their past key value states given to this model) of shape :obj:`(batch_size, 1)`
instead of all ``decoder_input_ids`` of shape :obj:`(batch_size, sequence_length)`.
labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`):
Labels for computing the masked language modeling loss. Indices should either be in ``[0, ...,
config.vocab_size]`` or -100 (see ``input_ids`` docstring). Tokens with indices set to ``-100`` are
ignored (masked), the loss is only computed for the tokens with labels in ``[0, ...,
config.vocab_size]``.
use_cache (:obj:`bool`, `optional`):
If set to :obj:`True`, :obj:`past_key_values` key value states are returned and can be used to speed up
decoding (see :obj:`past_key_values`).
If `past_key_values` are used, the user can optionally input only the last `decoder_input_ids`
(those that don't have their past key value states given to this model) of shape `(batch_size, 1)`
instead of all `decoder_input_ids` of shape `(batch_size, sequence_length)`.
labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
Labels for computing the masked language modeling loss. Indices should either be in `[0, ..., config.vocab_size]` or -100 (see `input_ids` docstring). Tokens with indices set to `-100` are
ignored (masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]`.
use_cache (`bool`, *optional*):
If set to `True`, `past_key_values` key value states are returned and can be used to speed up
decoding (see `past_key_values`).
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
output_attentions (:obj:`bool`, `optional`):
Whether or not to return the attentions tensors of all attention layers. See ``attentions`` under
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under
returned tensors for more detail.
output_hidden_states (:obj:`bool`, `optional`):
Whether or not to return the hidden states of all layers. See ``hidden_states`` under returned tensors
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors
for more detail.
return_dict (:obj:`bool`, `optional`):
Whether or not to return a :class:`~transformers.file_utils.ModelOutput` instead of a plain tuple.
return_dict (`bool`, *optional*):
Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple.
Returns:
Example::
Example:
>>> from transformers import {{cookiecutter.camelcase_modelname}}Tokenizer, {{cookiecutter.camelcase_modelname}}ForCausalLM
```python
>>> from transformers import {{cookiecutter.camelcase_modelname}}Tokenizer, {{cookiecutter.camelcase_modelname}}ForCausalLM
>>> tokenizer = {{cookiecutter.camelcase_modelname}}Tokenizer.from_pretrained('facebook/bart-large')
>>> model = {{cookiecutter.camelcase_modelname}}ForCausalLM.from_pretrained('facebook/bart-large', add_cross_attention=False)
>>> assert model.config.is_decoder, f"{model.__class__} has to be configured as a decoder."
>>> inputs = tokenizer("Hello, my dog is cute", return_tensors="pt")
>>> outputs = model(**inputs)
>>> tokenizer = {{cookiecutter.camelcase_modelname}}Tokenizer.from_pretrained('facebook/bart-large')
>>> model = {{cookiecutter.camelcase_modelname}}ForCausalLM.from_pretrained('facebook/bart-large', add_cross_attention=False)
>>> assert model.config.is_decoder, f"{model.__class__} has to be configured as a decoder."
>>> inputs = tokenizer("Hello, my dog is cute", return_tensors="pt")
>>> outputs = model(**inputs)
>>> logits = outputs.logits
"""
>>> logits = outputs.logits
```
"""
output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
output_hidden_states = (
......
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