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

PyTorch XLNet

Browse Source
This commit is contained in:
Lysandre 2020-01-17 14:59:31 -05:00 committed by Lysandre Debut
parent 83fa8d9fb5
commit cd656fb21a
2 changed files with 355 additions and 388 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
docs/source/model_doc/xlnet.rst
View File

@ -1,6 +1,22 @@
XLNet
----------------------------------------------------
The XLNet model was proposed in `XLNet: Generalized Autoregressive Pretraining for Language Understanding`_
by Zhilin Yang*, Zihang Dai*, Yiming Yang, Jaime Carbonell, Ruslan Salakhutdinov, Quoc V. Le.
XLnet is an extension of the Transformer-XL model pre-trained using an autoregressive method
to learn bidirectional contexts by maximizing the expected likelihood over all permutations
of the input sequence factorization order.
The specific attention pattern can be controlled at training and test time using the `perm_mask` input.
Due to the difficulty of training a fully auto-regressive model over various factorization order,
XLNet is pretrained using only a sub-set of the output tokens as target which are selected
with the `target_mapping` input.
To use XLNet for sequential decoding (i.e. not in fully bi-directional setting), use the `perm_mask` and
`target_mapping` inputs to control the attention span and outputs (see examples in `examples/run_generation.py`)
``XLNetConfig``
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

727
src/transformers/modeling_xlnet.py
View File

@ -26,7 +26,7 @@ from torch.nn import CrossEntropyLoss, MSELoss
from torch.nn import functional as F
from .configuration_xlnet import XLNetConfig
from .file_utils import add_start_docstrings
from .file_utils import add_start_docstrings, add_start_docstrings_to_callable
from .modeling_utils import PoolerAnswerClass, PoolerEndLogits, PoolerStartLogits, PreTrainedModel, SequenceSummary
@ -505,30 +505,10 @@ class XLNetPreTrainedModel(PreTrainedModel):
module.mask_emb.data.normal_(mean=0.0, std=self.config.initializer_range)
XLNET_START_DOCSTRING = r""" The XLNet model was proposed in
`XLNet: Generalized Autoregressive Pretraining for Language Understanding`_
by Zhilin Yang*, Zihang Dai*, Yiming Yang, Jaime Carbonell, Ruslan Salakhutdinov, Quoc V. Le.
XLnet is an extension of the Transformer-XL model pre-trained using an autoregressive method
to learn bidirectional contexts by maximizing the expected likelihood over all permutations
of the input sequence factorization order.
The specific attention pattern can be controlled at training and test time using the `perm_mask` input.
Due to the difficulty of training a fully auto-regressive model over various factorization order,
XLNet is pretrained using only a sub-set of the output tokens as target which are selected
with the `target_mapping` input.
To use XLNet for sequential decoding (i.e. not in fully bi-directional setting), use the `perm_mask` and
`target_mapping` inputs to control the attention span and outputs (see examples in `examples/run_generation.py`)
This model is a PyTorch `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.
.. _`XLNet: Generalized Autoregressive Pretraining for Language Understanding`:
http://arxiv.org/abs/1906.08237
.. _`torch.nn.Module`:
https://pytorch.org/docs/stable/nn.html#module
XLNET_START_DOCSTRING = r"""
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.XLNetConfig`): Model configuration class with all the parameters of the model.
@ -537,60 +517,54 @@ XLNET_START_DOCSTRING = r""" The XLNet model was proposed in
"""
XLNET_INPUTS_DOCSTRING = r"""
Inputs:
**input_ids**: ``torch.LongTensor`` of shape ``(batch_size, sequence_length)``:
Indices of input sequence tokens in the vocabulary.
XLNet is a model with relative position embeddings so you can either pad the inputs on
the right or on the left.
Indices can be obtained using :class:`transformers.XLNetTokenizer`.
Args:
input_ids (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`):
Indices of input sequence tokens in the vocabulary.
Indices can be obtained using :class:`transformers.BertTokenizer`.
See :func:`transformers.PreTrainedTokenizer.encode` and
:func:`transformers.PreTrainedTokenizer.convert_tokens_to_ids` for details.
**token_type_ids**: (`optional`) ``torch.LongTensor`` of shape ``(batch_size, sequence_length)``:
A parallel sequence of tokens (can be used to indicate various portions of the inputs).
The type indices in XLNet are NOT selected in the vocabulary, they can be arbitrary numbers and
the important thing is that they should be different for tokens which belong to different segments.
The model will compute relative segment differences from the given type indices:
0 if the segment id of two tokens are the same, 1 if not.
**attention_mask**: (`optional`) ``torch.FloatTensor`` of shape ``(batch_size, sequence_length)``:
:func:`transformers.PreTrainedTokenizer.encode_plus` for details.
`What are input IDs? <../glossary.html#input-ids>`__
attention_mask (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`, defaults to :obj:`None`):
Mask to avoid performing attention on padding token indices.
Mask values selected in ``[0, 1]``:
``1`` for tokens that are NOT MASKED, ``0`` for MASKED tokens.
**mems**: (`optional`)
list of ``torch.FloatTensor`` (one for each layer):
that contains pre-computed hidden-states (key and values in the attention blocks) as output by the model
(see `mems` output below). Can be used to speed up sequential decoding and attend to longer context.
To activate mems you need to set up config.mem_len to a positive value which will be the max number of tokens in
the memory output by the model. E.g. `model = XLNetModel.from_pretrained('xlnet-base-case, mem_len=1024)` will
instantiate a model which can use up to 1024 tokens of memory (in addition to the input it self).
**perm_mask**: (`optional`) ``torch.FloatTensor`` of shape ``(batch_size, sequence_length, sequence_length)``:
`What are attention masks? <../glossary.html#attention-mask>`__
mems (:obj:`List[torch.FloatTensor]` of length :obj:`config.n_layers`):
Contains pre-computed hidden-states (key and values in the attention blocks) as computed by the model
(see `mems` output below). Can be used to speed up sequential decoding. The token ids which have their mems
given to this model should not be passed as input ids as they have already been computed.
perm_mask (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length, sequence_length)`, `optional`, defaults to :obj:`None`):
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 None, 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**: (`optional`) ``torch.FloatTensor`` of shape ``(batch_size, num_predict, sequence_length)``:
target_mapping (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, num_predict, sequence_length)`, `optional`, defaults to :obj:`None`):
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**: (`optional`) ``torch.LongTensor`` of shape ``(batch_size, sequence_length)``:
A parallel sequence of tokens (can be used to indicate various portions of the inputs).
The type indices in XLNet are NOT selected in the vocabulary, they can be arbitrary numbers and
the important thing is that they should be different for tokens which belong to different segments.
The model will compute relative segment differences from the given type indices:
0 if the segment id of two tokens are the same, 1 if not.
**input_mask**: (`optional`) ``torch.FloatTensor`` of shape ``(batch_size, sequence_length)``:
token_type_ids (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`, defaults to :obj:`None`):
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
`What are token type IDs? <../glossary.html#token-type-ids>`_
input_mask (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`, defaults to :obj:`None`):
Mask to avoid performing attention on padding token indices.
Negative of `attention_mask`, i.e. with 0 for real tokens and 1 for padding.
Kept for compatibility with the original code base.
You can only uses one of `input_mask` and `attention_mask`
Mask values selected in ``[0, 1]``:
``1`` for tokens that are MASKED, ``0`` for tokens that are NOT MASKED.
**head_mask**: (`optional`) ``torch.FloatTensor`` of shape ``(num_heads,)`` or ``(num_layers, num_heads)``:
head_mask (:obj:`torch.FloatTensor` of shape :obj:`(num_heads,)` or :obj:`(num_layers, num_heads)`, `optional`, defaults to :obj:`None`):
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**: (`optional`) ``torch.FloatTensor`` of shape ``(batch_size, sequence_length, embedding_dim)``:
Optionally, instead of passing ``input_ids`` you can choose to directly pass an embedded representation.
:obj:`1` indicates the head is **not masked**, :obj:`0` indicates the head is **masked**.
input_embeds (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length, hidden_size)`, `optional`, defaults to :obj:`None`):
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
than the model's internal embedding lookup matrix.
"""
@ -599,36 +573,8 @@ XLNET_INPUTS_DOCSTRING = r"""
@add_start_docstrings(
"The bare XLNet Model transformer outputting raw hidden-states without any specific head on top.",
XLNET_START_DOCSTRING,
XLNET_INPUTS_DOCSTRING,
)
class XLNetModel(XLNetPreTrainedModel):
r"""
Outputs: `Tuple` comprising various elements depending on the configuration (config) and inputs:
**last_hidden_state**: ``torch.FloatTensor`` of shape ``(batch_size, sequence_length, hidden_size)``
Sequence of hidden-states at the last layer of the model.
**mems**: (`optional`, returned when ``config.mem_len > 0``)
list of ``torch.FloatTensor`` (one for each layer):
that contains pre-computed hidden-states (key and values in the attention blocks) as computed by the model
if config.mem_len > 0 else tuple of None. Can be used to speed up sequential decoding and attend to longer context.
See details in the docstring of the `mems` input above.
**hidden_states**: (`optional`, returned when ``config.output_hidden_states=True``)
list of ``torch.FloatTensor`` (one for the output of each layer + the output of the embeddings)
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**: (`optional`, returned when ``config.output_attentions=True``)
list 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.
When ``target_mapping is not None``, the attentions outputs are a list of 2-tuple of ``torch.FloatTensor``.
Examples::
tokenizer = XLNetTokenizer.from_pretrained('xlnet-large-cased')
model = XLNetModel.from_pretrained('xlnet-large-cased')
input_ids = torch.tensor(tokenizer.encode("Hello, my dog is cute", add_special_tokens=True)).unsqueeze(0) # Batch size 1
outputs = model(input_ids)
last_hidden_states = outputs[0] # The last hidden-state is the first element of the output tuple
"""
def __init__(self, config):
super().__init__(config)
@ -666,8 +612,8 @@ class XLNetModel(XLNetPreTrainedModel):
Creates causal attention mask. Float mask where 1.0 indicates masked, 0.0 indicates not-masked.
Args:
qlen: TODO Lysandre didn't fill
mlen: TODO Lysandre didn't fill
qlen: Sequence length
mlen: Mask length
::
@ -692,7 +638,7 @@ class XLNetModel(XLNetPreTrainedModel):
return ret
def cache_mem(self, curr_out, prev_mem):
"""cache hidden states into memory."""
# cache hidden states into memory.
if self.reuse_len is not None and self.reuse_len > 0:
curr_out = curr_out[: self.reuse_len]
@ -715,7 +661,7 @@ class XLNetModel(XLNetPreTrainedModel):
return pos_emb
def relative_positional_encoding(self, qlen, klen, bsz=None):
"""create relative positional encoding."""
# create relative positional encoding.
freq_seq = torch.arange(0, self.d_model, 2.0, dtype=torch.float)
inv_freq = 1 / torch.pow(10000, (freq_seq / self.d_model))
@ -753,6 +699,7 @@ class XLNetModel(XLNetPreTrainedModel):
pos_emb = pos_emb.to(next(self.parameters()))
return pos_emb
@add_start_docstrings_to_callable(XLNET_INPUTS_DOCSTRING)
def forward(
self,
input_ids=None,
@ -765,6 +712,36 @@ class XLNetModel(XLNetPreTrainedModel):
head_mask=None,
inputs_embeds=None,
):
r"""
Return:
:obj:`tuple(torch.FloatTensor)` comprising various elements depending on the configuration (config) and inputs:
last_hidden_state (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length, hidden_size)`):
Sequence of hidden-states at the last layer of the model.
mems (:obj:`List[torch.FloatTensor]` of length :obj:`config.n_layers`):
Contains pre-computed hidden-states (key and values in the attention blocks).
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 ``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 of the model at the output of each layer plus the initial embedding outputs.
attentions (:obj:`tuple(torch.FloatTensor)`, `optional`, returned 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 weights after the attention softmax, used to compute the weighted average in the self-attention
heads.
Examples::
tokenizer = XLNetTokenizer.from_pretrained('xlnet-large-cased')
model = XLNetModel.from_pretrained('xlnet-large-cased')
input_ids = torch.tensor(tokenizer.encode("Hello, my dog is cute", add_special_tokens=True)).unsqueeze(0) # Batch size 1
outputs = model(input_ids)
last_hidden_states = outputs[0] # The last hidden-state is the first element of the output tuple
"""
# the original code for XLNet uses shapes [len, bsz] with the batch dimension at the end
# but we want a unified interface in the library with the batch size on the first dimension
# so we move here the first dimension (batch) to the end
@ -950,50 +927,8 @@ class XLNetModel(XLNetPreTrainedModel):
"""XLNet Model with a language modeling head on top
(linear layer with weights tied to the input embeddings). """,
XLNET_START_DOCSTRING,
XLNET_INPUTS_DOCSTRING,
)
class XLNetLMHeadModel(XLNetPreTrainedModel):
r"""
**labels**: (`optional`) ``torch.LongTensor`` of shape ``(batch_size, sequence_length)``:
Labels for language modeling.
Note that the labels **are shifted** inside the model, i.e. you can set ``lm_labels = input_ids``
Indices are selected in ``[-1, 0, ..., config.vocab_size]``
All labels set to ``-100`` are ignored (masked), the loss is only
computed for labels in ``[0, ..., config.vocab_size]``
Outputs: `Tuple` comprising various elements depending on the configuration (config) and inputs:
**loss**: (`optional`, returned when ``labels`` is provided) ``torch.FloatTensor`` of shape ``(1,)``:
Language modeling loss.
**prediction_scores**: ``torch.FloatTensor`` of shape ``(batch_size, sequence_length, config.vocab_size)``
Prediction scores of the language modeling head (scores for each vocabulary token before SoftMax).
**mems**: (`optional`, returned when ``config.mem_len > 0``)
list of ``torch.FloatTensor`` (one for each layer):
that contains pre-computed hidden-states (key and values in the attention blocks) as computed by the model
if config.mem_len > 0 else tuple of None. Can be used to speed up sequential decoding and attend to longer context.
See details in the docstring of the `mems` input above.
**hidden_states**: (`optional`, returned when ``config.output_hidden_states=True``)
list of ``torch.FloatTensor`` (one for the output of each layer + the output of the embeddings)
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**: (`optional`, returned when ``config.output_attentions=True``)
list 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.
When ``target_mapping is not None``, the attentions outputs are a list of 2-tuple of ``torch.FloatTensor``.
Examples::
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=True)).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]
"""
def __init__(self, config):
super().__init__(config)
@ -1036,6 +971,7 @@ class XLNetLMHeadModel(XLNetPreTrainedModel):
return inputs
@add_start_docstrings_to_callable(XLNET_INPUTS_DOCSTRING)
def forward(
self,
input_ids=None,
@ -1049,6 +985,50 @@ class XLNetLMHeadModel(XLNetPreTrainedModel):
inputs_embeds=None,
labels=None,
):
r"""
labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`, defaults to :obj:`None`):
Labels for language modeling.
Note that the labels **are shifted** inside the model, i.e. you can set ``lm_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:
:obj:`tuple(torch.FloatTensor)` comprising various elements depending on the configuration (:obj:`~transformers.GPT2Config`) and inputs:
loss (:obj:`torch.FloatTensor` of shape `(1,)`, `optional`, returned when ``labels`` is provided)
Language modeling loss.
prediction_scores (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length, config.vocab_size)`):
Prediction scores of the language modeling head (scores for each vocabulary token before SoftMax).
mems (:obj:`List[torch.FloatTensor]` of length :obj:`config.n_layers`):
Contains pre-computed hidden-states (key and values in the attention blocks).
Can be used (see `past` 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 ``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 of the model at the output of each layer plus the initial embedding outputs.
attentions (:obj:`tuple(torch.FloatTensor)`, `optional`, returned 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 weights after the attention softmax, used to compute the weighted average in the self-attention
heads.
Examples::
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=True)).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]
"""
transformer_outputs = self.transformer(
input_ids,
attention_mask=attention_mask,
@ -1078,45 +1058,8 @@ class XLNetLMHeadModel(XLNetPreTrainedModel):
"""XLNet Model with a sequence classification/regression head on top (a linear layer on top of
the pooled output) e.g. for GLUE tasks. """,
XLNET_START_DOCSTRING,
XLNET_INPUTS_DOCSTRING,
)
class XLNetForSequenceClassification(XLNetPreTrainedModel):
r"""
**labels**: (`optional`) ``torch.LongTensor`` of shape ``(batch_size,)``:
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).
Outputs: `Tuple` comprising various elements depending on the configuration (config) and inputs:
**loss**: (`optional`, returned when ``labels`` is provided) ``torch.FloatTensor`` of shape ``(1,)``:
Classification (or regression if config.num_labels==1) loss.
**logits**: ``torch.FloatTensor`` of shape ``(batch_size, config.num_labels)``
Classification (or regression if config.num_labels==1) scores (before SoftMax).
**mems**: (`optional`, returned when ``config.mem_len > 0``)
list of ``torch.FloatTensor`` (one for each layer):
that contains pre-computed hidden-states (key and values in the attention blocks) as computed by the model
if config.mem_len > 0 else tuple of None. Can be used to speed up sequential decoding and attend to longer context.
See details in the docstring of the `mems` input above.
**hidden_states**: (`optional`, returned when ``config.output_hidden_states=True``)
list of ``torch.FloatTensor`` (one for the output of each layer + the output of the embeddings)
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**: (`optional`, returned when ``config.output_attentions=True``)
list 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.
When ``target_mapping is not None``, the attentions outputs are a list of 2-tuple of ``torch.FloatTensor``.
Examples::
tokenizer = XLNetTokenizer.from_pretrained('xlnet-large-cased')
model = XLNetForSequenceClassification.from_pretrained('xlnet-large-cased')
input_ids = torch.tensor(tokenizer.encode("Hello, my dog is cute", add_special_tokens=True)).unsqueeze(0) # Batch size 1
labels = torch.tensor([1]).unsqueeze(0) # Batch size 1
outputs = model(input_ids, labels=labels)
loss, logits = outputs[:2]
"""
def __init__(self, config):
super().__init__(config)
@ -1128,6 +1071,7 @@ class XLNetForSequenceClassification(XLNetPreTrainedModel):
self.init_weights()
@add_start_docstrings_to_callable(XLNET_INPUTS_DOCSTRING)
def forward(
self,
input_ids=None,
@ -1141,6 +1085,44 @@ class XLNetForSequenceClassification(XLNetPreTrainedModel):
inputs_embeds=None,
labels=None,
):
r"""
labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`, defaults to :obj:`None`)
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).
Outputs: `Tuple` comprising various elements depending on the configuration (config) and inputs:
loss (:obj:`torch.FloatTensor` of shape :obj:`(1,)`, `optional`, returned when :obj:`labels` is provided):
Classification (or regression if config.num_labels==1) loss.
logits (:obj:`torch.FloatTensor` of shape :obj:(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 (key and values in the attention blocks).
Can be used (see `past` 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 ``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 of the model at the output of each layer plus the initial embedding outputs.
attentions (:obj:`tuple(torch.FloatTensor)`, `optional`, returned 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 weights after the attention softmax, used to compute the weighted average in the self-attention
heads.
Examples::
tokenizer = XLNetTokenizer.from_pretrained('xlnet-large-cased')
model = XLNetForSequenceClassification.from_pretrained('xlnet-large-cased')
input_ids = torch.tensor(tokenizer.encode("Hello, my dog is cute", add_special_tokens=True)).unsqueeze(0) # Batch size 1
labels = torch.tensor([1]).unsqueeze(0) # Batch size 1
outputs = model(input_ids, labels=labels)
loss, logits = outputs[:2]
"""
transformer_outputs = self.transformer(
input_ids,
attention_mask=attention_mask,
@ -1174,64 +1156,10 @@ class XLNetForSequenceClassification(XLNetPreTrainedModel):
@add_start_docstrings(
"""XLNet Model with a token classification head on top (a linear layer on top of
the hidden-states output) e.g. for Named-Entity-Recognition (NER) tasks. """,
the hidden-states output) e.g. for Named-Entity-Recognition (NER) tasks. """,
XLNET_START_DOCSTRING,
XLNET_INPUTS_DOCSTRING,
)
class XLNetForTokenClassification(XLNetPreTrainedModel):
r"""
Inputs:
**input_ids**: ``torch.LongTensor`` of shape ``(batch_size, num_choices, sequence_length)``:
Indices of input sequence tokens in the vocabulary.
The second dimension of the input (`num_choices`) indicates the number of choices to scores.
**token_type_ids**: (`optional`) ``torch.LongTensor`` of shape ``(batch_size, sequence_length)``:
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``
**attention_mask**: (`optional`) ``torch.FloatTensor`` of shape ``(batch_size, sequence_length)``:
Mask to avoid performing attention on padding token indices.
Mask values selected in ``[0, 1]``:
``1`` for tokens that are NOT MASKED, ``0`` for MASKED tokens.
**head_mask**: (`optional`) ``torch.FloatTensor`` of shape ``(num_heads,)`` or ``(num_layers, num_heads)``:
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**: (`optional`) ``torch.FloatTensor`` of shape ``(batch_size, sequence_length, embedding_dim)``:
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.
**labels**: (`optional`) ``torch.LongTensor`` of shape ``(batch_size,)``:
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)
Outputs: `Tuple` comprising various elements depending on the configuration (config) and inputs:
**loss**: (`optional`, returned when ``labels`` is provided) ``torch.FloatTensor`` of shape ``(1,)``:
Classification loss.
**scores**: ``torch.FloatTensor`` of shape ``(batch_size, sequence_length, config.num_labels)``
Classification scores (before SoftMax).
**mems**: (`optional`, returned when ``config.mem_len > 0``)
list of ``torch.FloatTensor`` (one for each layer):
that contains pre-computed hidden-states (key and values in the attention blocks) as computed by the model
if config.mem_len > 0 else tuple of None. Can be used to speed up sequential decoding and attend to longer context.
See details in the docstring of the `mems` input above.
**hidden_states**: (`optional`, returned when ``config.output_hidden_states=True``)
list of ``torch.FloatTensor`` (one for the output of each layer + the output of the embeddings)
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**: (`optional`, returned when ``config.output_attentions=True``)
list 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.
Examples::
tokenizer = XLNetTokenizer.from_pretrained('xlnet-large-cased')
model = XLNetForSequenceClassification.from_pretrained('xlnet-large-cased')
input_ids = torch.tensor(tokenizer.encode("Hello, my dog is cute")).unsqueeze(0) # Batch size 1
labels = torch.tensor([1] * input_ids.size(1)).unsqueeze(0) # Batch size 1
outputs = model(input_ids, labels=labels)
scores = outputs[0]
"""
def __init__(self, config):
super().__init__(config)
@ -1242,6 +1170,7 @@ class XLNetForTokenClassification(XLNetPreTrainedModel):
self.init_weights()
@add_start_docstrings_to_callable(XLNET_INPUTS_DOCSTRING)
def forward(
self,
input_ids=None,
@ -1255,6 +1184,44 @@ class XLNetForTokenClassification(XLNetPreTrainedModel):
inputs_embeds=None,
labels=None,
):
r"""
labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`, defaults to :obj:`None`):
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:
`tuple(torch.FloatTensor)` comprising various elements depending on the configuration (config) and inputs:
loss (:obj:`torch.FloatTensor` of shape :obj:`(1,)`, `optional`, returned when :obj:`labels` is provided):
Classification loss.
logits (:obj:`torch.FloatTensor` of shape :obj:(batch_size, config.num_labels)`):
Classification scores (before SoftMax).
mems (:obj:`List[torch.FloatTensor]` of length :obj:`config.n_layers`):
Contains pre-computed hidden-states (key and values in the attention blocks).
Can be used (see `past` 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 ``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 of the model at the output of each layer plus the initial embedding outputs.
attentions (:obj:`tuple(torch.FloatTensor)`, `optional`, returned 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 weights after the attention softmax, used to compute the weighted average in the self-attention
heads.
Examples::
tokenizer = XLNetTokenizer.from_pretrained('xlnet-large-cased')
model = XLNetForSequenceClassification.from_pretrained('xlnet-large-cased')
input_ids = torch.tensor(tokenizer.encode("Hello, my dog is cute")).unsqueeze(0) # Batch size 1
labels = torch.tensor([1] * input_ids.size(1)).unsqueeze(0) # Batch size 1
outputs = model(input_ids, labels=labels)
scores = outputs[0]
"""
outputs = self.transformer(
input_ids,
@ -1292,67 +1259,8 @@ class XLNetForTokenClassification(XLNetPreTrainedModel):
"""XLNet Model with a multiple choice classification head on top (a linear layer on top of
the pooled output and a softmax) e.g. for RACE/SWAG tasks. """,
XLNET_START_DOCSTRING,
XLNET_INPUTS_DOCSTRING,
)
class XLNetForMultipleChoice(XLNetPreTrainedModel):
r"""
Inputs:
**input_ids**: ``torch.LongTensor`` of shape ``(batch_size, num_choices, sequence_length)``:
Indices of input sequence tokens in the vocabulary.
The second dimension of the input (`num_choices`) indicates the number of choices to scores.
**token_type_ids**: (`optional`) ``torch.LongTensor`` of shape ``(batch_size, num_choices, sequence_length)``:
Segment token indices to indicate first and second portions of the inputs.
The second dimension of the input (`num_choices`) indicates the number of choices to score.
Indices are selected in ``[0, 1]``: ``0`` corresponds to a `sentence A` token, ``1``
**attention_mask**: (`optional`) ``torch.FloatTensor`` of shape ``(batch_size, num_choices, sequence_length)``:
Mask to avoid performing attention on padding token indices.
The second dimension of the input (`num_choices`) indicates the number of choices to score.
Mask values selected in ``[0, 1]``:
``1`` for tokens that are NOT MASKED, ``0`` for MASKED tokens.
**head_mask**: (`optional`) ``torch.FloatTensor`` of shape ``(num_heads,)`` or ``(num_layers, num_heads)``:
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**: (`optional`) ``torch.FloatTensor`` of shape ``(batch_size, sequence_length, embedding_dim)``:
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.
**labels**: (`optional`) ``torch.LongTensor`` of shape ``(batch_size,)``:
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)
Outputs: `Tuple` comprising various elements depending on the configuration (config) and inputs:
**loss**: (`optional`, returned when ``labels`` is provided) ``torch.FloatTensor`` of shape ``(1,)``:
Classification loss.
**classification_scores**: ``torch.FloatTensor`` of shape ``(batch_size, num_choices)`` where `num_choices` is the size of the second dimension
of the input tensors. (see `input_ids` above).
Classification scores (before SoftMax).
**mems**: (`optional`, returned when ``config.mem_len > 0``)
list of ``torch.FloatTensor`` (one for each layer):
that contains pre-computed hidden-states (key and values in the attention blocks) as computed by the model
if config.mem_len > 0 else tuple of None. Can be used to speed up sequential decoding and attend to longer context.
See details in the docstring of the `mems` input above.
**hidden_states**: (`optional`, returned when ``config.output_hidden_states=True``)
list of ``torch.FloatTensor`` (one for the output of each layer + the output of the embeddings)
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**: (`optional`, returned when ``config.output_attentions=True``)
list 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.
When ``target_mapping is not None``, the attentions outputs are a list of 2-tuple of ``torch.FloatTensor``.
Examples::
tokenizer = XLNetTokenizer.from_pretrained('xlnet-base-cased')
model = XLNetForMultipleChoice.from_pretrained('xlnet-base-cased')
choices = ["Hello, my dog is cute", "Hello, my cat is amazing"]
input_ids = torch.tensor([tokenizer.encode(s) for s in choices]).unsqueeze(0) # Batch size 1, 2 choices
labels = torch.tensor(1).unsqueeze(0) # Batch size 1
outputs = model(input_ids, labels=labels)
loss, classification_scores = outputs[:2]
"""
def __init__(self, config):
super().__init__(config)
@ -1363,6 +1271,7 @@ class XLNetForMultipleChoice(XLNetPreTrainedModel):
self.init_weights()
@add_start_docstrings_to_callable(XLNET_INPUTS_DOCSTRING)
def forward(
self,
input_ids=None,
@ -1376,6 +1285,47 @@ class XLNetForMultipleChoice(XLNetPreTrainedModel):
head_mask=None,
inputs_embeds=None,
):
r"""
labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`, defaults to :obj:`None`):
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)
Returns:
:obj:`tuple(torch.FloatTensor)` comprising various elements depending on the configuration (config) and inputs:
loss (:obj:`torch.FloatTensor`` of shape ``(1,)`, `optional`, returned when :obj:`labels` is provided):
Classification loss.
classification_scores (:obj:`torch.FloatTensor` of shape :obj:`(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 (key and values in the attention blocks).
Can be used (see `past` 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 ``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 of the model at the output of each layer plus the initial embedding outputs.
attentions (:obj:`tuple(torch.FloatTensor)`, `optional`, returned 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 weights after the attention softmax, used to compute the weighted average in the self-attention
heads.
Examples::
tokenizer = XLNetTokenizer.from_pretrained('xlnet-base-cased')
model = XLNetForMultipleChoice.from_pretrained('xlnet-base-cased')
choices = ["Hello, my dog is cute", "Hello, my cat is amazing"]
input_ids = torch.tensor([tokenizer.encode(s) for s in choices]).unsqueeze(0) # Batch size 1, 2 choices
labels = torch.tensor(1).unsqueeze(0) # Batch size 1
outputs = model(input_ids, labels=labels)
loss, classification_scores = outputs[:2]
"""
num_choices = input_ids.shape[1]
flat_input_ids = input_ids.view(-1, input_ids.size(-1))
@ -1416,51 +1366,8 @@ class XLNetForMultipleChoice(XLNetPreTrainedModel):
"""XLNet Model with a span classification head on top for extractive question-answering tasks like SQuAD (a linear layers on top of
the hidden-states output to compute `span start logits` and `span end logits`). """,
XLNET_START_DOCSTRING,
XLNET_INPUTS_DOCSTRING,
)
class XLNetForQuestionAnsweringSimple(XLNetPreTrainedModel):
r"""
**start_positions**: (`optional`) ``torch.LongTensor`` of shape ``(batch_size,)``:
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 are not taken into account for computing the loss.
**end_positions**: (`optional`) ``torch.LongTensor`` of shape ``(batch_size,)``:
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 are not taken into account for computing the loss.
Outputs: `Tuple` comprising various elements depending on the configuration (config) and inputs:
**loss**: (`optional`, returned if both ``start_positions`` and ``end_positions`` are provided) ``torch.FloatTensor`` of shape ``(1,)``:
Classification loss as the sum of start token, end token (and is_impossible if provided) classification losses.
**start_scores**: ``torch.FloatTensor`` of shape ``(batch_size, sequence_length,)``
Span-start scores (before SoftMax).
**end_scores**: ``torch.FloatTensor`` of shape ``(batch_size, sequence_length,)``
Span-end scores (before SoftMax).
**mems**: (`optional`, returned when ``config.mem_len > 0``)
list of ``torch.FloatTensor`` (one for each layer):
that contains pre-computed hidden-states (key and values in the attention blocks) as computed by the model
if config.mem_len > 0 else tuple of None. Can be used to speed up sequential decoding and attend to longer context.
See details in the docstring of the `mems` input above.
**hidden_states**: (`optional`, returned when ``config.output_hidden_states=True``)
list of ``torch.FloatTensor`` (one for the output of each layer + the output of the embeddings)
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**: (`optional`, returned when ``config.output_attentions=True``)
list 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.
When ``target_mapping is not None``, the attentions outputs are a list of 2-tuple of ``torch.FloatTensor``.
Examples::
tokenizer = XLMTokenizer.from_pretrained('xlm-mlm-en-2048')
model = XLMForQuestionAnswering.from_pretrained('xlnet-large-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)
loss, start_scores, end_scores = outputs[:2]
"""
def __init__(self, config):
super().__init__(config)
@ -1471,6 +1378,7 @@ class XLNetForQuestionAnsweringSimple(XLNetPreTrainedModel):
self.init_weights()
@add_start_docstrings_to_callable(XLNET_INPUTS_DOCSTRING)
def forward(
self,
input_ids=None,
@ -1485,6 +1393,51 @@ class XLNetForQuestionAnsweringSimple(XLNetPreTrainedModel):
start_positions=None,
end_positions=None,
):
r"""
start_positions (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`, defaults to :obj:`None`):
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 are not taken into account for computing the loss.
end_positions (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`, defaults to :obj:`None`):
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 are not taken into account for computing the loss.
Returns:
:obj:`tuple(torch.FloatTensor)` comprising various elements depending on the configuration (config) and inputs:
loss (:obj:`torch.FloatTensor` of shape :obj:`(1,)`, `optional`, returned when :obj:`labels` is provided):
Total span extraction loss is the sum of a Cross-Entropy for the start and end positions.
start_scores (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length,)`):
Span-start scores (before SoftMax).
end_scores (:obj:`torch.FloatTensor` of shape :obj:`(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 (key and values in the attention blocks).
Can be used (see `past` 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 ``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 of the model at the output of each layer plus the initial embedding outputs.
attentions (:obj:`tuple(torch.FloatTensor)`, `optional`, returned 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 weights after the attention softmax, used to compute the weighted average in the self-attention
heads.
Examples::
tokenizer = XLMTokenizer.from_pretrained('xlm-mlm-en-2048')
model = XLMForQuestionAnswering.from_pretrained('xlnet-large-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)
loss, start_scores, end_scores = outputs[:2]
"""
outputs = self.transformer(
input_ids,
@ -1530,69 +1483,8 @@ class XLNetForQuestionAnsweringSimple(XLNetPreTrainedModel):
"""XLNet Model with a span classification head on top for extractive question-answering tasks like SQuAD (a linear layers on top of
the hidden-states output to compute `span start logits` and `span end logits`). """,
XLNET_START_DOCSTRING,
XLNET_INPUTS_DOCSTRING,
)
class XLNetForQuestionAnswering(XLNetPreTrainedModel):
r"""
**start_positions**: (`optional`) ``torch.LongTensor`` of shape ``(batch_size,)``:
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 are not taken into account for computing the loss.
**end_positions**: (`optional`) ``torch.LongTensor`` of shape ``(batch_size,)``:
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 are not taken into account for computing the loss.
**is_impossible**: (`optional`) ``torch.LongTensor`` of shape ``(batch_size,)``:
Labels whether a question has an answer or no answer (SQuAD 2.0)
**cls_index**: (`optional`) ``torch.LongTensor`` of shape ``(batch_size,)``:
Labels for position (index) of the classification token to use as input for computing plausibility of the answer.
**p_mask**: (`optional`) ``torch.FloatTensor`` of shape ``(batch_size, sequence_length)``:
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.
Outputs: `Tuple` comprising various elements depending on the configuration (config) and inputs:
**loss**: (`optional`, returned if both ``start_positions`` and ``end_positions`` are provided) ``torch.FloatTensor`` of shape ``(1,)``:
Classification loss as the sum of start token, end token (and is_impossible if provided) classification losses.
**start_top_log_probs**: (`optional`, returned if ``start_positions`` or ``end_positions`` is not provided)
``torch.FloatTensor`` of shape ``(batch_size, config.start_n_top)``
Log probabilities for the top config.start_n_top start token possibilities (beam-search).
**start_top_index**: (`optional`, returned if ``start_positions`` or ``end_positions`` is not provided)
``torch.LongTensor`` of shape ``(batch_size, config.start_n_top)``
Indices for the top config.start_n_top start token possibilities (beam-search).
**end_top_log_probs**: (`optional`, returned if ``start_positions`` or ``end_positions`` is not provided)
``torch.FloatTensor`` of shape ``(batch_size, config.start_n_top * config.end_n_top)``
Log probabilities for the top ``config.start_n_top * config.end_n_top`` end token possibilities (beam-search).
**end_top_index**: (`optional`, returned if ``start_positions`` or ``end_positions`` is not provided)
``torch.LongTensor`` of shape ``(batch_size, config.start_n_top * config.end_n_top)``
Indices for the top ``config.start_n_top * config.end_n_top`` end token possibilities (beam-search).
**cls_logits**: (`optional`, returned if ``start_positions`` or ``end_positions`` is not provided)
``torch.FloatTensor`` of shape ``(batch_size,)``
Log probabilities for the ``is_impossible`` label of the answers.
**mems**: (`optional`, returned when ``config.mem_len > 0``)
list of ``torch.FloatTensor`` (one for each layer):
that contains pre-computed hidden-states (key and values in the attention blocks) as computed by the model
if config.mem_len > 0 else tuple of None. Can be used to speed up sequential decoding and attend to longer context.
See details in the docstring of the `mems` input above.
**hidden_states**: (`optional`, returned when ``config.output_hidden_states=True``)
list of ``torch.FloatTensor`` (one for the output of each layer + the output of the embeddings)
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**: (`optional`, returned when ``config.output_attentions=True``)
list 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.
When ``target_mapping is not None``, the attentions outputs are a list of 2-tuple of ``torch.FloatTensor``.
Examples::
tokenizer = XLNetTokenizer.from_pretrained('xlnet-large-cased')
model = XLMForQuestionAnswering.from_pretrained('xlnet-large-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)
loss, start_scores, end_scores = outputs[:2]
"""
def __init__(self, config):
super().__init__(config)
@ -1606,6 +1498,7 @@ class XLNetForQuestionAnswering(XLNetPreTrainedModel):
self.init_weights()
@add_start_docstrings_to_callable(XLNET_INPUTS_DOCSTRING)
def forward(
self,
input_ids=None,
@ -1623,6 +1516,64 @@ class XLNetForQuestionAnswering(XLNetPreTrainedModel):
cls_index=None,
p_mask=None,
):
r"""
start_positions (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`, defaults to :obj:`None`):
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 are not taken into account for computing the loss.
end_positions (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`, defaults to :obj:`None`):
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 are not taken into account for computing the loss.
is_impossible (``torch.LongTensor`` of shape ``(batch_size,)``, `optional`, defaults to :obj:`None`):
Labels whether a question has an answer or no answer (SQuAD 2.0)
cls_index (``torch.LongTensor`` of shape ``(batch_size,)``, `optional`, defaults to :obj:`None`):
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`, defaults to :obj:`None`):
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:
:obj:`tuple(torch.FloatTensor)` comprising various elements depending on the configuration (config) and inputs:
loss (:obj:`torch.FloatTensor` of shape :obj:`(1,)`, `optional`, returned if both :obj:`start_positions` and :obj:`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):
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):
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 (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 (key and values in the attention blocks).
Can be used (see `past` 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 ``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 of the model at the output of each layer plus the initial embedding outputs.
attentions (:obj:`tuple(torch.FloatTensor)`, `optional`, returned 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 weights after the attention softmax, used to compute the weighted average in the self-attention
heads.
Examples::
tokenizer = XLNetTokenizer.from_pretrained('xlnet-large-cased')
model = XLMForQuestionAnswering.from_pretrained('xlnet-large-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)
loss, start_scores, end_scores = outputs[:2]
"""
transformer_outputs = self.transformer(
input_ids,
attention_mask=attention_mask,