# coding=utf-8
# Copyright 2021 Google Research and The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" PyTorch FNet model. """
import warnings
from dataclasses import dataclass
from functools import partial
from typing import Optional, Tuple
import torch
import torch.utils.checkpoint
from packaging import version
from torch import nn
from torch.nn import CrossEntropyLoss, MSELoss
from ...file_utils import is_scipy_available
if is_scipy_available():
from scipy import linalg
from ...activations import ACT2FN
from ...file_utils import (
add_code_sample_docstrings,
add_start_docstrings,
add_start_docstrings_to_model_forward,
replace_return_docstrings,
)
from ...modeling_outputs import (
BaseModelOutput,
BaseModelOutputWithPooling,
MaskedLMOutput,
ModelOutput,
MultipleChoiceModelOutput,
NextSentencePredictorOutput,
QuestionAnsweringModelOutput,
SequenceClassifierOutput,
TokenClassifierOutput,
)
from ...modeling_utils import PreTrainedModel, apply_chunking_to_forward
from ...utils import logging
from .configuration_fnet import FNetConfig
logger = logging.get_logger(__name__)
_CHECKPOINT_FOR_DOC = "google/fnet-base"
_CONFIG_FOR_DOC = "FNetConfig"
_TOKENIZER_FOR_DOC = "FNetTokenizer"
FNET_PRETRAINED_MODEL_ARCHIVE_LIST = [
"google/fnet-base",
"google/fnet-large"
# See all FNet models at https://huggingface.co/models?filter=fnet
]
# Adapted from https://github.com/google-research/google-research/blob/master/f_net/fourier.py
def _two_dim_matmul(x, matrix_dim_one, matrix_dim_two):
"""Applies 2D matrix multiplication to 3D input arrays."""
seq_length = x.shape[1]
matrix_dim_one = matrix_dim_one[:seq_length, :seq_length]
x = x.type(torch.complex64)
return torch.einsum("bij,jk,ni->bnk", x, matrix_dim_two, matrix_dim_one)
# # Adapted from https://github.com/google-research/google-research/blob/master/f_net/fourier.py
def two_dim_matmul(x, matrix_dim_one, matrix_dim_two):
return _two_dim_matmul(x, matrix_dim_one, matrix_dim_two)
# Adapted from https://github.com/google-research/google-research/blob/master/f_net/fourier.py
def fftn(x):
"""
Applies n-dimensional Fast Fourier Transform (FFT) to input array.
Args:
x: Input n-dimensional array.
Returns:
n-dimensional Fourier transform of input n-dimensional array.
"""
out = x
for axis in reversed(range(x.ndim)[1:]): # We don't need to apply FFT to last axis
out = torch.fft.fft(out, axis=axis)
return out
class FNetEmbeddings(nn.Module):
"""Construct the embeddings from word, position and token_type embeddings."""
def __init__(self, config):
super().__init__()
self.word_embeddings = nn.Embedding(config.vocab_size, config.hidden_size, padding_idx=config.pad_token_id)
self.position_embeddings = nn.Embedding(config.max_position_embeddings, config.hidden_size)
self.token_type_embeddings = nn.Embedding(config.type_vocab_size, config.hidden_size)
# self.LayerNorm is not snake-cased to stick with TensorFlow model variable name and be able to load
# any TensorFlow checkpoint file
self.LayerNorm = nn.LayerNorm(config.hidden_size, eps=config.layer_norm_eps)
# NOTE: This is the project layer and will be needed. The original code allows for different embedding and different model dimensions.
self.projection = nn.Linear(config.hidden_size, config.hidden_size)
self.dropout = nn.Dropout(config.hidden_dropout_prob)
# position_ids (1, len position emb) is contiguous in memory and exported when serialized
self.register_buffer("position_ids", torch.arange(config.max_position_embeddings).expand((1, -1)))
if version.parse(torch.__version__) > version.parse("1.6.0"):
self.register_buffer(
"token_type_ids",
torch.zeros(self.position_ids.size(), dtype=torch.long, device=self.position_ids.device),
persistent=False,
)
def forward(self, input_ids=None, token_type_ids=None, position_ids=None, inputs_embeds=None):
if input_ids is not None:
input_shape = input_ids.size()
else:
input_shape = inputs_embeds.size()[:-1]
seq_length = input_shape[1]
if position_ids is None:
position_ids = self.position_ids[:, :seq_length]
# Setting the token_type_ids to the registered buffer in constructor where it is all zeros, which usually occurs
# when its auto-generated, registered buffer helps users when tracing the model without passing token_type_ids, solves
# issue #5664
if token_type_ids is None:
if hasattr(self, "token_type_ids"):
buffered_token_type_ids = self.token_type_ids[:, :seq_length]
buffered_token_type_ids_expanded = buffered_token_type_ids.expand(input_shape[0], seq_length)
token_type_ids = buffered_token_type_ids_expanded
else:
token_type_ids = torch.zeros(input_shape, dtype=torch.long, device=self.position_ids.device)
if inputs_embeds is None:
inputs_embeds = self.word_embeddings(input_ids)
token_type_embeddings = self.token_type_embeddings(token_type_ids)
embeddings = inputs_embeds + token_type_embeddings
position_embeddings = self.position_embeddings(position_ids)
embeddings += position_embeddings
embeddings = self.LayerNorm(embeddings)
embeddings = self.projection(embeddings)
embeddings = self.dropout(embeddings)
return embeddings
class FNetBasicFourierTransform(nn.Module):
def __init__(self, config):
super().__init__()
self._init_fourier_transform(config)
def _init_fourier_transform(self, config):
if not config.use_tpu_fourier_optimizations:
self.fourier_transform = partial(torch.fft.fftn, dim=(1, 2))
elif config.max_position_embeddings <= 4096:
if is_scipy_available():
self.register_buffer(
"dft_mat_hidden", torch.tensor(linalg.dft(config.hidden_size), dtype=torch.complex64)
)
self.register_buffer(
"dft_mat_seq", torch.tensor(linalg.dft(config.tpu_short_seq_length), dtype=torch.complex64)
)
self.fourier_transform = partial(
two_dim_matmul, matrix_dim_one=self.dft_mat_seq, matrix_dim_two=self.dft_mat_hidden
)
else:
logging.warning(
"SciPy is needed for DFT matrix calculation and is not found. Using TPU optimized fast fourier transform instead."
)
self.fourier_transform = fftn
else:
self.fourier_transform = fftn
def forward(self, hidden_states):
# NOTE: We do not use torch.vmap as it is not integrated into PyTorch stable versions.
# Interested users can modify the code to use vmap from the nightly versions, getting the vmap from here:
# https://pytorch.org/docs/master/generated/torch.vmap.html. Note that fourier transform methods will need
# change accordingly.
outputs = self.fourier_transform(hidden_states).real
return (outputs,)
class FNetBasicOutput(nn.Module):
def __init__(self, config):
super().__init__()
self.LayerNorm = nn.LayerNorm(config.hidden_size, eps=config.layer_norm_eps)
def forward(self, hidden_states, input_tensor):
hidden_states = self.LayerNorm(input_tensor + hidden_states)
return hidden_states
class FNetFourierTransform(nn.Module):
def __init__(self, config):
super().__init__()
self.self = FNetBasicFourierTransform(config)
self.output = FNetBasicOutput(config)
def forward(self, hidden_states):
self_outputs = self.self(hidden_states)
fourier_output = self.output(self_outputs[0], hidden_states)
outputs = (fourier_output,)
return outputs
# Copied from transformers.models.bert.modeling_bert.BertIntermediate with Bert->FNet
class FNetIntermediate(nn.Module):
def __init__(self, config):
super().__init__()
self.dense = nn.Linear(config.hidden_size, config.intermediate_size)
if isinstance(config.hidden_act, str):
self.intermediate_act_fn = ACT2FN[config.hidden_act]
else:
self.intermediate_act_fn = config.hidden_act
def forward(self, hidden_states):
hidden_states = self.dense(hidden_states)
hidden_states = self.intermediate_act_fn(hidden_states)
return hidden_states
# Copied from transformers.models.bert.modeling_bert.BertOutput with Bert->FNet
class FNetOutput(nn.Module):
def __init__(self, config):
super().__init__()
self.dense = nn.Linear(config.intermediate_size, config.hidden_size)
self.LayerNorm = nn.LayerNorm(config.hidden_size, eps=config.layer_norm_eps)
self.dropout = nn.Dropout(config.hidden_dropout_prob)
def forward(self, hidden_states, input_tensor):
hidden_states = self.dense(hidden_states)
hidden_states = self.dropout(hidden_states)
hidden_states = self.LayerNorm(hidden_states + input_tensor)
return hidden_states
class FNetLayer(nn.Module):
def __init__(self, config):
super().__init__()
self.chunk_size_feed_forward = config.chunk_size_feed_forward
self.seq_len_dim = 1 # The dimension which has the sequence length
self.fourier = FNetFourierTransform(config)
self.intermediate = FNetIntermediate(config)
self.output = FNetOutput(config)
def forward(self, hidden_states):
self_fourier_outputs = self.fourier(hidden_states)
fourier_output = self_fourier_outputs[0]
layer_output = apply_chunking_to_forward(
self.feed_forward_chunk, self.chunk_size_feed_forward, self.seq_len_dim, fourier_output
)
outputs = (layer_output,)
return outputs
def feed_forward_chunk(self, fourier_output):
intermediate_output = self.intermediate(fourier_output)
layer_output = self.output(intermediate_output, fourier_output)
return layer_output
class FNetEncoder(nn.Module):
def __init__(self, config):
super().__init__()
self.config = config
self.layer = nn.ModuleList([FNetLayer(config) for _ in range(config.num_hidden_layers)])
self.gradient_checkpointing = False
def forward(self, hidden_states, output_hidden_states=False, return_dict=True):
all_hidden_states = () if output_hidden_states else None
for i, layer_module in enumerate(self.layer):
if output_hidden_states:
all_hidden_states = all_hidden_states + (hidden_states,)
if self.gradient_checkpointing and self.training:
def create_custom_forward(module):
def custom_forward(*inputs):
return module(*inputs)
return custom_forward
layer_outputs = torch.utils.checkpoint.checkpoint(create_custom_forward(layer_module), hidden_states)
else:
layer_outputs = layer_module(hidden_states)
hidden_states = layer_outputs[0]
if output_hidden_states:
all_hidden_states = all_hidden_states + (hidden_states,)
if not return_dict:
return tuple(v for v in [hidden_states, all_hidden_states] if v is not None)
return BaseModelOutput(last_hidden_state=hidden_states, hidden_states=all_hidden_states)
# Copied from transformers.models.bert.modeling_bert.BertPooler with Bert->FNet
class FNetPooler(nn.Module):
def __init__(self, config):
super().__init__()
self.dense = nn.Linear(config.hidden_size, config.hidden_size)
self.activation = nn.Tanh()
def forward(self, hidden_states):
# We "pool" the model by simply taking the hidden state corresponding
# to the first token.
first_token_tensor = hidden_states[:, 0]
pooled_output = self.dense(first_token_tensor)
pooled_output = self.activation(pooled_output)
return pooled_output
# Copied from transformers.models.bert.modeling_bert.BertPredictionHeadTransform with Bert->FNet
class FNetPredictionHeadTransform(nn.Module):
def __init__(self, config):
super().__init__()
self.dense = nn.Linear(config.hidden_size, config.hidden_size)
if isinstance(config.hidden_act, str):
self.transform_act_fn = ACT2FN[config.hidden_act]
else:
self.transform_act_fn = config.hidden_act
self.LayerNorm = nn.LayerNorm(config.hidden_size, eps=config.layer_norm_eps)
def forward(self, hidden_states):
hidden_states = self.dense(hidden_states)
hidden_states = self.transform_act_fn(hidden_states)
hidden_states = self.LayerNorm(hidden_states)
return hidden_states
class FNetLMPredictionHead(nn.Module):
def __init__(self, config):
super().__init__()
self.transform = FNetPredictionHeadTransform(config)
# The output weights are the same as the input embeddings, but there is
# an output-only bias for each token.
self.decoder = nn.Linear(config.hidden_size, config.vocab_size)
self.bias = nn.Parameter(torch.zeros(config.vocab_size))
self.decoder.bias = self.bias
def forward(self, hidden_states):
hidden_states = self.transform(hidden_states)
hidden_states = self.decoder(hidden_states)
return hidden_states
def _tie_weights(self):
# To tie those two weights if they get disconnected (on TPU or when the bias is resized)
self.bias = self.decoder.bias
class FNetOnlyMLMHead(nn.Module):
def __init__(self, config):
super().__init__()
self.predictions = FNetLMPredictionHead(config)
def forward(self, sequence_output):
prediction_scores = self.predictions(sequence_output)
return prediction_scores
# Copied from transformers.models.bert.modeling_bert.BertOnlyNSPHead with Bert->FNet
class FNetOnlyNSPHead(nn.Module):
def __init__(self, config):
super().__init__()
self.seq_relationship = nn.Linear(config.hidden_size, 2)
def forward(self, pooled_output):
seq_relationship_score = self.seq_relationship(pooled_output)
return seq_relationship_score
# Copied from transformers.models.bert.modeling_bert.BertPreTrainingHeads with Bert->FNet
class FNetPreTrainingHeads(nn.Module):
def __init__(self, config):
super().__init__()
self.predictions = FNetLMPredictionHead(config)
self.seq_relationship = nn.Linear(config.hidden_size, 2)
def forward(self, sequence_output, pooled_output):
prediction_scores = self.predictions(sequence_output)
seq_relationship_score = self.seq_relationship(pooled_output)
return prediction_scores, seq_relationship_score
class FNetPreTrainedModel(PreTrainedModel):
"""
An abstract class to handle weights initialization and a simple interface for downloading and loading pretrained
models.
"""
config_class = FNetConfig
base_model_prefix = "fnet"
supports_gradient_checkpointing = True
_keys_to_ignore_on_load_missing = [r"position_ids"]
def _init_weights(self, module):
"""Initialize the weights"""
if isinstance(module, nn.Linear):
# Slightly different from the TF version which uses truncated_normal for initialization
# cf https://github.com/pytorch/pytorch/pull/5617
module.weight.data.normal_(mean=0.0, std=self.config.initializer_range)
# NOTE: Original code uses same initialization as weights for biases as well.
if module.bias is not None:
module.bias.data.zero_()
elif isinstance(module, nn.Embedding):
module.weight.data.normal_(mean=0.0, std=self.config.initializer_range)
if module.padding_idx is not None:
module.weight.data[module.padding_idx].zero_()
elif isinstance(module, nn.LayerNorm):
module.bias.data.zero_()
module.weight.data.fill_(1.0)
def _set_gradient_checkpointing(self, module, value=False):
if isinstance(module, FNetEncoder):
module.gradient_checkpointing = value
@dataclass
class FNetForPreTrainingOutput(ModelOutput):
"""
Output type of :class:`~transformers.FNetForPreTraining`.
Args:
loss (`optional`, returned when ``labels`` is provided, ``torch.FloatTensor`` of shape :obj:`(1,)`):
Total loss as the sum of the masked language modeling loss and the next sequence prediction
(classification) loss.
prediction_logits (: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).
seq_relationship_logits (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, 2)`):
Prediction scores of the next sequence prediction (classification) head (scores of True/False continuation
before SoftMax).
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 of the model at the output of
each layer plus the initial embedding outputs.
"""
loss: Optional[torch.FloatTensor] = None
prediction_logits: torch.FloatTensor = None
seq_relationship_logits: torch.FloatTensor = None
hidden_states: Optional[Tuple[torch.FloatTensor]] = None
FNET_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.FNetConfig`): 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.
"""
FNET_INPUTS_DOCSTRING = r"""
Args:
input_ids (:obj:`torch.LongTensor` of shape :obj:`({0})`):
Indices of input sequence tokens in the vocabulary.
Indices can be obtained using :class:`transformers.FNetTokenizer`. See
:func:`transformers.PreTrainedTokenizer.encode` and :func:`transformers.PreTrainedTokenizer.__call__` for
details.
`What are input IDs? <../glossary.html#input-ids>`__
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]``:
- 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 position IDs? <../glossary.html#position-ids>`_
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
than the model's internal embedding lookup matrix.
output_hidden_states (:obj:`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.
"""
[docs]@add_start_docstrings(
"The bare FNet Model transformer outputting raw hidden-states without any specific head on top.",
FNET_START_DOCSTRING,
)
class FNetModel(FNetPreTrainedModel):
"""
The model can behave as an encoder, following the architecture described in `FNet: Mixing Tokens with Fourier
Transforms <https://arxiv.org/abs/2105.03824>`__ by James Lee-Thorp, Joshua Ainslie, Ilya Eckstein, Santiago
Ontanon.
"""
def __init__(self, config, add_pooling_layer=True):
super().__init__(config)
self.config = config
self.embeddings = FNetEmbeddings(config)
self.encoder = FNetEncoder(config)
self.pooler = FNetPooler(config) if add_pooling_layer else None
self.init_weights()
def get_input_embeddings(self):
return self.embeddings.word_embeddings
def set_input_embeddings(self, value):
self.embeddings.word_embeddings = value
[docs] @add_start_docstrings_to_model_forward(FNET_INPUTS_DOCSTRING.format("batch_size, sequence_length"))
@add_code_sample_docstrings(
tokenizer_class=_TOKENIZER_FOR_DOC,
checkpoint=_CHECKPOINT_FOR_DOC,
output_type=BaseModelOutput,
config_class=_CONFIG_FOR_DOC,
)
def forward(
self,
input_ids=None,
token_type_ids=None,
position_ids=None,
inputs_embeds=None,
output_hidden_states=None,
return_dict=None,
):
output_hidden_states = (
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
)
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
if input_ids is not None and inputs_embeds is not None:
raise ValueError("You cannot specify both input_ids and inputs_embeds at the same time")
elif input_ids is not None:
input_shape = input_ids.size()
batch_size, seq_length = input_shape
elif inputs_embeds is not None:
input_shape = inputs_embeds.size()[:-1]
batch_size, seq_length = input_shape
else:
raise ValueError("You have to specify either input_ids or inputs_embeds")
if (
self.config.use_tpu_fourier_optimizations
and seq_length <= 4096
and self.config.tpu_short_seq_length != seq_length
):
raise ValueError(
"The `tpu_short_seq_length` in FNetConfig should be set equal to the sequence length being passed to the model when using TPU optimizations."
)
device = input_ids.device if input_ids is not None else inputs_embeds.device
if token_type_ids is None:
if hasattr(self.embeddings, "token_type_ids"):
buffered_token_type_ids = self.embeddings.token_type_ids[:, :seq_length]
buffered_token_type_ids_expanded = buffered_token_type_ids.expand(batch_size, seq_length)
token_type_ids = buffered_token_type_ids_expanded
else:
token_type_ids = torch.zeros(input_shape, dtype=torch.long, device=device)
embedding_output = self.embeddings(
input_ids=input_ids,
position_ids=position_ids,
token_type_ids=token_type_ids,
inputs_embeds=inputs_embeds,
)
encoder_outputs = self.encoder(
embedding_output,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
sequence_output = encoder_outputs[0]
pooler_output = self.pooler(sequence_output) if self.pooler is not None else None
if not return_dict:
return (sequence_output, pooler_output) + encoder_outputs[1:]
return BaseModelOutputWithPooling(
last_hidden_state=sequence_output,
pooler_output=pooler_output,
hidden_states=encoder_outputs.hidden_states,
)
[docs]@add_start_docstrings(
"""
FNet Model with two heads on top as done during the pretraining: a `masked language modeling` head and a `next
sentence prediction (classification)` head.
""",
FNET_START_DOCSTRING,
)
class FNetForPreTraining(FNetPreTrainedModel):
def __init__(self, config):
super().__init__(config)
self.fnet = FNetModel(config)
self.cls = FNetPreTrainingHeads(config)
self.init_weights()
def get_output_embeddings(self):
return self.cls.predictions.decoder
def set_output_embeddings(self, new_embeddings):
self.cls.predictions.decoder = new_embeddings
[docs] @add_start_docstrings_to_model_forward(FNET_INPUTS_DOCSTRING.format("batch_size, sequence_length"))
@replace_return_docstrings(output_type=FNetForPreTrainingOutput, config_class=_CONFIG_FOR_DOC)
def forward(
self,
input_ids=None,
token_type_ids=None,
position_ids=None,
inputs_embeds=None,
labels=None,
next_sentence_label=None,
output_hidden_states=None,
return_dict=None,
):
r"""
labels (:obj:`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]``
next_sentence_label (``torch.LongTensor`` of shape ``(batch_size,)``, `optional`):
Labels for computing the next sequence prediction (classification) loss. Input should be a sequence pair
(see :obj:`input_ids` docstring) Indices should be in ``[0, 1]``:
- 0 indicates sequence B is a continuation of sequence A,
- 1 indicates sequence B is a random sequence.
kwargs (:obj:`Dict[str, any]`, optional, defaults to `{}`):
Used to hide legacy arguments that have been deprecated.
Returns:
Example::
>>> from transformers import FNetTokenizer, FNetForPreTraining
>>> import torch
>>> tokenizer = FNetTokenizer.from_pretrained('google/fnet-base')
>>> model = FNetForPreTraining.from_pretrained('google/fnet-base')
>>> inputs = tokenizer("Hello, my dog is cute", return_tensors="pt")
>>> outputs = model(**inputs)
>>> prediction_logits = outputs.prediction_logits
>>> seq_relationship_logits = outputs.seq_relationship_logits
"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
outputs = self.fnet(
input_ids,
token_type_ids=token_type_ids,
position_ids=position_ids,
inputs_embeds=inputs_embeds,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
sequence_output, pooled_output = outputs[:2]
prediction_scores, seq_relationship_score = self.cls(sequence_output, pooled_output)
total_loss = None
if labels is not None and next_sentence_label is not None:
loss_fct = CrossEntropyLoss()
masked_lm_loss = loss_fct(prediction_scores.view(-1, self.config.vocab_size), labels.view(-1))
next_sentence_loss = loss_fct(seq_relationship_score.view(-1, 2), next_sentence_label.view(-1))
total_loss = masked_lm_loss + next_sentence_loss
if not return_dict:
output = (prediction_scores, seq_relationship_score) + outputs[2:]
return ((total_loss,) + output) if total_loss is not None else output
return FNetForPreTrainingOutput(
loss=total_loss,
prediction_logits=prediction_scores,
seq_relationship_logits=seq_relationship_score,
hidden_states=outputs.hidden_states,
)
[docs]@add_start_docstrings(
"""FNet Model with a `next sentence prediction (classification)` head on top. """,
FNET_START_DOCSTRING,
)
class FNetForNextSentencePrediction(FNetPreTrainedModel):
def __init__(self, config):
super().__init__(config)
self.fnet = FNetModel(config)
self.cls = FNetOnlyNSPHead(config)
self.init_weights()
[docs] @add_start_docstrings_to_model_forward(FNET_INPUTS_DOCSTRING.format("batch_size, sequence_length"))
@replace_return_docstrings(output_type=NextSentencePredictorOutput, config_class=_CONFIG_FOR_DOC)
def forward(
self,
input_ids=None,
token_type_ids=None,
position_ids=None,
inputs_embeds=None,
labels=None,
output_hidden_states=None,
return_dict=None,
**kwargs,
):
r"""
labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`):
Labels for computing the next sequence prediction (classification) loss. Input should be a sequence pair
(see ``input_ids`` docstring). Indices should be in ``[0, 1]``:
- 0 indicates sequence B is a continuation of sequence A,
- 1 indicates sequence B is a random sequence.
Returns:
Example::
>>> from transformers import FNetTokenizer, FNetForNextSentencePrediction
>>> import torch
>>> tokenizer = FNetTokenizer.from_pretrained('google/fnet-base')
>>> model = FNetForNextSentencePrediction.from_pretrained('google/fnet-base')
>>> prompt = "In Italy, pizza served in formal settings, such as at a restaurant, is presented unsliced."
>>> next_sentence = "The sky is blue due to the shorter wavelength of blue light."
>>> encoding = tokenizer(prompt, next_sentence, return_tensors='pt')
>>> outputs = model(**encoding, labels=torch.LongTensor([1]))
>>> logits = outputs.logits
>>> assert logits[0, 0] < logits[0, 1] # next sentence was random
"""
if "next_sentence_label" in kwargs:
warnings.warn(
"The `next_sentence_label` argument is deprecated and will be removed in a future version, use `labels` instead.",
FutureWarning,
)
labels = kwargs.pop("next_sentence_label")
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
outputs = self.fnet(
input_ids,
token_type_ids=token_type_ids,
position_ids=position_ids,
inputs_embeds=inputs_embeds,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
pooled_output = outputs[1]
seq_relationship_scores = self.cls(pooled_output)
next_sentence_loss = None
if labels is not None:
loss_fct = CrossEntropyLoss()
next_sentence_loss = loss_fct(seq_relationship_scores.view(-1, 2), labels.view(-1))
if not return_dict:
output = (seq_relationship_scores,) + outputs[2:]
return ((next_sentence_loss,) + output) if next_sentence_loss is not None else output
return NextSentencePredictorOutput(
loss=next_sentence_loss,
logits=seq_relationship_scores,
hidden_states=outputs.hidden_states,
)
[docs]@add_start_docstrings(
"""
FNet Model transformer with a sequence classification/regression head on top (a linear layer on top of the pooled
output) e.g. for GLUE tasks.
""",
FNET_START_DOCSTRING,
)
class FNetForSequenceClassification(FNetPreTrainedModel):
def __init__(self, config):
super().__init__(config)
self.num_labels = config.num_labels
self.fnet = FNetModel(config)
self.dropout = nn.Dropout(config.hidden_dropout_prob)
self.classifier = nn.Linear(config.hidden_size, config.num_labels)
self.init_weights()
[docs] @add_start_docstrings_to_model_forward(FNET_INPUTS_DOCSTRING.format("batch_size, sequence_length"))
@add_code_sample_docstrings(
tokenizer_class=_TOKENIZER_FOR_DOC,
checkpoint=_CHECKPOINT_FOR_DOC,
output_type=SequenceClassifierOutput,
config_class=_CONFIG_FOR_DOC,
)
def forward(
self,
input_ids=None,
token_type_ids=None,
position_ids=None,
inputs_embeds=None,
labels=None,
output_hidden_states=None,
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).
"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
outputs = self.fnet(
input_ids,
token_type_ids=token_type_ids,
position_ids=position_ids,
inputs_embeds=inputs_embeds,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
pooled_output = outputs[1]
pooled_output = self.dropout(pooled_output)
logits = self.classifier(pooled_output)
loss = None
if labels is not None:
if self.num_labels == 1:
# We are doing regression
loss_fct = MSELoss()
loss = loss_fct(logits.view(-1), labels.view(-1))
else:
loss_fct = CrossEntropyLoss()
loss = loss_fct(logits.view(-1, self.num_labels), labels.view(-1))
if not return_dict:
output = (logits,) + outputs[2:]
return ((loss,) + output) if loss is not None else output
return SequenceClassifierOutput(loss=loss, logits=logits, hidden_states=outputs.hidden_states)
[docs]@add_start_docstrings(
"""
FNet 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.
""",
FNET_START_DOCSTRING,
)
class FNetForTokenClassification(FNetPreTrainedModel):
def __init__(self, config):
super().__init__(config)
self.num_labels = config.num_labels
self.fnet = FNetModel(config)
self.dropout = nn.Dropout(config.hidden_dropout_prob)
self.classifier = nn.Linear(config.hidden_size, config.num_labels)
self.init_weights()
[docs] @add_start_docstrings_to_model_forward(FNET_INPUTS_DOCSTRING.format("batch_size, sequence_length"))
@add_code_sample_docstrings(
tokenizer_class=_TOKENIZER_FOR_DOC,
checkpoint=_CHECKPOINT_FOR_DOC,
output_type=TokenClassifierOutput,
config_class=_CONFIG_FOR_DOC,
)
def forward(
self,
input_ids=None,
token_type_ids=None,
position_ids=None,
inputs_embeds=None,
labels=None,
output_hidden_states=None,
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]``.
"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
outputs = self.fnet(
input_ids,
token_type_ids=token_type_ids,
position_ids=position_ids,
inputs_embeds=inputs_embeds,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
sequence_output = outputs[0]
sequence_output = self.dropout(sequence_output)
logits = self.classifier(sequence_output)
loss = None
if labels is not None:
loss_fct = CrossEntropyLoss()
# Only keep active parts of the loss
loss = loss_fct(logits.view(-1, self.num_labels), labels.view(-1))
if not return_dict:
output = (logits,) + outputs[2:]
return ((loss,) + output) if loss is not None else output
return TokenClassifierOutput(loss=loss, logits=logits, hidden_states=outputs.hidden_states)
[docs]@add_start_docstrings(
"""
FNet 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`).
""",
FNET_START_DOCSTRING,
)
class FNetForQuestionAnswering(FNetPreTrainedModel):
def __init__(self, config):
super().__init__(config)
self.num_labels = config.num_labels
self.fnet = FNetModel(config)
self.qa_outputs = nn.Linear(config.hidden_size, config.num_labels)
self.init_weights()
[docs] @add_start_docstrings_to_model_forward(FNET_INPUTS_DOCSTRING.format("batch_size, sequence_length"))
@add_code_sample_docstrings(
tokenizer_class=_TOKENIZER_FOR_DOC,
checkpoint=_CHECKPOINT_FOR_DOC,
output_type=QuestionAnsweringModelOutput,
config_class=_CONFIG_FOR_DOC,
)
def forward(
self,
input_ids=None,
token_type_ids=None,
position_ids=None,
inputs_embeds=None,
start_positions=None,
end_positions=None,
output_hidden_states=None,
return_dict=None,
):
r"""
start_positions (:obj:`torch.LongTensor` of shape :obj:`(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
sequence are not taken into account for computing the loss.
end_positions (:obj:`torch.LongTensor` of shape :obj:`(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
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
outputs = self.fnet(
input_ids,
token_type_ids=token_type_ids,
position_ids=position_ids,
inputs_embeds=inputs_embeds,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
sequence_output = outputs[0]
logits = self.qa_outputs(sequence_output)
start_logits, end_logits = logits.split(1, dim=-1)
start_logits = start_logits.squeeze(-1).contiguous()
end_logits = end_logits.squeeze(-1).contiguous()
total_loss = None
if start_positions is not None and end_positions is not None:
# If we are on multi-GPU, split add a dimension
if len(start_positions.size()) > 1:
start_positions = start_positions.squeeze(-1)
if len(end_positions.size()) > 1:
end_positions = end_positions.squeeze(-1)
# sometimes the start/end positions are outside our model inputs, we ignore these terms
ignored_index = start_logits.size(1)
start_positions = start_positions.clamp(0, ignored_index)
end_positions = end_positions.clamp(0, ignored_index)
loss_fct = CrossEntropyLoss(ignore_index=ignored_index)
start_loss = loss_fct(start_logits, start_positions)
end_loss = loss_fct(end_logits, end_positions)
total_loss = (start_loss + end_loss) / 2
if not return_dict:
output = (start_logits, end_logits) + outputs[2:]
return ((total_loss,) + output) if total_loss is not None else output
return QuestionAnsweringModelOutput(
loss=total_loss, start_logits=start_logits, end_logits=end_logits, hidden_states=outputs.hidden_states
)