/
modeling_electra.py
925 lines (766 loc) 路 42.3 KB
/
modeling_electra.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
import logging
import os
import warnings
import torch
import torch.nn as nn
from torch.nn import CrossEntropyLoss, MSELoss
from .activations import get_activation
from .configuration_electra import ElectraConfig
from .file_utils import add_code_sample_docstrings, add_start_docstrings, add_start_docstrings_to_callable
from .modeling_bert import BertEmbeddings, BertEncoder, BertLayerNorm, BertPreTrainedModel
from .modeling_utils import SequenceSummary
logger = logging.getLogger(__name__)
_TOKENIZER_FOR_DOC = "ElectraTokenizer"
ELECTRA_PRETRAINED_MODEL_ARCHIVE_LIST = [
"google/electra-small-generator",
"google/electra-base-generator",
"google/electra-large-generator",
"google/electra-small-discriminator",
"google/electra-base-discriminator",
"google/electra-large-discriminator",
# See all ELECTRA models at https://huggingface.co/models?filter=electra
]
def load_tf_weights_in_electra(model, config, tf_checkpoint_path, discriminator_or_generator="discriminator"):
""" Load tf checkpoints in a pytorch model.
"""
try:
import re
import numpy as np
import tensorflow as tf
except ImportError:
logger.error(
"Loading a TensorFlow model in PyTorch, requires TensorFlow to be installed. Please see "
"https://www.tensorflow.org/install/ for installation instructions."
)
raise
tf_path = os.path.abspath(tf_checkpoint_path)
logger.info("Converting TensorFlow checkpoint from {}".format(tf_path))
# Load weights from TF model
init_vars = tf.train.list_variables(tf_path)
names = []
arrays = []
for name, shape in init_vars:
logger.info("Loading TF weight {} with shape {}".format(name, shape))
array = tf.train.load_variable(tf_path, name)
names.append(name)
arrays.append(array)
for name, array in zip(names, arrays):
original_name: str = name
try:
if isinstance(model, ElectraForMaskedLM):
name = name.replace("electra/embeddings/", "generator/embeddings/")
if discriminator_or_generator == "generator":
name = name.replace("electra/", "discriminator/")
name = name.replace("generator/", "electra/")
name = name.replace("dense_1", "dense_prediction")
name = name.replace("generator_predictions/output_bias", "generator_lm_head/bias")
name = name.split("/")
# print(original_name, name)
# adam_v and adam_m are variables used in AdamWeightDecayOptimizer to calculated m and v
# which are not required for using pretrained model
if any(n in ["global_step", "temperature"] for n in name):
logger.info("Skipping {}".format(original_name))
continue
pointer = model
for m_name in name:
if re.fullmatch(r"[A-Za-z]+_\d+", m_name):
scope_names = re.split(r"_(\d+)", m_name)
else:
scope_names = [m_name]
if scope_names[0] == "kernel" or scope_names[0] == "gamma":
pointer = getattr(pointer, "weight")
elif scope_names[0] == "output_bias" or scope_names[0] == "beta":
pointer = getattr(pointer, "bias")
elif scope_names[0] == "output_weights":
pointer = getattr(pointer, "weight")
elif scope_names[0] == "squad":
pointer = getattr(pointer, "classifier")
else:
pointer = getattr(pointer, scope_names[0])
if len(scope_names) >= 2:
num = int(scope_names[1])
pointer = pointer[num]
if m_name.endswith("_embeddings"):
pointer = getattr(pointer, "weight")
elif m_name == "kernel":
array = np.transpose(array)
try:
assert pointer.shape == array.shape, original_name
except AssertionError as e:
e.args += (pointer.shape, array.shape)
raise
print("Initialize PyTorch weight {}".format(name), original_name)
pointer.data = torch.from_numpy(array)
except AttributeError as e:
print("Skipping {}".format(original_name), name, e)
continue
return model
class ElectraEmbeddings(BertEmbeddings):
"""Construct the embeddings from word, position and token_type embeddings."""
def __init__(self, config):
super().__init__(config)
self.word_embeddings = nn.Embedding(config.vocab_size, config.embedding_size, padding_idx=config.pad_token_id)
self.position_embeddings = nn.Embedding(config.max_position_embeddings, config.embedding_size)
self.token_type_embeddings = nn.Embedding(config.type_vocab_size, config.embedding_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 = BertLayerNorm(config.embedding_size, eps=config.layer_norm_eps)
class ElectraDiscriminatorPredictions(nn.Module):
"""Prediction module for the discriminator, made up of two dense layers."""
def __init__(self, config):
super().__init__()
self.dense = nn.Linear(config.hidden_size, config.hidden_size)
self.dense_prediction = nn.Linear(config.hidden_size, 1)
self.config = config
def forward(self, discriminator_hidden_states):
hidden_states = self.dense(discriminator_hidden_states)
hidden_states = get_activation(self.config.hidden_act)(hidden_states)
logits = self.dense_prediction(hidden_states).squeeze()
return logits
class ElectraGeneratorPredictions(nn.Module):
"""Prediction module for the generator, made up of two dense layers."""
def __init__(self, config):
super().__init__()
self.LayerNorm = BertLayerNorm(config.embedding_size)
self.dense = nn.Linear(config.hidden_size, config.embedding_size)
def forward(self, generator_hidden_states):
hidden_states = self.dense(generator_hidden_states)
hidden_states = get_activation("gelu")(hidden_states)
hidden_states = self.LayerNorm(hidden_states)
return hidden_states
class ElectraPreTrainedModel(BertPreTrainedModel):
""" An abstract class to handle weights initialization and
a simple interface for downloading and loading pretrained models.
"""
config_class = ElectraConfig
load_tf_weights = load_tf_weights_in_electra
base_model_prefix = "electra"
ELECTRA_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.ElectraConfig`): 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.
"""
ELECTRA_INPUTS_DOCSTRING = r"""
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.ElectraTokenizer`.
See :func:`transformers.PreTrainedTokenizer.encode` and
:func:`transformers.PreTrainedTokenizer.__call__` 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.
`What are attention masks? <../glossary.html#attention-mask>`__
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>`_
position_ids (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`, defaults to :obj:`None`):
Indices of positions of each input sequence tokens in the position embeddings.
Selected in the range ``[0, config.max_position_embeddings - 1]``.
`What are position IDs? <../glossary.html#position-ids>`_
head_mask (:obj:`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]``:
:obj:`1` indicates the head is **not masked**, :obj:`0` indicates the head is **masked**.
inputs_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.
encoder_hidden_states (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length, hidden_size)`, `optional`, defaults to :obj:`None`):
Sequence of hidden-states at the output of the last layer of the encoder. Used in the cross-attention
if the model is configured as a decoder.
encoder_attention_mask (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`, defaults to :obj:`None`):
Mask to avoid performing attention on the padding token indices of the encoder input. This mask
is used in the cross-attention if the model is configured as a decoder.
Mask values selected in ``[0, 1]``:
``1`` for tokens that are NOT MASKED, ``0`` for MASKED tokens.
output_attentions (:obj:`bool`, `optional`, defaults to :obj:`None`):
If set to ``True``, the attentions tensors of all attention layers are returned. See ``attentions`` under returned tensors for more detail.
"""
@add_start_docstrings(
"The bare Electra Model transformer outputting raw hidden-states without any specific head on top. Identical to "
"the BERT model except that it uses an additional linear layer between the embedding layer and the encoder if the "
"hidden size and embedding size are different."
""
"Both the generator and discriminator checkpoints may be loaded into this model.",
ELECTRA_START_DOCSTRING,
)
class ElectraModel(ElectraPreTrainedModel):
config_class = ElectraConfig
def __init__(self, config):
super().__init__(config)
self.embeddings = ElectraEmbeddings(config)
if config.embedding_size != config.hidden_size:
self.embeddings_project = nn.Linear(config.embedding_size, config.hidden_size)
self.encoder = BertEncoder(config)
self.config = config
self.init_weights()
def get_input_embeddings(self):
return self.embeddings.word_embeddings
def set_input_embeddings(self, value):
self.embeddings.word_embeddings = value
def _prune_heads(self, heads_to_prune):
""" Prunes heads of the model.
heads_to_prune: dict of {layer_num: list of heads to prune in this layer}
See base class PreTrainedModel
"""
for layer, heads in heads_to_prune.items():
self.encoder.layer[layer].attention.prune_heads(heads)
@add_start_docstrings_to_callable(ELECTRA_INPUTS_DOCSTRING)
@add_code_sample_docstrings(tokenizer_class=_TOKENIZER_FOR_DOC, checkpoint="google/electra-small-discriminator")
def forward(
self,
input_ids=None,
attention_mask=None,
token_type_ids=None,
position_ids=None,
head_mask=None,
inputs_embeds=None,
output_attentions=None,
output_hidden_states=None,
):
r"""
Return:
:obj:`tuple(torch.FloatTensor)` comprising various elements depending on the configuration (:class:`~transformers.ElectraConfig`) and inputs:
last_hidden_state (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length, hidden_size)`):
Sequence of hidden-states at the output of the last layer of the model.
hidden_states (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_hidden_states=True`` is passed or when ``config.output_hidden_states=True``):
Tuple of :obj:`torch.FloatTensor` (one for the output of the embeddings + one for the output of each layer)
of shape :obj:`(batch_size, sequence_length, hidden_size)`.
Hidden-states of the model at the output of each layer plus the initial embedding outputs.
attentions (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_attentions=True`` is passed or when ``config.output_attentions=True``):
Tuple of :obj:`torch.FloatTensor` (one for each layer) of shape
:obj:`(batch_size, num_heads, sequence_length, sequence_length)`.
Attentions weights after the attention softmax, used to compute the weighted average in the self-attention
heads.
"""
output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
output_hidden_states = (
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
)
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()
elif inputs_embeds is not None:
input_shape = inputs_embeds.size()[:-1]
else:
raise ValueError("You have to specify either input_ids or inputs_embeds")
device = input_ids.device if input_ids is not None else inputs_embeds.device
if attention_mask is None:
attention_mask = torch.ones(input_shape, device=device)
if token_type_ids is None:
token_type_ids = torch.zeros(input_shape, dtype=torch.long, device=device)
extended_attention_mask = self.get_extended_attention_mask(attention_mask, input_shape, device)
head_mask = self.get_head_mask(head_mask, self.config.num_hidden_layers)
hidden_states = self.embeddings(
input_ids=input_ids, position_ids=position_ids, token_type_ids=token_type_ids, inputs_embeds=inputs_embeds
)
if hasattr(self, "embeddings_project"):
hidden_states = self.embeddings_project(hidden_states)
hidden_states = self.encoder(
hidden_states,
attention_mask=extended_attention_mask,
head_mask=head_mask,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
)
return hidden_states
class ElectraClassificationHead(nn.Module):
"""Head for sentence-level classification tasks."""
def __init__(self, config):
super().__init__()
self.dense = nn.Linear(config.hidden_size, config.hidden_size)
self.dropout = nn.Dropout(config.hidden_dropout_prob)
self.out_proj = nn.Linear(config.hidden_size, config.num_labels)
def forward(self, features, **kwargs):
x = features[:, 0, :] # take <s> token (equiv. to [CLS])
x = self.dropout(x)
x = self.dense(x)
x = get_activation("gelu")(x) # although BERT uses tanh here, it seems Electra authors used gelu here
x = self.dropout(x)
x = self.out_proj(x)
return x
@add_start_docstrings(
"""ELECTRA Model transformer with a sequence classification/regression head on top (a linear layer on top of
the pooled output) e.g. for GLUE tasks. """,
ELECTRA_START_DOCSTRING,
)
class ElectraForSequenceClassification(ElectraPreTrainedModel):
def __init__(self, config):
super().__init__(config)
self.num_labels = config.num_labels
self.electra = ElectraModel(config)
self.classifier = ElectraClassificationHead(config)
self.init_weights()
@add_start_docstrings_to_callable(ELECTRA_INPUTS_DOCSTRING)
@add_code_sample_docstrings(tokenizer_class=_TOKENIZER_FOR_DOC, checkpoint="google/electra-small-discriminator")
def forward(
self,
input_ids=None,
attention_mask=None,
token_type_ids=None,
position_ids=None,
head_mask=None,
inputs_embeds=None,
labels=None,
output_attentions=None,
output_hidden_states=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 :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).
Returns:
:obj:`tuple(torch.FloatTensor)` comprising various elements depending on the configuration (:class:`~transformers.BertConfig`) and inputs:
loss (:obj:`torch.FloatTensor` of shape :obj:`(1,)`, `optional`, returned when :obj:`label` 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).
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.
attentions (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_attentions=True`` is passed or when ``config.output_attentions=True``):
Tuple of :obj:`torch.FloatTensor` (one for each layer) of shape
:obj:`(batch_size, num_heads, sequence_length, sequence_length)`.
Attentions weights after the attention softmax, used to compute the weighted average in the self-attention
heads.
"""
discriminator_hidden_states = self.electra(
input_ids,
attention_mask,
token_type_ids,
position_ids,
head_mask,
inputs_embeds,
output_attentions,
output_hidden_states,
)
sequence_output = discriminator_hidden_states[0]
logits = self.classifier(sequence_output)
outputs = (logits,) + discriminator_hidden_states[1:] # add hidden states and attention if they are here
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))
outputs = (loss,) + outputs
return outputs # (loss), logits, (hidden_states), (attentions)
@add_start_docstrings(
"""
Electra model with a binary classification head on top as used during pre-training for identifying generated
tokens.
It is recommended to load the discriminator checkpoint into that model.""",
ELECTRA_START_DOCSTRING,
)
class ElectraForPreTraining(ElectraPreTrainedModel):
def __init__(self, config):
super().__init__(config)
self.electra = ElectraModel(config)
self.discriminator_predictions = ElectraDiscriminatorPredictions(config)
self.init_weights()
@add_start_docstrings_to_callable(ELECTRA_INPUTS_DOCSTRING)
def forward(
self,
input_ids=None,
attention_mask=None,
token_type_ids=None,
position_ids=None,
head_mask=None,
inputs_embeds=None,
labels=None,
output_attentions=None,
output_hidden_states=None,
):
r"""
labels (``torch.LongTensor`` of shape ``(batch_size, sequence_length)``, `optional`, defaults to :obj:`None`):
Labels for computing the ELECTRA loss. Input should be a sequence of tokens (see :obj:`input_ids` docstring)
Indices should be in ``[0, 1]``.
``0`` indicates the token is an original token,
``1`` indicates the token was replaced.
Returns:
:obj:`tuple(torch.FloatTensor)` comprising various elements depending on the configuration (:class:`~transformers.ElectraConfig`) and inputs:
loss (`optional`, returned when ``labels`` is provided) ``torch.FloatTensor`` of shape ``(1,)``:
Total loss of the ELECTRA objective.
scores (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length)`)
Prediction scores of the head (scores for each token 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.
attentions (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_attentions=True`` is passed or when ``config.output_attentions=True``):
Tuple of :obj:`torch.FloatTensor` (one for each layer) of shape
:obj:`(batch_size, num_heads, sequence_length, sequence_length)`.
Attentions weights after the attention softmax, used to compute the weighted average in the self-attention
heads.
Examples::
>>> from transformers import ElectraTokenizer, ElectraForPreTraining
>>> import torch
>>> tokenizer = ElectraTokenizer.from_pretrained('google/electra-small-discriminator')
>>> model = ElectraForPreTraining.from_pretrained('google/electra-small-discriminator')
>>> input_ids = torch.tensor(tokenizer.encode("Hello, my dog is cute", add_special_tokens=True)).unsqueeze(0) # Batch size 1
>>> scores = model(input_ids)[0]
"""
discriminator_hidden_states = self.electra(
input_ids,
attention_mask,
token_type_ids,
position_ids,
head_mask,
inputs_embeds,
output_attentions,
output_hidden_states,
)
discriminator_sequence_output = discriminator_hidden_states[0]
logits = self.discriminator_predictions(discriminator_sequence_output)
output = (logits,)
if labels is not None:
loss_fct = nn.BCEWithLogitsLoss()
if attention_mask is not None:
active_loss = attention_mask.view(-1, discriminator_sequence_output.shape[1]) == 1
active_logits = logits.view(-1, discriminator_sequence_output.shape[1])[active_loss]
active_labels = labels[active_loss]
loss = loss_fct(active_logits, active_labels.float())
else:
loss = loss_fct(logits.view(-1, discriminator_sequence_output.shape[1]), labels.float())
output = (loss,) + output
output += discriminator_hidden_states[1:]
return output # (loss), scores, (hidden_states), (attentions)
@add_start_docstrings(
"""
Electra model with a language modeling head on top.
Even though both the discriminator and generator may be loaded into this model, the generator is
the only model of the two to have been trained for the masked language modeling task.""",
ELECTRA_START_DOCSTRING,
)
class ElectraForMaskedLM(ElectraPreTrainedModel):
def __init__(self, config):
super().__init__(config)
self.electra = ElectraModel(config)
self.generator_predictions = ElectraGeneratorPredictions(config)
self.generator_lm_head = nn.Linear(config.embedding_size, config.vocab_size)
self.init_weights()
def get_output_embeddings(self):
return self.generator_lm_head
@add_start_docstrings_to_callable(ELECTRA_INPUTS_DOCSTRING)
@add_code_sample_docstrings(tokenizer_class=_TOKENIZER_FOR_DOC, checkpoint="google/electra-small-generator")
def forward(
self,
input_ids=None,
attention_mask=None,
token_type_ids=None,
position_ids=None,
head_mask=None,
inputs_embeds=None,
labels=None,
output_attentions=None,
output_hidden_states=None,
**kwargs
):
r"""
labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`, defaults to :obj:`None`):
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]``
kwargs (:obj:`Dict[str, any]`, optional, defaults to `{}`):
Used to hide legacy arguments that have been deprecated.
Returns:
:obj:`tuple(torch.FloatTensor)` comprising various elements depending on the configuration (:class:`~transformers.ElectraConfig`) and inputs:
masked_lm_loss (`optional`, returned when ``labels`` is provided) ``torch.FloatTensor`` of shape ``(1,)``:
Masked 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).
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.
attentions (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_attentions=True`` is passed or when ``config.output_attentions=True``):
Tuple of :obj:`torch.FloatTensor` (one for each layer) of shape
:obj:`(batch_size, num_heads, sequence_length, sequence_length)`.
Attentions weights after the attention softmax, used to compute the weighted average in the self-attention
heads.
"""
if "masked_lm_labels" in kwargs:
warnings.warn(
"The `masked_lm_labels` argument is deprecated and will be removed in a future version, use `labels` instead.",
DeprecationWarning,
)
labels = kwargs.pop("masked_lm_labels")
assert kwargs == {}, f"Unexpected keyword arguments: {list(kwargs.keys())}."
generator_hidden_states = self.electra(
input_ids,
attention_mask,
token_type_ids,
position_ids,
head_mask,
inputs_embeds,
output_attentions,
output_hidden_states,
)
generator_sequence_output = generator_hidden_states[0]
prediction_scores = self.generator_predictions(generator_sequence_output)
prediction_scores = self.generator_lm_head(prediction_scores)
output = (prediction_scores,)
# Masked language modeling softmax layer
if labels is not None:
loss_fct = nn.CrossEntropyLoss() # -100 index = padding token
loss = loss_fct(prediction_scores.view(-1, self.config.vocab_size), labels.view(-1))
output = (loss,) + output
output += generator_hidden_states[1:]
return output # (masked_lm_loss), prediction_scores, (hidden_states), (attentions)
@add_start_docstrings(
"""
Electra model with a token classification head on top.
Both the discriminator and generator may be loaded into this model.""",
ELECTRA_START_DOCSTRING,
)
class ElectraForTokenClassification(ElectraPreTrainedModel):
def __init__(self, config):
super().__init__(config)
self.electra = ElectraModel(config)
self.dropout = nn.Dropout(config.hidden_dropout_prob)
self.classifier = nn.Linear(config.hidden_size, config.num_labels)
self.init_weights()
@add_start_docstrings_to_callable(ELECTRA_INPUTS_DOCSTRING)
@add_code_sample_docstrings(tokenizer_class=_TOKENIZER_FOR_DOC, checkpoint="google/electra-small-discriminator")
def forward(
self,
input_ids=None,
attention_mask=None,
token_type_ids=None,
position_ids=None,
head_mask=None,
inputs_embeds=None,
labels=None,
output_attentions=None,
output_hidden_states=None,
):
r"""
labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`, defaults to :obj:`None`):
Labels for computing the token classification loss.
Indices should be in ``[0, ..., config.num_labels - 1]``.
Returns:
:obj:`tuple(torch.FloatTensor)` comprising various elements depending on the configuration (:class:`~transformers.ElectraConfig`) and inputs:
loss (:obj:`torch.FloatTensor` of shape :obj:`(1,)`, `optional`, returned when ``labels`` is provided) :
Classification loss.
scores (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length, config.num_labels)`)
Classification scores (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.
attentions (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_attentions=True`` is passed or when ``config.output_attentions=True``):
Tuple of :obj:`torch.FloatTensor` (one for each layer) of shape
:obj:`(batch_size, num_heads, sequence_length, sequence_length)`.
Attentions weights after the attention softmax, used to compute the weighted average in the self-attention
heads.
"""
discriminator_hidden_states = self.electra(
input_ids,
attention_mask,
token_type_ids,
position_ids,
head_mask,
inputs_embeds,
output_attentions,
output_hidden_states,
)
discriminator_sequence_output = discriminator_hidden_states[0]
discriminator_sequence_output = self.dropout(discriminator_sequence_output)
logits = self.classifier(discriminator_sequence_output)
output = (logits,)
if labels is not None:
loss_fct = nn.CrossEntropyLoss()
# Only keep active parts of the loss
if attention_mask is not None:
active_loss = attention_mask.view(-1) == 1
active_logits = logits.view(-1, self.config.num_labels)[active_loss]
active_labels = labels.view(-1)[active_loss]
loss = loss_fct(active_logits, active_labels)
else:
loss = loss_fct(logits.view(-1, self.config.num_labels), labels.view(-1))
output = (loss,) + output
output += discriminator_hidden_states[1:]
return output # (loss), scores, (hidden_states), (attentions)
@add_start_docstrings(
"""
ELECTRA 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`).""",
ELECTRA_START_DOCSTRING,
)
class ElectraForQuestionAnswering(ElectraPreTrainedModel):
config_class = ElectraConfig
base_model_prefix = "electra"
def __init__(self, config):
super().__init__(config)
self.num_labels = config.num_labels
self.electra = ElectraModel(config)
self.qa_outputs = nn.Linear(config.hidden_size, config.num_labels)
self.init_weights()
@add_start_docstrings_to_callable(ELECTRA_INPUTS_DOCSTRING.format("(batch_size, sequence_length)"))
@add_code_sample_docstrings(tokenizer_class=_TOKENIZER_FOR_DOC, checkpoint="google/electra-small-discriminator")
def forward(
self,
input_ids=None,
attention_mask=None,
token_type_ids=None,
position_ids=None,
head_mask=None,
inputs_embeds=None,
start_positions=None,
end_positions=None,
output_attentions=None,
output_hidden_states=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 (:class:`~transformers.ElectraConfig`) 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).
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.
attentions (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_attentions=True`` is passed or when ``config.output_attentions=True``):
Tuple of :obj:`torch.FloatTensor` (one for each layer) of shape
:obj:`(batch_size, num_heads, sequence_length, sequence_length)`.
Attentions weights after the attention softmax, used to compute the weighted average in the self-attention
heads.
"""
discriminator_hidden_states = self.electra(
input_ids,
attention_mask=attention_mask,
token_type_ids=token_type_ids,
position_ids=position_ids,
head_mask=head_mask,
inputs_embeds=inputs_embeds,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
)
sequence_output = discriminator_hidden_states[0]
logits = self.qa_outputs(sequence_output)
start_logits, end_logits = logits.split(1, dim=-1)
start_logits = start_logits.squeeze(-1)
end_logits = end_logits.squeeze(-1)
outputs = (start_logits, end_logits,) + discriminator_hidden_states[1:]
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.clamp_(0, ignored_index)
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
outputs = (total_loss,) + outputs
return outputs # (loss), start_logits, end_logits, (hidden_states), (attentions)
@add_start_docstrings(
"""ELECTRA Model with a multiple choice classification head on top (a linear layer on top of
the pooled output and a softmax) e.g. for RocStories/SWAG tasks. """,
ELECTRA_INPUTS_DOCSTRING,
)
class ElectraForMultipleChoice(ElectraPreTrainedModel):
def __init__(self, config):
super().__init__(config)
self.electra = ElectraModel(config)
self.summary = SequenceSummary(config)
self.classifier = nn.Linear(config.hidden_size, 1)
self.init_weights()
@add_start_docstrings_to_callable(ELECTRA_INPUTS_DOCSTRING.format("(batch_size, num_choices, sequence_length)"))
@add_code_sample_docstrings(tokenizer_class=_TOKENIZER_FOR_DOC, checkpoint="google/electra-small-discriminator")
def forward(
self,
input_ids=None,
attention_mask=None,
token_type_ids=None,
position_ids=None,
head_mask=None,
inputs_embeds=None,
labels=None,
output_attentions=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-1]`` 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 (:class:`~transformers.ElectraConfig`) 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).
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.
attentions (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_attentions=True`` is passed or when ``config.output_attentions=True``):
Tuple of :obj:`torch.FloatTensor` (one for each layer) of shape
:obj:`(batch_size, num_heads, sequence_length, sequence_length)`.
Attentions weights after the attention softmax, used to compute the weighted average in the self-attention
heads.
"""
num_choices = input_ids.shape[1] if input_ids is not None else inputs_embeds.shape[1]
input_ids = input_ids.view(-1, input_ids.size(-1)) if input_ids is not None else None
attention_mask = attention_mask.view(-1, attention_mask.size(-1)) if attention_mask is not None else None
token_type_ids = token_type_ids.view(-1, token_type_ids.size(-1)) if token_type_ids is not None else None
position_ids = position_ids.view(-1, position_ids.size(-1)) if position_ids is not None else None
inputs_embeds = (
inputs_embeds.view(-1, inputs_embeds.size(-2), inputs_embeds.size(-1))
if inputs_embeds is not None
else None
)
discriminator_hidden_states = self.electra(
input_ids,
attention_mask=attention_mask,
token_type_ids=token_type_ids,
position_ids=position_ids,
head_mask=head_mask,
inputs_embeds=inputs_embeds,
output_attentions=output_attentions,
)
sequence_output = discriminator_hidden_states[0]
pooled_output = self.summary(sequence_output)
logits = self.classifier(pooled_output)
reshaped_logits = logits.view(-1, num_choices)
outputs = (reshaped_logits,) + discriminator_hidden_states[
1:
] # add hidden states and attention if they are here
if labels is not None:
loss_fct = CrossEntropyLoss()
loss = loss_fct(reshaped_logits, labels)
outputs = (loss,) + outputs
return outputs # (loss), reshaped_logits, (hidden_states), (attentions)