Skip to content

[All Seq2Seq model + CLM models that can be used with EncoderDecoder] Add cross-attention weights to outputs #8071

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 8 commits into from
Nov 6, 2020
Merged

[All Seq2Seq model + CLM models that can be used with EncoderDecoder] Add cross-attention weights to outputs #8071

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
8 commits
Select commit Hold shift + click to select a range
925092d
Output cross-attention with decoder attention output
Oct 26, 2020
4c182fa
Update src/transformers/modeling_bert.py
patrickvonplaten Nov 6, 2020
bd7eb88
Merge remote-tracking branch 'main/master' into output-crossattention
patrickvonplaten Nov 6, 2020
8f8575f
add cross-attention for t5 and bart as well
patrickvonplaten Nov 6, 2020
d9a5e35
fix tests
patrickvonplaten Nov 6, 2020
518acf6
correct typo in docs
patrickvonplaten Nov 6, 2020
9225cab
add sylvains and sams comments
patrickvonplaten Nov 6, 2020
0205b70
correct typo
patrickvonplaten Nov 6, 2020
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
36 changes: 36 additions & 0 deletions docs/source/main_classes/output.rst
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -65,12 +65,34 @@ BaseModelOutputWithPooling
:members:


BaseModelOutputWithCrossAttentions
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

.. autoclass:: transformers.modeling_outputs.BaseModelOutputWithCrossAttentions
:members:


BaseModelOutputWithPoolingAndCrossAttentions
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

.. autoclass:: transformers.modeling_outputs.BaseModelOutputWithPoolingAndCrossAttentions
:members:


BaseModelOutputWithPast
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

.. autoclass:: transformers.modeling_outputs.BaseModelOutputWithPast
:members:


BaseModelOutputWithPastAndCrossAttentions
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

.. autoclass:: transformers.modeling_outputs.BaseModelOutputWithPastAndCrossAttentions
:members:


Seq2SeqModelOutput
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Expand All @@ -85,6 +107,20 @@ CausalLMOutput
:members:


CausalLMOutputWithCrossAttentions
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

.. autoclass:: transformers.modeling_outputs.CausalLMOutputWithCrossAttentions
:members:


CausalLMOutputWithPastAndCrossAttentions
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

.. autoclass:: transformers.modeling_outputs.CausalLMOutputWithPastAndCrossAttentions
:members:


CausalLMOutputWithPast
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Expand Down
28 changes: 21 additions & 7 deletions src/transformers/modeling_bart.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -35,7 +35,7 @@
)
from .modeling_outputs import (
BaseModelOutput,
BaseModelOutputWithPast,
BaseModelOutputWithPastAndCrossAttentions,
Seq2SeqLMOutput,
Seq2SeqModelOutput,
Seq2SeqQuestionAnsweringModelOutput,
Expand Down Expand Up @@ -451,11 +451,12 @@ def forward(
assert self.encoder_attn.cache_key != self.self_attn.cache_key
if self.normalize_before:
x = self.encoder_attn_layer_norm(x)
x, _ = self.encoder_attn(
x, cross_attn_weights = self.encoder_attn(
query=x,
key=encoder_hidden_states,
key_padding_mask=encoder_attn_mask,
layer_state=layer_state, # mutates layer state
output_attentions=output_attentions,
)
x = F.dropout(x, p=self.dropout, training=self.training)
x = residual + x
Expand All @@ -477,7 +478,8 @@ def forward(
x,
self_attn_weights,
layer_state,
) # just self_attn weights for now, following t5, layer_state = cache for decoding
cross_attn_weights,
) # layer_state = cache for decoding


class BartDecoder(nn.Module):
Expand Down Expand Up @@ -590,6 +592,7 @@ def forward(
# decoder layers
all_hidden_states = () if output_hidden_states else None
all_self_attns = () if output_attentions else None
all_cross_attentions = () if output_attentions else None
next_decoder_cache: List[Dict] = []
for idx, decoder_layer in enumerate(self.layers):
# add LayerDrop (see https://arxiv.org/abs/1909.11556 for description)
Expand All @@ -601,7 +604,7 @@ def forward(

layer_state = past_key_values[idx] if past_key_values is not None else None

x, layer_self_attn, layer_past = decoder_layer(
x, layer_self_attn, layer_past, layer_cross_attn = decoder_layer(
x,
encoder_hidden_states,
encoder_attn_mask=encoder_padding_mask,
Expand All @@ -616,6 +619,7 @@ def forward(

if output_attentions:
all_self_attns += (layer_self_attn,)
all_cross_attentions += (layer_cross_attn,)

if self.layer_norm: # if config.add_final_layer_norm (mBART)
x = self.layer_norm(x)
Expand All @@ -628,9 +632,15 @@ def forward(

next_cache = next_decoder_cache if use_cache else None
if not return_dict:
return tuple(v for v in [x, next_cache, all_hidden_states, all_self_attns] if v is not None)
return BaseModelOutputWithPast(
last_hidden_state=x, past_key_values=next_cache, hidden_states=all_hidden_states, attentions=all_self_attns
return tuple(
v for v in [x, next_cache, all_hidden_states, all_self_attns, all_cross_attentions] if v is not None
)
return BaseModelOutputWithPastAndCrossAttentions(
last_hidden_state=x,
past_key_values=next_cache,
hidden_states=all_hidden_states,
attentions=all_self_attns,
cross_attentions=all_cross_attentions,
)


Expand Down Expand Up @@ -934,6 +944,7 @@ def forward(
past_key_values=decoder_outputs.past_key_values,
decoder_hidden_states=decoder_outputs.hidden_states,
decoder_attentions=decoder_outputs.attentions,
cross_attentions=decoder_outputs.cross_attentions,
encoder_last_hidden_state=encoder_outputs.last_hidden_state,
encoder_hidden_states=encoder_outputs.hidden_states,
encoder_attentions=encoder_outputs.attentions,
Expand Down Expand Up @@ -1078,6 +1089,7 @@ def forward(
past_key_values=outputs.past_key_values,
decoder_hidden_states=outputs.decoder_hidden_states,
decoder_attentions=outputs.decoder_attentions,
cross_attentions=outputs.cross_attentions,
encoder_last_hidden_state=outputs.encoder_last_hidden_state,
encoder_hidden_states=outputs.encoder_hidden_states,
encoder_attentions=outputs.encoder_attentions,
Expand Down Expand Up @@ -1207,6 +1219,7 @@ def forward(
past_key_values=outputs.past_key_values,
decoder_hidden_states=outputs.decoder_hidden_states,
decoder_attentions=outputs.decoder_attentions,
cross_attentions=outputs.cross_attentions,
encoder_last_hidden_state=outputs.encoder_last_hidden_state,
encoder_hidden_states=outputs.encoder_hidden_states,
encoder_attentions=outputs.encoder_attentions,
Expand Down Expand Up @@ -1317,6 +1330,7 @@ def forward(
past_key_values=outputs.past_key_values,
decoder_hidden_states=outputs.decoder_hidden_states,
decoder_attentions=outputs.decoder_attentions,
cross_attentions=outputs.cross_attentions,
encoder_last_hidden_state=outputs.encoder_last_hidden_state,
encoder_hidden_states=outputs.encoder_hidden_states,
encoder_attentions=outputs.encoder_attentions,
Expand Down
36 changes: 24 additions & 12 deletions src/transformers/modeling_bert.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -37,9 +37,9 @@
replace_return_docstrings,
)
from .modeling_outputs import (
BaseModelOutput,
BaseModelOutputWithPooling,
CausalLMOutput,
BaseModelOutputWithCrossAttentions,
BaseModelOutputWithPoolingAndCrossAttentions,
CausalLMOutputWithCrossAttentions,
MaskedLMOutput,
MultipleChoiceModelOutput,
NextSentencePredictorOutput,
Expand Down Expand Up @@ -449,7 +449,8 @@ def forward(
return_dict=False,
):
all_hidden_states = () if output_hidden_states else None
all_attentions = () if output_attentions else None
all_self_attentions = () if output_attentions else None
all_cross_attentions = () if output_attentions and self.config.add_cross_attention else None
for i, layer_module in enumerate(self.layer):
if output_hidden_states:
all_hidden_states = all_hidden_states + (hidden_states,)
Expand Down Expand Up @@ -483,15 +484,24 @@ def custom_forward(*inputs):
)
hidden_states = layer_outputs[0]
if output_attentions:
all_attentions = all_attentions + (layer_outputs[1],)
all_self_attentions = all_self_attentions + (layer_outputs[1],)
if self.config.add_cross_attention:
all_cross_attentions = all_cross_attentions + (layer_outputs[2],)

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, all_attentions] if v is not None)
return BaseModelOutput(
last_hidden_state=hidden_states, hidden_states=all_hidden_states, attentions=all_attentions
return tuple(
v
for v in [hidden_states, all_hidden_states, all_self_attentions, all_cross_attentions]
if v is not None
)
return BaseModelOutputWithCrossAttentions(
last_hidden_state=hidden_states,
hidden_states=all_hidden_states,
attentions=all_self_attentions,
cross_attentions=all_cross_attentions,
)


Expand Down Expand Up @@ -752,7 +762,7 @@ class PreTrainedModel
@add_code_sample_docstrings(
tokenizer_class=_TOKENIZER_FOR_DOC,
checkpoint="bert-base-uncased",
output_type=BaseModelOutputWithPooling,
output_type=BaseModelOutputWithPoolingAndCrossAttentions,
config_class=_CONFIG_FOR_DOC,
)
def forward(
Expand Down Expand Up @@ -843,11 +853,12 @@ def forward(
if not return_dict:
return (sequence_output, pooled_output) + encoder_outputs[1:]

return BaseModelOutputWithPooling(
return BaseModelOutputWithPoolingAndCrossAttentions(
last_hidden_state=sequence_output,
pooler_output=pooled_output,
hidden_states=encoder_outputs.hidden_states,
attentions=encoder_outputs.attentions,
cross_attentions=encoder_outputs.cross_attentions,
)


Expand Down Expand Up @@ -984,7 +995,7 @@ def get_output_embeddings(self):
return self.cls.predictions.decoder

@add_start_docstrings_to_model_forward(BERT_INPUTS_DOCSTRING.format("batch_size, sequence_length"))
@replace_return_docstrings(output_type=CausalLMOutput, config_class=_CONFIG_FOR_DOC)
@replace_return_docstrings(output_type=CausalLMOutputWithCrossAttentions, config_class=_CONFIG_FOR_DOC)
def forward(
self,
input_ids=None,
Expand Down Expand Up @@ -1063,11 +1074,12 @@ def forward(
output = (prediction_scores,) + outputs[2:]
return ((lm_loss,) + output) if lm_loss is not None else output

return CausalLMOutput(
return CausalLMOutputWithCrossAttentions(
loss=lm_loss,
logits=prediction_scores,
hidden_states=outputs.hidden_states,
attentions=outputs.attentions,
cross_attentions=outputs.cross_attentions,
)

def prepare_inputs_for_generation(self, input_ids, attention_mask=None, **model_kwargs):
Expand Down
12 changes: 7 additions & 5 deletions src/transformers/modeling_bert_generation.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -28,7 +28,7 @@
replace_return_docstrings,
)
from .modeling_bert import BertEncoder
from .modeling_outputs import BaseModelOutput, CausalLMOutput
from .modeling_outputs import BaseModelOutputWithCrossAttentions, CausalLMOutputWithCrossAttentions
from .modeling_utils import PreTrainedModel
from .utils import logging

Expand Down Expand Up @@ -297,7 +297,7 @@ class PreTrainedModel
@add_code_sample_docstrings(
tokenizer_class=_TOKENIZER_FOR_DOC,
checkpoint="google/bert_for_seq_generation_L-24_bbc_encoder",
output_type=BaseModelOutput,
output_type=BaseModelOutputWithCrossAttentions,
config_class=_CONFIG_FOR_DOC,
)
def forward(
Expand Down Expand Up @@ -381,10 +381,11 @@ def forward(
if not return_dict:
return (sequence_output,) + encoder_outputs[1:]

return BaseModelOutput(
return BaseModelOutputWithCrossAttentions(
last_hidden_state=sequence_output,
hidden_states=encoder_outputs.hidden_states,
attentions=encoder_outputs.attentions,
cross_attentions=encoder_outputs.cross_attentions,
)


Expand Down Expand Up @@ -422,7 +423,7 @@ def get_output_embeddings(self):
return self.lm_head.decoder

@add_start_docstrings_to_model_forward(BERT_GENERATION_INPUTS_DOCSTRING.format("batch_size, sequence_length"))
@replace_return_docstrings(output_type=CausalLMOutput, config_class=_CONFIG_FOR_DOC)
@replace_return_docstrings(output_type=CausalLMOutputWithCrossAttentions, config_class=_CONFIG_FOR_DOC)
def forward(
self,
input_ids=None,
Expand Down Expand Up @@ -499,11 +500,12 @@ def forward(
output = (prediction_scores,) + outputs[1:]
return ((lm_loss,) + output) if lm_loss is not None else output

return CausalLMOutput(
return CausalLMOutputWithCrossAttentions(
loss=lm_loss,
logits=prediction_scores,
hidden_states=outputs.hidden_states,
attentions=outputs.attentions,
cross_attentions=outputs.cross_attentions,
)

def prepare_inputs_for_generation(self, input_ids, attention_mask=None, **model_kwargs):
Expand Down
24 changes: 17 additions & 7 deletions src/transformers/modeling_electra.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -34,7 +34,7 @@
replace_return_docstrings,
)
from .modeling_outputs import (
BaseModelOutput,
BaseModelOutputWithCrossAttentions,
MaskedLMOutput,
MultipleChoiceModelOutput,
QuestionAnsweringModelOutput,
Expand Down Expand Up @@ -445,7 +445,8 @@ def forward(
return_dict=False,
):
all_hidden_states = () if output_hidden_states else None
all_attentions = () if output_attentions else None
all_self_attentions = () if output_attentions else None
all_cross_attentions = () if output_attentions and self.config.add_cross_attention else None
for i, layer_module in enumerate(self.layer):
if output_hidden_states:
all_hidden_states = all_hidden_states + (hidden_states,)
Expand Down Expand Up @@ -479,15 +480,24 @@ def custom_forward(*inputs):
)
hidden_states = layer_outputs[0]
if output_attentions:
all_attentions = all_attentions + (layer_outputs[1],)
all_self_attentions = all_self_attentions + (layer_outputs[1],)
if self.config.add_cross_attention:
all_cross_attentions = all_cross_attentions + (layer_outputs[2],)

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, all_attentions] if v is not None)
return BaseModelOutput(
last_hidden_state=hidden_states, hidden_states=all_hidden_states, attentions=all_attentions
return tuple(
v
for v in [hidden_states, all_hidden_states, all_self_attentions, all_cross_attentions]
if v is not None
)
return BaseModelOutputWithCrossAttentions(
last_hidden_state=hidden_states,
hidden_states=all_hidden_states,
attentions=all_self_attentions,
cross_attentions=all_cross_attentions,
)


Expand Down Expand Up @@ -697,7 +707,7 @@ class PreTrainedModel
@add_code_sample_docstrings(
tokenizer_class=_TOKENIZER_FOR_DOC,
checkpoint="google/electra-small-discriminator",
output_type=BaseModelOutput,
output_type=BaseModelOutputWithCrossAttentions,
config_class=_CONFIG_FOR_DOC,
)
def forward(
Expand Down
1 change: 1 addition & 0 deletions src/transformers/modeling_encoder_decoder.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -426,6 +426,7 @@ def forward(
past_key_values=None, # TODO(PVP) - need to implement cache for BERT, etc... before this works
decoder_hidden_states=decoder_outputs.hidden_states,
decoder_attentions=decoder_outputs.attentions,
cross_attentions=decoder_outputs.cross_attentions,
encoder_last_hidden_state=encoder_outputs.last_hidden_state,
encoder_hidden_states=encoder_outputs.hidden_states,
encoder_attentions=encoder_outputs.attentions,
Expand Down
Loading