SUMIN/HuggingFace_transformer
1
0
Fork 0
Code Issues Pull Requests Actions 7 Packages Projects Releases Wiki Activity

Reformat (#9482)

Browse Source
This commit is contained in:
Julien Plu
2021-01-10 15:10:15 +01:00
committed by GitHub
parent 96f1f74aaf
commit 4f7022d68d
19 changed files with 151 additions and 415 deletions
Download Patch File Download Diff File Expand all files Collapse all files

35
src/transformers/models/mpnet/modeling_tf_mpnet.py
View File

@@ -805,6 +805,7 @@ class TFMPNetModel(TFMPNetPreTrainedModel):
)
return outputs
# Copied from transformers.models.bert.modeling_tf_bert.TFBertModel.serving_output
def serving_output(self, output):
hs = tf.convert_to_tensor(output.hidden_states) if self.config.output_hidden_states else None
attns = tf.convert_to_tensor(output.attentions) if self.config.output_attentions else None
@@ -942,15 +943,12 @@ class TFMPNetForMaskedLM(TFMPNetPreTrainedModel, TFMaskedLanguageModelingLoss):
attentions=outputs.attentions,
)
# Copied from transformers.models.bert.modeling_tf_bert.TFBertForMaskedLM.serving_output
def serving_output(self, output):
hs = tf.convert_to_tensor(output.hidden_states) if self.config.output_hidden_states else None
attns = tf.convert_to_tensor(output.attentions) if self.config.output_attentions else None
return TFMaskedLMOutput(
logits=output.logits,
hidden_states=hs,
attentions=attns,
)
return TFMaskedLMOutput(logits=output.logits, hidden_states=hs, attentions=attns)
class TFMPNetClassificationHead(tf.keras.layers.Layer):
@@ -1069,15 +1067,12 @@ class TFMPNetForSequenceClassification(TFMPNetPreTrainedModel, TFSequenceClassif
attentions=outputs.attentions,
)
# Copied from transformers.models.bert.modeling_tf_bert.TFBertForSequenceClassification.serving_output
def serving_output(self, output):
hs = tf.convert_to_tensor(output.hidden_states) if self.config.output_hidden_states else None
attns = tf.convert_to_tensor(output.attentions) if self.config.output_attentions else None
return TFSequenceClassifierOutput(
logits=output.logits,
hidden_states=hs,
attentions=attns,
)
return TFSequenceClassifierOutput(logits=output.logits, hidden_states=hs, attentions=attns)
@add_start_docstrings(
@@ -1216,15 +1211,12 @@ class TFMPNetForMultipleChoice(TFMPNetPreTrainedModel, TFMultipleChoiceLoss):
return self.serving_output(output)
# Copied from transformers.models.bert.modeling_tf_bert.TFBertForMultipleChoice.serving_output
def serving_output(self, output):
hs = tf.convert_to_tensor(output.hidden_states) if self.config.output_hidden_states else None
attns = tf.convert_to_tensor(output.attentions) if self.config.output_attentions else None
return TFMultipleChoiceModelOutput(
logits=output.logits,
hidden_states=hs,
attentions=attns,
)
return TFMultipleChoiceModelOutput(logits=output.logits, hidden_states=hs, attentions=attns)
@add_start_docstrings(
@@ -1321,15 +1313,12 @@ class TFMPNetForTokenClassification(TFMPNetPreTrainedModel, TFTokenClassificatio
attentions=outputs.attentions,
)
# Copied from transformers.models.bert.modeling_tf_bert.TFBertForTokenClassification.serving_output
def serving_output(self, output):
hs = tf.convert_to_tensor(output.hidden_states) if self.config.output_hidden_states else None
attns = tf.convert_to_tensor(output.attentions) if self.config.output_attentions else None
return TFTokenClassifierOutput(
logits=output.logits,
hidden_states=hs,
attentions=attns,
)
return TFTokenClassifierOutput(logits=output.logits, hidden_states=hs, attentions=attns)
@add_start_docstrings(
@@ -1439,13 +1428,11 @@ class TFMPNetForQuestionAnswering(TFMPNetPreTrainedModel, TFQuestionAnsweringLos
attentions=outputs.attentions,
)
# Copied from transformers.models.bert.modeling_tf_bert.TFBertForQuestionAnswering.serving_output
def serving_output(self, output):
hs = tf.convert_to_tensor(output.hidden_states) if self.config.output_hidden_states else None
attns = tf.convert_to_tensor(output.attentions) if self.config.output_attentions else None
return TFQuestionAnsweringModelOutput(
start_logits=output.start_logits,
end_logits=output.end_logits,
hidden_states=hs,
attentions=attns,
start_logits=output.start_logits, end_logits=output.end_logits, hidden_states=hs, attentions=attns
)