[All models] Extend config.output_attentions with output_attentions function arguments (#4538)

* DOC: Replace instances of ``config.output_attentions`` with function argument ``output_attentions``

* DOC: Apply Black Formatting

* Fix errors where output_attentions was undefined

* Remove output_attentions in classes per review

* Fix regressions on tests having `output_attention`

* Fix further regressions in tests relating to `output_attentions`

Ensure proper propagation of `output_attentions` as a function parameter
to all model subclasses

* Fix more regressions in `test_output_attentions`

* Fix issues with BertEncoder

* Rename related variables to `output_attentions`

* fix pytorch tests

* fix bert and gpt2 tf

* Fix most TF tests for `test_output_attentions`

* Fix linter errors and more TF tests

* fix conflicts

* DOC: Apply Black Formatting

* Fix errors where output_attentions was undefined

* Remove output_attentions in classes per review

* Fix regressions on tests having `output_attention`

* fix conflicts

* fix conflicts

* fix conflicts

* fix conflicts

* fix pytorch tests

* fix conflicts

* fix conflicts

* Fix linter errors and more TF tests

* fix tf tests

* make style

* fix isort

* improve output_attentions

* improve tensorflow

Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>

templates/adding_a_new_model/modeling_tf_xxx.py

@@ -288,7 +288,7 @@ class TFXxxModel(TFXxxPreTrainedModel):
             list of ``tf.Tensor`` (one for the output of each layer + the output of the embeddings)
             of shape ``(batch_size, sequence_length, hidden_size)``:
             Hidden-states of the model at the output of each layer plus the initial embedding outputs.
-        **attentions**: (`optional`, returned when ``config.output_attentions=True``)
+        **attentions**: (`optional`, returned when ``output_attentions=True``)
             list of ``tf.Tensor`` (one for each layer) of shape ``(batch_size, num_heads, sequence_length, sequence_length)``:
             Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads.
 
@@ -329,7 +329,7 @@ class TFXxxForMaskedLM(TFXxxPreTrainedModel):
             list of ``Numpy array`` or ``tf.Tensor`` (one for the output of each layer + the output of the embeddings)
             of shape ``(batch_size, sequence_length, hidden_size)``:
             Hidden-states of the model at the output of each layer plus the initial embedding outputs.
-        **attentions**: (`optional`, returned when ``config.output_attentions=True``)
+        **attentions**: (`optional`, returned when ``output_attentions=True``)
             list of ``Numpy array`` or ``tf.Tensor`` (one for each layer) of shape ``(batch_size, num_heads, sequence_length, sequence_length)``:
             Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads.
 
@@ -378,7 +378,7 @@ class TFXxxForSequenceClassification(TFXxxPreTrainedModel):
             list of ``Numpy array`` or ``tf.Tensor`` (one for the output of each layer + the output of the embeddings)
             of shape ``(batch_size, sequence_length, hidden_size)``:
             Hidden-states of the model at the output of each layer plus the initial embedding outputs.
-        **attentions**: (`optional`, returned when ``config.output_attentions=True``)
+        **attentions**: (`optional`, returned when ``output_attentions=True``)
             list of ``Numpy array`` or ``tf.Tensor`` (one for each layer) of shape ``(batch_size, num_heads, sequence_length, sequence_length)``:
             Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads.
 
@@ -433,7 +433,7 @@ class TFXxxForTokenClassification(TFXxxPreTrainedModel):
             list of ``Numpy array`` or ``tf.Tensor`` (one for the output of each layer + the output of the embeddings)
             of shape ``(batch_size, sequence_length, hidden_size)``:
             Hidden-states of the model at the output of each layer plus the initial embedding outputs.
-        **attentions**: (`optional`, returned when ``config.output_attentions=True``)
+        **attentions**: (`optional`, returned when ``output_attentions=True``)
             list of ``Numpy array`` or ``tf.Tensor`` (one for each layer) of shape ``(batch_size, num_heads, sequence_length, sequence_length)``:
             Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads.
 
@@ -490,7 +490,7 @@ class TFXxxForQuestionAnswering(TFXxxPreTrainedModel):
             list of ``Numpy array`` or ``tf.Tensor`` (one for the output of each layer + the output of the embeddings)
             of shape ``(batch_size, sequence_length, hidden_size)``:
             Hidden-states of the model at the output of each layer plus the initial embedding outputs.
-        **attentions**: (`optional`, returned when ``config.output_attentions=True``)
+        **attentions**: (`optional`, returned when ``output_attentions=True``)
             list of ``Numpy array`` or ``tf.Tensor`` (one for each layer) of shape ``(batch_size, num_heads, sequence_length, sequence_length)``:
             Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads.
 

templates/adding_a_new_model/modeling_xxx.py

@@ -285,7 +285,7 @@ class XxxModel(XxxPreTrainedModel):
             list of ``torch.FloatTensor`` (one for the output of each layer + the output of the embeddings)
             of shape ``(batch_size, sequence_length, hidden_size)``:
             Hidden-states of the model at the output of each layer plus the initial embedding outputs.
-        **attentions**: (`optional`, returned when ``config.output_attentions=True``)
+        **attentions**: (`optional`, returned when ``output_attentions=True``)
             list of ``torch.FloatTensor`` (one for each layer) of shape ``(batch_size, num_heads, sequence_length, sequence_length)``:
             Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads.
 
@@ -403,7 +403,7 @@ class XxxForMaskedLM(XxxPreTrainedModel):
             list of ``torch.FloatTensor`` (one for the output of each layer + the output of the embeddings)
             of shape ``(batch_size, sequence_length, hidden_size)``:
             Hidden-states of the model at the output of each layer plus the initial embedding outputs.
-        **attentions**: (`optional`, returned when ``config.output_attentions=True``)
+        **attentions**: (`optional`, returned when ``output_attentions=True``)
             list of ``torch.FloatTensor`` (one for each layer) of shape ``(batch_size, num_heads, sequence_length, sequence_length)``:
             Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads.
 
@@ -483,7 +483,7 @@ class XxxForSequenceClassification(XxxPreTrainedModel):
             list of ``torch.FloatTensor`` (one for the output of each layer + the output of the embeddings)
             of shape ``(batch_size, sequence_length, hidden_size)``:
             Hidden-states of the model at the output of each layer plus the initial embedding outputs.
-        **attentions**: (`optional`, returned when ``config.output_attentions=True``)
+        **attentions**: (`optional`, returned when ``output_attentions=True``)
             list of ``torch.FloatTensor`` (one for each layer) of shape ``(batch_size, num_heads, sequence_length, sequence_length)``:
             Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads.
 
@@ -569,7 +569,7 @@ class XxxForTokenClassification(XxxPreTrainedModel):
             list of ``torch.FloatTensor`` (one for the output of each layer + the output of the embeddings)
             of shape ``(batch_size, sequence_length, hidden_size)``:
             Hidden-states of the model at the output of each layer plus the initial embedding outputs.
-        **attentions**: (`optional`, returned when ``config.output_attentions=True``)
+        **attentions**: (`optional`, returned when ``output_attentions=True``)
             list of ``torch.FloatTensor`` (one for each layer) of shape ``(batch_size, num_heads, sequence_length, sequence_length)``:
             Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads.
 
@@ -663,7 +663,7 @@ class XxxForQuestionAnswering(XxxPreTrainedModel):
             list of ``torch.FloatTensor`` (one for the output of each layer + the output of the embeddings)
             of shape ``(batch_size, sequence_length, hidden_size)``:
             Hidden-states of the model at the output of each layer plus the initial embedding outputs.
-        **attentions**: (`optional`, returned when ``config.output_attentions=True``)
+        **attentions**: (`optional`, returned when ``output_attentions=True``)
             list of ``torch.FloatTensor`` (one for each layer) of shape ``(batch_size, num_heads, sequence_length, sequence_length)``:
             Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads.