Fix examples (#11990)

src/transformers/models/visual_bert/modeling_visual_bert.py

@@ -728,6 +728,9 @@ class VisualBertModel(VisualBertPreTrainedModel):
         return_dict=None,
     ):
         r"""
+
+        Returns:
+
         Example::
 
             >>> # Assumption: `get_visual_embeddings(image)` gets the visual embeddings of the image.
@@ -1016,6 +1019,7 @@ class VisualBertForMultipleChoice(VisualBertPreTrainedModel):
     @add_start_docstrings_to_model_forward(
         VISUAL_BERT_INPUTS_DOCSTRING.format("batch_size, num_choices, sequence_length")
     )
+    @replace_return_docstrings(output_type=MultipleChoiceModelOutput, config_class=_CONFIG_FOR_DOC)
     def forward(
         self,
         input_ids=None,
@@ -1039,6 +1043,8 @@ class VisualBertForMultipleChoice(VisualBertPreTrainedModel):
                 num_choices-1]`` where :obj:`num_choices` is the size of the second dimension of the input tensors.
                 (See :obj:`input_ids` above)
 
+        Returns:
+
         Example::
 
             >>> from transformers import BertTokenizer, VisualBertForMultipleChoice
@@ -1160,6 +1166,7 @@ class VisualBertForQuestionAnswering(VisualBertPreTrainedModel):
         self.init_weights()
 
     @add_start_docstrings_to_model_forward(VISUAL_BERT_INPUTS_DOCSTRING.format("batch_size, sequence_length"))
+    @replace_return_docstrings(output_type=SequenceClassifierOutput, config_class=_CONFIG_FOR_DOC)
     def forward(
         self,
         input_ids=None,
@@ -1182,6 +1189,7 @@ class VisualBertForQuestionAnswering(VisualBertPreTrainedModel):
                 Labels for computing the sequence classification/regression loss. Indices should be in :obj:`[0, ...,
                 config.num_labels - 1]`. A KLDivLoss is computed between the labels and the returned logits.
 
+        Returns:
 
         Example::
 
@@ -1280,6 +1288,7 @@ class VisualBertForVisualReasoning(VisualBertPreTrainedModel):
         self.init_weights()
 
     @add_start_docstrings_to_model_forward(VISUAL_BERT_INPUTS_DOCSTRING.format("batch_size, sequence_length"))
+    @replace_return_docstrings(output_type=SequenceClassifierOutput, config_class=_CONFIG_FOR_DOC)
     def forward(
         self,
         input_ids=None,
@@ -1302,6 +1311,8 @@ class VisualBertForVisualReasoning(VisualBertPreTrainedModel):
                 Labels for computing the sequence classification/regression loss. Indices should be in :obj:`[0, ...,
                 config.num_labels - 1]`. A classification loss is computed (Cross-Entropy) against these labels.
 
+        Returns:
+
         Example::
 
             >>> # Assumption: `get_visual_embeddings(image)` gets the visual embeddings of the image in the batch.
@@ -1433,6 +1444,7 @@ class VisualBertForRegionToPhraseAlignment(VisualBertPreTrainedModel):
         self.init_weights()
 
     @add_start_docstrings_to_model_forward(VISUAL_BERT_INPUTS_DOCSTRING.format("batch_size, sequence_length"))
+    @replace_return_docstrings(output_type=SequenceClassifierOutput, config_class=_CONFIG_FOR_DOC)
     def forward(
         self,
         input_ids=None,
@@ -1459,6 +1471,8 @@ class VisualBertForRegionToPhraseAlignment(VisualBertPreTrainedModel):
                 Labels for computing the masked language modeling loss. KLDivLoss is computed against these labels and
                 the outputs from the attention layer.
 
+        Returns:
+
         Example::
 
             >>> # Assumption: `get_visual_embeddings(image)` gets the visual embeddings of the image in the batch.