Change model outputs types to self-document outputs (#5438)

* [WIP] Proposal for model outputs

* All Bert models

* Make CI green maybe?

* Fix ONNX test

* Isolate ModelOutput from pt and tf

* Formatting

* Add Electra models

* Auto-generate docstrings from outputs

* Add TF outputs

* Add some BERT models

* Revert TF side

* Remove last traces of TF changes

* Fail with a clear error message

* Add Albert and work through Bart

* Add CTRL and DistilBert

* Formatting

* Progress on Bart

* Renames and finish Bart

* Formatting

* Fix last test

* Add DPR

* Finish Electra and add FlauBERT

* Add GPT2

* Add Longformer

* Add MMBT

* Add MobileBert

* Add GPT

* Formatting

* Add Reformer

* Add Roberta

* Add T5

* Add Transformer XL

* Fix test

* Add XLM + fix XLMForTokenClassification

* Style + XLMRoberta

* Add XLNet

* Formatting

* Add doc of return_tuple arg

tests/test_modeling_common.py

@@ -220,7 +220,6 @@ class ModelTesterMixin:
 
     def test_torchscript(self):
         config, inputs_dict = self.model_tester.prepare_config_and_inputs_for_common()
-
         self._create_and_check_torchscript(config, inputs_dict)
 
     def test_torchscript_output_attentions(self):
@@ -230,7 +229,6 @@ class ModelTesterMixin:
 
     def test_torchscript_output_hidden_state(self):
         config, inputs_dict = self.model_tester.prepare_config_and_inputs_for_common()
-
         config.output_hidden_states = True
         self._create_and_check_torchscript(config, inputs_dict)
 

tests/test_modeling_t5.py

@@ -355,6 +355,7 @@ class T5ModelTest(ModelTesterMixin, unittest.TestCase):
         import tempfile
 
         config_and_inputs = self.model_tester.prepare_config_and_inputs()
+        config_and_inputs[0].return_tuple = True
         model = T5Model(config_and_inputs[0])
         with tempfile.TemporaryDirectory() as tmpdirname:
             torch.onnx.export(

tests/test_modeling_tf_common.py

@@ -319,7 +319,7 @@ class TFModelTesterMixin:
             outputs_dict = model(input_ids)
             hidden_states = outputs_dict[0]
 
-            # Add a dense layer on top to test intetgration with other keras modules
+            # Add a dense layer on top to test integration with other keras modules
             outputs = tf.keras.layers.Dense(2, activation="softmax", name="outputs")(hidden_states)
 
             # Compile extended model

tests/test_modeling_xlm.py

@@ -347,6 +347,7 @@ class XLMModelTest(ModelTesterMixin, unittest.TestCase):
             XLMForQuestionAnswering,
             XLMForSequenceClassification,
             XLMForQuestionAnsweringSimple,
+            XLMForTokenClassification,
         )
         if is_torch_available()
         else ()

tests/test_modeling_xlnet.py

@@ -35,6 +35,7 @@ if is_torch_available():
         XLNetForSequenceClassification,
         XLNetForTokenClassification,
         XLNetForQuestionAnswering,
+        XLNetForQuestionAnsweringSimple,
     )
     from transformers.modeling_xlnet import XLNET_PRETRAINED_MODEL_ARCHIVE_LIST
 
@@ -458,6 +459,7 @@ class XLNetModelTest(ModelTesterMixin, unittest.TestCase):
             XLNetForTokenClassification,
             XLNetForSequenceClassification,
             XLNetForQuestionAnswering,
+            XLNetForQuestionAnsweringSimple,
             XLNetForMultipleChoice,
         )
         if is_torch_available()