Fix aria tests (#39879)

* fix aria tests

* awful bug

* fix copies

* fix tests

* fix style

* revert this

src/transformers/models/aria/modeling_aria.py

@@ -1014,18 +1014,9 @@ class AriaModel(AriaPreTrainedModel):
         past_key_values: Optional[Cache] = None,
         inputs_embeds: Optional[torch.FloatTensor] = None,
         use_cache: Optional[bool] = None,
-        output_attentions: Optional[bool] = None,
-        output_hidden_states: Optional[bool] = None,
-        return_dict: Optional[bool] = None,
         cache_position: Optional[torch.LongTensor] = None,
         **kwargs: Unpack[FlashAttentionKwargs],
     ) -> Union[tuple, AriaModelOutputWithPast]:
-        output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
-        output_hidden_states = (
-            output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
-        )
-        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
-
         if inputs_embeds is None:
             inputs_embeds = self.get_input_embeddings()(input_ids)
 
@@ -1037,7 +1028,7 @@ class AriaModel(AriaPreTrainedModel):
                 vision_feature_layer=self.config.vision_feature_layer,
             )
             image_features = image_features.to(inputs_embeds.device, inputs_embeds.dtype)
-            special_image_mask = self._get_image_mask(
+            special_image_mask = self.get_placeholder_mask(
                 input_ids, inputs_embeds=inputs_embeds, image_features=image_features
             )
             inputs_embeds = inputs_embeds.masked_scatter(special_image_mask, image_features)
@@ -1048,9 +1039,6 @@ class AriaModel(AriaPreTrainedModel):
             past_key_values=past_key_values,
             inputs_embeds=inputs_embeds,
             use_cache=use_cache,
-            output_attentions=output_attentions,
-            output_hidden_states=output_hidden_states,
-            return_dict=True,
             cache_position=cache_position,
             **kwargs,
         )
@@ -1156,9 +1144,6 @@ class AriaForConditionalGeneration(AriaPreTrainedModel, GenerationMixin):
         inputs_embeds: Optional[torch.FloatTensor] = None,
         labels: Optional[torch.LongTensor] = None,
         use_cache: Optional[bool] = None,
-        output_attentions: Optional[bool] = None,
-        output_hidden_states: Optional[bool] = None,
-        return_dict: Optional[bool] = None,
         logits_to_keep: Union[int, torch.Tensor] = 0,
         cache_position: Optional[torch.LongTensor] = None,
         **kwargs: Unpack[TransformersKwargs],
@@ -1223,12 +1208,6 @@ class AriaForConditionalGeneration(AriaPreTrainedModel, GenerationMixin):
         >>> print(generated_texts[1])
         Assistant: The bridge is in San Francisco.
         ```"""
-        output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
-        output_hidden_states = (
-            output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
-        )
-        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
-
         outputs = self.model(
             input_ids=input_ids,
             pixel_values=pixel_values,
@@ -1238,9 +1217,6 @@ class AriaForConditionalGeneration(AriaPreTrainedModel, GenerationMixin):
             past_key_values=past_key_values,
             inputs_embeds=inputs_embeds,
             use_cache=use_cache,
-            output_attentions=output_attentions,
-            output_hidden_states=output_hidden_states,
-            return_dict=return_dict,
             cache_position=cache_position,
             **kwargs,
         )

src/transformers/models/aria/modular_aria.py

@@ -1414,18 +1414,9 @@ class AriaModel(LlavaModel):
         past_key_values: Optional[Cache] = None,
         inputs_embeds: Optional[torch.FloatTensor] = None,
         use_cache: Optional[bool] = None,
-        output_attentions: Optional[bool] = None,
-        output_hidden_states: Optional[bool] = None,
-        return_dict: Optional[bool] = None,
         cache_position: Optional[torch.LongTensor] = None,
         **kwargs: Unpack[FlashAttentionKwargs],
     ) -> Union[tuple, AriaModelOutputWithPast]:
-        output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
-        output_hidden_states = (
-            output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
-        )
-        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
-
         if inputs_embeds is None:
             inputs_embeds = self.get_input_embeddings()(input_ids)
 
@@ -1437,7 +1428,7 @@ class AriaModel(LlavaModel):
                 vision_feature_layer=self.config.vision_feature_layer,
             )
             image_features = image_features.to(inputs_embeds.device, inputs_embeds.dtype)
-            special_image_mask = self._get_image_mask(
+            special_image_mask = self.get_placeholder_mask(
                 input_ids, inputs_embeds=inputs_embeds, image_features=image_features
             )
             inputs_embeds = inputs_embeds.masked_scatter(special_image_mask, image_features)
@@ -1448,9 +1439,6 @@ class AriaModel(LlavaModel):
             past_key_values=past_key_values,
             inputs_embeds=inputs_embeds,
             use_cache=use_cache,
-            output_attentions=output_attentions,
-            output_hidden_states=output_hidden_states,
-            return_dict=True,
             cache_position=cache_position,
             **kwargs,
         )
@@ -1498,9 +1486,6 @@ class AriaForConditionalGeneration(LlavaForConditionalGeneration):
         inputs_embeds: Optional[torch.FloatTensor] = None,
         labels: Optional[torch.LongTensor] = None,
         use_cache: Optional[bool] = None,
-        output_attentions: Optional[bool] = None,
-        output_hidden_states: Optional[bool] = None,
-        return_dict: Optional[bool] = None,
         logits_to_keep: Union[int, torch.Tensor] = 0,
         cache_position: Optional[torch.LongTensor] = None,
         **kwargs: Unpack[TransformersKwargs],
@@ -1565,12 +1550,6 @@ class AriaForConditionalGeneration(LlavaForConditionalGeneration):
         >>> print(generated_texts[1])
         Assistant: The bridge is in San Francisco.
         ```"""
-        output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
-        output_hidden_states = (
-            output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
-        )
-        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
-
         outputs = self.model(
             input_ids=input_ids,
             pixel_values=pixel_values,
@@ -1580,9 +1559,6 @@ class AriaForConditionalGeneration(LlavaForConditionalGeneration):
             past_key_values=past_key_values,
             inputs_embeds=inputs_embeds,
             use_cache=use_cache,
-            output_attentions=output_attentions,
-            output_hidden_states=output_hidden_states,
-            return_dict=return_dict,
             cache_position=cache_position,
             **kwargs,
         )

tests/models/aria/test_modeling_aria.py

@@ -137,8 +137,8 @@ class AriaVisionText2TextModelTester:
 
     def get_config(self):
         return AriaConfig(
-            text_config=self.text_config,
+            text_config=self.text_config.to_dict(),
-            vision_config=self.vision_config,
+            vision_config=self.vision_config.to_dict(),
             ignore_index=self.ignore_index,
             image_token_index=self.image_token_index,
             projector_hidden_act=self.projector_hidden_act,