Clean-up composite configs (#34603)

* remove manual assignment tie-word-embeddings

* remove another unused attribute

* fix tests

* fix tests

* remove unnecessary overwrites

* fix

* decoder=True

* clean pix2struct

* run-all

* forgot `_tied_weights_keys` when adding Emu3

* also Aria + fix-copies

* and clean aria

tests/models/blip_2/test_modeling_blip_2.py

@@ -1123,6 +1123,13 @@ class Blip2ModelTest(ModelTesterMixin, PipelineTesterMixin, GenerationTesterMixi
 
     def setUp(self):
         self.model_tester = Blip2ModelTester(self)
+        common_properties = ["image_token_index", "num_query_tokens", "image_text_hidden_size"]
+        self.config_tester = ConfigTester(
+            self, config_class=Blip2Config, has_text_modality=False, common_properties=common_properties
+        )
+
+    def test_config(self):
+        self.config_tester.run_common_tests()
 
     def test_for_conditional_generation(self):
         config_and_inputs = self.model_tester.prepare_config_and_inputs()

tests/models/instructblip/test_modeling_instructblip.py

@@ -158,7 +158,10 @@ class InstructBlipVisionModelTest(ModelTesterMixin, unittest.TestCase):
     def setUp(self):
         self.model_tester = InstructBlipVisionModelTester(self)
         self.config_tester = ConfigTester(
-            self, config_class=InstructBlipVisionConfig, has_text_modality=False, hidden_size=37
+            self,
+            config_class=InstructBlipConfig,
+            has_text_modality=False,
+            common_properties=["num_query_tokens", "image_token_index"],
         )
 
     def test_config(self):

tests/models/instructblipvideo/test_modeling_instructblipvideo.py

@@ -163,8 +163,9 @@ class InstructBlipVideoVisionModelTest(ModelTesterMixin, unittest.TestCase):
 
     def setUp(self):
         self.model_tester = InstructBlipVideoVisionModelTester(self)
+        common_properties = ["num_query_tokens", "video_token_index"]
         self.config_tester = ConfigTester(
-            self, config_class=InstructBlipVideoVisionConfig, has_text_modality=False, hidden_size=37
+            self, config_class=InstructBlipVideoConfig, has_text_modality=False, common_properties=common_properties
         )
 
     def test_config(self):

tests/test_modeling_common.py

@@ -2283,7 +2283,7 @@ class ModelTesterMixin:
 
     def test_tied_weights_keys(self):
         config, _ = self.model_tester.prepare_config_and_inputs_for_common()
-        config.tie_word_embeddings = True
+        config.get_text_config().tie_word_embeddings = True
         for model_class in self.all_model_classes:
             model_tied = model_class(config)