[PretrainedConfig] Fix save pretrained config for edge case (#7943)

* fix config save

* add test

* add config class variable and another test

* line break

* fix fsmt and typo

* god am I making many errors today :-/

* Update src/transformers/configuration_utils.py
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

src/transformers/configuration_encoder_decoder.py

@@ -70,6 +70,7 @@ class EncoderDecoderConfig(PretrainedConfig):
         >>> model = EncoderDecoderModel.from_pretrained('my-model', config=encoder_decoder_config)
     """
     model_type = "encoder_decoder"
+    is_composition = True
 
     def __init__(self, **kwargs):
         super().__init__(**kwargs)

src/transformers/configuration_fsmt.py

@@ -126,9 +126,9 @@ class FSMTConfig(PretrainedConfig):
     # update the defaults from config file
     def __init__(
         self,
-        langs,
-        src_vocab_size,
-        tgt_vocab_size,
+        langs=["en", "de"],
+        src_vocab_size=42024,
+        tgt_vocab_size=42024,
         activation_function="relu",
         d_model=1024,
         max_length=200,

src/transformers/configuration_rag.py

@@ -77,6 +77,7 @@ RAG_CONFIG_DOC = r"""
 @add_start_docstrings(RAG_CONFIG_DOC)
 class RagConfig(PretrainedConfig):
     model_type = "rag"
+    is_composition = True
 
     def __init__(
         self,

src/transformers/configuration_utils.py

@@ -41,6 +41,10 @@ class PretrainedConfig(object):
     Class attributes (overridden by derived classes)
         - **model_type** (:obj:`str`): An identifier for the model type, serialized into the JSON file, and used to
           recreate the correct object in :class:`~transformers.AutoConfig`.
+        - **is_composition** (:obj:`bool`): Whether the config class is composed of multiple
+          sub-configs. In this case the config has to be initialized from two or more configs of
+          type :class:`~transformers.PretrainedConfig` like: :class:`~transformers.EncoderDecoderConfig` or
+          :class:`~RagConfig`.
 
     Args:
         name_or_path (:obj:`str`, `optional`, defaults to :obj:`""`):
@@ -145,6 +149,7 @@ class PretrainedConfig(object):
           use BFloat16 scalars (only used by some TensorFlow models).
     """
     model_type: str = ""
+    is_composition: bool = False
 
     def __init__(self, **kwargs):
         # Attributes with defaults
@@ -476,11 +481,18 @@ class PretrainedConfig(object):
         # get the default config dict
         default_config_dict = PretrainedConfig().to_dict()
 
+        # get class specific config dict
+        class_config_dict = self.__class__().to_dict() if not self.is_composition else {}
+
         serializable_config_dict = {}
 
         # only serialize values that differ from the default config
         for key, value in config_dict.items():
-            if key not in default_config_dict or value != default_config_dict[key]:
+            if (
+                key not in default_config_dict
+                or value != default_config_dict[key]
+                or (key in class_config_dict and value != class_config_dict[key])
+            ):
                 serializable_config_dict[key] = value
 
         return serializable_config_dict

tests/test_configuration_common.py

@@ -66,9 +66,16 @@ class ConfigTester(object):
         self.parent.assertEqual(len(config.id2label), 3)
         self.parent.assertEqual(len(config.label2id), 3)
 
+    def check_config_can_be_init_without_params(self):
+        if self.config_class.is_composition:
+            return
+        config = self.config_class()
+        self.parent.assertIsNotNone(config)
+
     def run_common_tests(self):
         self.create_and_test_config_common_properties()
         self.create_and_test_config_to_json_string()
         self.create_and_test_config_to_json_file()
         self.create_and_test_config_from_and_save_pretrained()
         self.create_and_test_config_with_num_labels()
+        self.check_config_can_be_init_without_params()

tests/test_modeling_prophetnet.py

@@ -901,6 +901,15 @@ class ProphetNetModelTest(ModelTesterMixin, unittest.TestCase):
         config_and_inputs = self.model_tester.prepare_config_and_inputs()
         self.model_tester.check_model_with_attn_mask(*config_and_inputs)
 
+    def test_config_save(self):
+        config = self.model_tester.prepare_config_and_inputs()[0]
+        config.add_cross_attention = False
+        with tempfile.TemporaryDirectory() as tmp_dirname:
+            config.save_pretrained(tmp_dirname)
+            config = ProphetNetConfig.from_pretrained(tmp_dirname)
+
+        self.assertFalse(config.add_cross_attention)
+
     @unittest.skipIf(torch_device == "cpu", "Cant do half precision")
     def test_fp16_forward(self):
         config_and_inputs = self.model_tester.prepare_config_and_inputs()