[bnb] Fix bnb config json serialization (#24137)

* fix bnb config json serialization

* forward contrib credits from discussions

---------

Co-authored-by: Andrechang <[email protected]>

Author: 2 people

src/transformers/configuration_utils.py

@@ -784,6 +784,13 @@ def to_diff_dict(self) -> Dict[str, Any]:
             ):
                 serializable_config_dict[key] = value
 
+        if hasattr(self, "quantization_config"):
+            serializable_config_dict["quantization_config"] = (
+                self.quantization_config.to_dict()
+                if not isinstance(self.quantization_config, dict)
+                else self.quantization_config
+            )
+
         self.dict_torch_dtype_to_str(serializable_config_dict)
 
         return serializable_config_dict

tests/bitsandbytes/test_4bit.py

@@ -111,6 +111,19 @@ def tearDown(self):
         gc.collect()
         torch.cuda.empty_cache()
 
+    def test_quantization_config_json_serialization(self):
+        r"""
+        A simple test to check if the quantization config is correctly serialized and deserialized
+        """
+        config = self.model_4bit.config
+
+        self.assertTrue(hasattr(config, "quantization_config"))
+
+        _ = config.to_dict()
+        _ = config.to_diff_dict()
+
+        _ = config.to_json_string()
+
     def test_memory_footprint(self):
         r"""
         A simple test to check if the model conversion has been done correctly by checking on the

tests/bitsandbytes/test_mixed_int8.py

@@ -118,6 +118,19 @@ def tearDown(self):
         gc.collect()
         torch.cuda.empty_cache()
 
+    def test_quantization_config_json_serialization(self):
+        r"""
+        A simple test to check if the quantization config is correctly serialized and deserialized
+        """
+        config = self.model_8bit.config
+
+        self.assertTrue(hasattr(config, "quantization_config"))
+
+        _ = config.to_dict()
+        _ = config.to_diff_dict()
+
+        _ = config.to_json_string()
+
     def test_memory_footprint(self):
         r"""
         A simple test to check if the model conversion has been done correctly by checking on the