Remove duplicate values from pytorch model json (microsoft#600)

## Describe your changes
Pytorch model adds values from the hf model config into model attributes
to make them easily accessible. However, this makes the model json
messy/confusing with values in model_attributes that are already in
`model_config`. This PR removes information from model_attributes that
is already in the hf_config.

## Checklist before requesting a review
- [ ] Add unit tests for this change.
- [ ] Make sure all tests can pass.
- [ ] Update documents if necessary.
- [ ] Format your code by running `pre-commit run --all-files`
- [ ] Is this a user-facing change? If yes, give a description of this
change to be included in the release notes.

## (Optional) Issue link

olive/model/__init__.py

@@ -724,11 +724,16 @@ def to_json(self, check_object: bool = False):
                 "hf_config": self.hf_config,
             }
         )
-        # convert script_dir and model_script to string
-        # the original config has them as serialized ResourcePath
-        for resource_name in ["script_dir", "model_script"]:
-            if self.resource_paths[resource_name]:
-                config["config"][resource_name] = self.get_resource(resource_name)
+        # clean up redundant information in model_attributes
+        config["config"].pop("model_attributes", None)
+        # using a copy of self.model_attributes since config["config"]["model_attributes"] is already
+        # serialized and might not match self.model_attributes
+        model_attributes = deepcopy(self.model_attributes)
+        if model_attributes and self.hf_config:
+            for key, value in self.get_hf_model_config().to_dict().items():
+                if key in model_attributes and model_attributes[key] == value:
+                    del model_attributes[key]
+        config["config"]["model_attributes"] = model_attributes or None
         return serialize_to_json(config, check_object)