Don't save `processor_config.json` if a processor has no extra attribute (#28584)

* not save if empty

* fix

* fix

* fix

* fix

* fix

---------
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>

src/transformers/processing_utils.py

@@ -234,6 +234,9 @@ class ProcessorMixin(PushToHubMixin):
         # If we save using the predefined names, we can load using `from_pretrained`
         output_processor_file = os.path.join(save_directory, PROCESSOR_NAME)
 
+        # For now, let's not save to `processor_config.json` if the processor doesn't have extra attributes and
+        # `auto_map` is not specified.
+        if set(self.to_dict().keys()) != {"processor_class"}:
             self.to_json_file(output_processor_file)
             logger.info(f"processor saved in {output_processor_file}")
 
@@ -246,6 +249,8 @@ class ProcessorMixin(PushToHubMixin):
                 token=kwargs.get("token"),
             )
 
+        if set(self.to_dict().keys()) == {"processor_class"}:
+            return []
         return [output_processor_file]
 
     @classmethod

tests/models/auto/test_processor_auto.py

@@ -101,6 +101,12 @@ class AutoFeatureExtractorTest(unittest.TestCase):
             # save in new folder
             processor.save_pretrained(tmpdirname)
 
+            if not os.path.isfile(os.path.join(tmpdirname, PROCESSOR_NAME)):
+                # create one manually in order to perform this test's objective
+                config_dict = {"processor_class": "Wav2Vec2Processor"}
+                with open(os.path.join(tmpdirname, PROCESSOR_NAME), "w") as fp:
+                    json.dump(config_dict, fp)
+
             # drop `processor_class` in tokenizer config
             with open(os.path.join(tmpdirname, TOKENIZER_CONFIG_FILE), "r") as f:
                 config_dict = json.load(f)
@@ -123,6 +129,7 @@ class AutoFeatureExtractorTest(unittest.TestCase):
             # save in new folder
             processor.save_pretrained(tmpdirname)
 
+            if os.path.isfile(os.path.join(tmpdirname, PROCESSOR_NAME)):
                 # drop `processor_class` in processor
                 with open(os.path.join(tmpdirname, PROCESSOR_NAME), "r") as f:
                     config_dict = json.load(f)
@@ -153,6 +160,7 @@ class AutoFeatureExtractorTest(unittest.TestCase):
             # save in new folder
             processor.save_pretrained(tmpdirname)
 
+            if os.path.isfile(os.path.join(tmpdirname, PROCESSOR_NAME)):
                 # drop `processor_class` in processor
                 with open(os.path.join(tmpdirname, PROCESSOR_NAME), "r") as f:
                     config_dict = json.load(f)

tests/test_processing_common.py

@@ -75,8 +75,9 @@ class ProcessorTesterMixin:
         processor_first = self.get_processor()
 
         with tempfile.TemporaryDirectory() as tmpdirname:
-            saved_file = processor_first.save_pretrained(tmpdirname)[0]
-            check_json_file_has_correct_format(saved_file)
+            saved_files = processor_first.save_pretrained(tmpdirname)
+            if len(saved_files) > 0:
+                check_json_file_has_correct_format(saved_files[0])
                 processor_second = self.processor_class.from_pretrained(tmpdirname)
 
                 self.assertEqual(processor_second.to_dict(), processor_first.to_dict())