def test_push_to_hub_dynamic_processor(self):
CustomFeatureExtractor.register_for_auto_class()
CustomTokenizer.register_for_auto_class()
CustomProcessor.register_for_auto_class()
feature_extractor = CustomFeatureExtractor.from_pretrained(SAMPLE_PROCESSOR_CONFIG_DIR)
with tempfile.TemporaryDirectory() as tmp_dir:
vocab_file = os.path.join(tmp_dir, "vocab.txt")
with open(vocab_file, "w", encoding="utf-8") as vocab_writer:
vocab_writer.write("".join([x + "\n" for x in self.vocab_tokens]))
tokenizer = CustomTokenizer(vocab_file)
processor = CustomProcessor(feature_extractor, tokenizer)
with tempfile.TemporaryDirectory() as tmp_dir:
repo = Repository(tmp_dir, clone_from=f"{USER}/test-dynamic-processor", use_auth_token=self._token)
processor.save_pretrained(tmp_dir)
# This has added the proper auto_map field to the feature extractor config
self.assertDictEqual(
processor.feature_extractor.auto_map,
{
"AutoFeatureExtractor": "custom_feature_extraction.CustomFeatureExtractor",
"AutoProcessor": "custom_processing.CustomProcessor",
},
)
# This has added the proper auto_map field to the tokenizer config
with open(os.path.join(tmp_dir, "tokenizer_config.json")) as f:
tokenizer_config = json.load(f)
self.assertDictEqual(
tokenizer_config["auto_map"],
{
"AutoTokenizer": ["custom_tokenization.CustomTokenizer", None],
"AutoProcessor": "custom_processing.CustomProcessor",
},
)
# The code has been copied from fixtures
self.assertTrue(os.path.isfile(os.path.join(tmp_dir, "custom_feature_extraction.py")))
self.assertTrue(os.path.isfile(os.path.join(tmp_dir, "custom_tokenization.py")))
self.assertTrue(os.path.isfile(os.path.join(tmp_dir, "custom_processing.py")))
repo.push_to_hub()
new_processor = AutoProcessor.from_pretrained(f"{USER}/test-dynamic-processor", trust_remote_code=True)
# Can't make an isinstance check because the new_processor is from the CustomProcessor class of a dynamic module
self.assertEqual(new_processor.__class__.__name__, "CustomProcessor")
def test_new_processor_registration(self):
try:
AutoConfig.register("custom", CustomConfig)
AutoFeatureExtractor.register(CustomConfig, CustomFeatureExtractor)
AutoTokenizer.register(CustomConfig, slow_tokenizer_class=CustomTokenizer)
AutoProcessor.register(CustomConfig, CustomProcessor)
# Trying to register something existing in the Transformers library will raise an error
with self.assertRaises(ValueError):
AutoProcessor.register(Wav2Vec2Config, Wav2Vec2Processor)
# Now that the config is registered, it can be used as any other config with the auto-API
feature_extractor = CustomFeatureExtractor.from_pretrained(SAMPLE_PROCESSOR_CONFIG_DIR)
with tempfile.TemporaryDirectory() as tmp_dir:
vocab_file = os.path.join(tmp_dir, "vocab.txt")
with open(vocab_file, "w", encoding="utf-8") as vocab_writer:
vocab_writer.write("".join([x + "\n" for x in self.vocab_tokens]))
tokenizer = CustomTokenizer(vocab_file)
processor = CustomProcessor(feature_extractor, tokenizer)
with tempfile.TemporaryDirectory() as tmp_dir:
processor.save_pretrained(tmp_dir)
new_processor = AutoProcessor.from_pretrained(tmp_dir)
self.assertIsInstance(new_processor, CustomProcessor)
finally:
if "custom" in CONFIG_MAPPING._extra_content:
del CONFIG_MAPPING._extra_content["custom"]
if CustomConfig in FEATURE_EXTRACTOR_MAPPING._extra_content:
del FEATURE_EXTRACTOR_MAPPING._extra_content[CustomConfig]
if CustomConfig in TOKENIZER_MAPPING._extra_content:
del TOKENIZER_MAPPING._extra_content[CustomConfig]
if CustomConfig in PROCESSOR_MAPPING._extra_content:
del PROCESSOR_MAPPING._extra_content[CustomConfig]
def test_new_tokenizer_registration(self):
try:
AutoConfig.register("custom", CustomConfig)
AutoTokenizer.register(CustomConfig,
slow_tokenizer_class=CustomTokenizer)
# Trying to register something existing in the Transformers library will raise an error
with self.assertRaises(ValueError):
AutoTokenizer.register(BertConfig,
slow_tokenizer_class=BertTokenizer)
tokenizer = CustomTokenizer.from_pretrained(SMALL_MODEL_IDENTIFIER)
with tempfile.TemporaryDirectory() as tmp_dir:
tokenizer.save_pretrained(tmp_dir)
new_tokenizer = AutoTokenizer.from_pretrained(tmp_dir)
self.assertIsInstance(new_tokenizer, CustomTokenizer)
finally:
if "custom" in CONFIG_MAPPING._extra_content:
del CONFIG_MAPPING._extra_content["custom"]
if CustomConfig in TOKENIZER_MAPPING._extra_content:
del TOKENIZER_MAPPING._extra_content[CustomConfig]