I am trying to use a custom model with insanely fast whisper. This is the code that I have so far: ... def __init__(self): self.insanely_fast_whisper_model = PreTrainedModel( config=PretrainedConfig(name_or_path=WHISPER_MODEL_PATH), ) ... async def _transcribe_insanely_fast_whisper(self, file_content: bytes = None) -> str: pipe = pipeline( "automatic-speech-recognition", model=self.insanely_fast_whisper_model, # "openai/whisper-large-v3", device="cpu", # "cuda:0", # or mps for Mac devices model_kwargs={"attn_implementation": "flash_attention_2"} if is_flash_attn_2_available() else {"attn_implementation": "sdpa"}, ) outputs = pipe( file_content, chunk_length_s=30, batch_size=24, return_timestamps=True, ) return outputs I've tried many things but whenever I am running the code I am getting: File "/usr/local/lib/python3.12/site-packages/transformers/pipelines/__init__.py", line 1097, in pipeline return pipeline_class(model=model, framework=framework, task=task, **kwargs) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/usr/local/lib/python3.12/site-packages/transformers/pipelines/automatic_speech_recognition.py", line 210, in __init__ feature_extractor._processor_class AttributeError: 'NoneType' object has no attribute '_processor_class' I don't know a lot about models and transformers and how to use them. Tried reading the documentation but didn't find an answer. Additionally here is the directory structure that WHISPER_MODEL_PATH is pointing to, for what it's worth: . ├── dataset │ └── whisper-large-v3-ro-ct2 │ ├── config.json │ ├── preprocessor_config.json │ ├── tokenizer.json │ └── vocabulary.json ├── Dockerfile ├── readme.md └── setup_python.sh Please, let me know if you need any additional info. Continue reading...
