I was using faster-whisper and the downloaded model utilized relative symlink (to avoid duplications I suppose) but the webui (or the ctranslate2) doesn't like it:
Traceback (most recent call last):
File "C:\AI\LLM\Subs-AI\venv\Lib\site-packages\streamlit\runtime\scriptrunner\script_runner.py", line 565, in _run_script
exec(code, module.__dict__)
File "C:\AI\LLM\Subs-AI\venv\Lib\site-packages\subsai\webui.py", line 545, in <module>
run()
File "C:\AI\LLM\Subs-AI\venv\Lib\site-packages\subsai\webui.py", line 538, in run
webui()
File "C:\AI\LLM\Subs-AI\venv\Lib\site-packages\subsai\webui.py", line 318, in webui
subs = _transcribe(file_path, stt_model_name, model_config)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\AI\LLM\Subs-AI\venv\Lib\site-packages\streamlit\runtime\caching\cache_utils.py", line 194, in wrapper
return cached_func(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\AI\LLM\Subs-AI\venv\Lib\site-packages\streamlit\runtime\caching\cache_utils.py", line 223, in __call__
return self._get_or_create_cached_value(args, kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\AI\LLM\Subs-AI\venv\Lib\site-packages\streamlit\runtime\caching\cache_utils.py", line 248, in _get_or_create_cached_value
return self._handle_cache_miss(cache, value_key, func_args, func_kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\AI\LLM\Subs-AI\venv\Lib\site-packages\streamlit\runtime\caching\cache_utils.py", line 302, in _handle_cache_miss
computed_value = self._info.func(*func_args, **func_kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\AI\LLM\Subs-AI\venv\Lib\site-packages\subsai\webui.py", line 189, in _transcribe
model = subs_ai.create_model(model_name, model_config=model_config)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\AI\LLM\Subs-AI\venv\Lib\site-packages\subsai\main.py", line 96, in create_model
return AVAILABLE_MODELS[model_name]['class'](model_config)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\AI\LLM\Subs-AI\venv\Lib\site-packages\subsai\models\faster_whisper_model.py", line 240, in __init__
self.model = WhisperModel(model_size_or_path=self._model_size_or_path,
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\AI\LLM\Subs-AI\venv\Lib\site-packages\faster_whisper\transcribe.py", line 120, in __init__
self.model = ctranslate2.models.Whisper(
^^^^^^^^^^^^^^^^^^^^^^^^^^^
RuntimeError: Unable to open file 'model.bin' in model 'C:\Users\Victor\.cache\huggingface\hub\models--guillaumekln--faster-whisper-large-v2\snapshots\f541c54c566e32dc1fbce16f98df699208837e8b'
models--guillaumekln--faster-whisper-large-v2\snapshots\f541c54c566e32dc1fbce16f98df699208837e8b is a folder that contains model files, those files are symlinks to the actual files at models--guillaumekln--faster-whisper-large-v2\blobs folder.
If I copy those files over and rename them, this error goes away. And the rest works flawlessly.
I was using faster-whisper and the downloaded model utilized relative symlink (to avoid duplications I suppose) but the webui (or the ctranslate2) doesn't like it:
models--guillaumekln--faster-whisper-large-v2\snapshots\f541c54c566e32dc1fbce16f98df699208837e8bis a folder that contains model files, those files are symlinks to the actual files atmodels--guillaumekln--faster-whisper-large-v2\blobsfolder.If I copy those files over and rename them, this error goes away. And the rest works flawlessly.