-
Notifications
You must be signed in to change notification settings - Fork 44
Description
Not really sure if this falls in the catagory of tensor dimension issue, but i'm keep getting this error no matter if max_seq_len is adjusted or default
Traceback (most recent call last):
File "/Users/zhuoruizhang/Desktop/csm/voice_clone.py", line 173, in
main()
File "/Users/zhuoruizhang/Desktop/csm/voice_clone.py", line 164, in main
generate_speech_with_context(
File "/Users/zhuoruizhang/Desktop/csm/voice_clone.py", line 141, in generate_speech_with_context
audio = generator.generate(
File "/Users/zhuoruizhang/Desktop/csm/.venv/lib/python3.10/site-packages/torch/utils/_contextlib.py", line 116, in decorate_context
return func(*args, **kwargs)
File "/Users/zhuoruizhang/Desktop/csm/generator.py", line 157, in generate
audio = self._audio_tokenizer.decode(torch.stack(samples).permute(1, 2, 0)).squeeze(0).squeeze(0)
RuntimeError: stack expects a non-empty TensorList