The version of torch

[yutokondo101010]

Thanks for your great works. Would you describe the version of torch and torchaudio?
The error "magic_number = pickle_module.load(f, **pickle_load_args)
_pickle.UnpicklingError: invalid load key, 'v'." occured in my practice.

[enamigi]

Hi. I had this problem earlier today. I installed with python 3.9 because Triton can not be installed with pip with later version.
The torch version is 2.4.1 and torch audio 0.19.1

I installed in WSL in windows. And i got this message. Seems to be an error of not finding the data it expects.

It seems it's an error because there are placeholders in ./dreamvoice/ckpts for all pt files. I had installed previously on Windows and copied the files to this location and could execute.

I had files before about 130 Bytes and the real checkpoints are 300 MB, 452 MB, and 104.9 MB. After I copied all files into the ckpts directories, the program ran smoothly. I had to pip install some other libraries that were missing: webrtcvad, xformers, and SentencePiece

[HuZhetao]

Hi. I had this problem earlier today. I installed with python 3.9 because Triton can not be installed with pip with later version. The torch version is 2.4.1 and torch audio 0.19.1

I installed in WSL in windows. And i got this message. Seems to be an error of not finding the data it expects.

It seems it's an error because there are placeholders in ./dreamvoice/ckpts for all pt files. I had installed previously on Windows and copied the files to this location and could execute.

I had files before about 130 Bytes and the real checkpoints are 300 MB, 452 MB, and 104.9 MB. After I copied all files into the ckpts directories, the program ran smoothly. I had to pip install some other libraries that were missing: webrtcvad, xformers, and SentencePiece

I have meet just the same questions， it would be great if i seen your comment before i try to run the program