How to finetune musicgen?

[Frei-2]

I am trying to finetune musicgen following the docs provided in MUSICGEN.md, but I find the docs is hard to follow and encounter a lot of errors. Details of errors are listed here.

Specifically, when I am trying to run the finetune command, I use the example dataset provided in egs/example and the command I use is:

dora run solver=musicgen/musicgen_base_32khz model/lm/model_scale=small continue_from=/cache_dir/musicgen-small conditioner=text2music dset=audio/example

• I run the command in the root directory of this project, i.e., /audiocraft-main.
• /cache_dir/musicgen-small is the local path I place pretrained model I download from facebook/musicgen-small because our server cann't access hugging face.

Thanks in advance for any help with this particular issue and would appreciate any general tips for something else I might be doing wrong.