search

lucidrains / audiolm-pytorch

Implementation of AudioLM, a SOTA Language Modeling Approach to Audio Generation out of Google Research, in Pytorch
MIT License
2.36k stars 255 forks source link

What is the idea environment to run this repo? #159

Closed sulkytejas closed 1 year ago

sulkytejas commented 1 year ago

I tried running it on my mac m1 pro, but MPS keeps crashing, giving me an error raise subprocess.CalledProcessError(returncode=process.returncode, cmd=cmd)

Then I tried on google collab, but then it kept failing on AttributeError: module 'fairseq' has no attribute 'checkpoint_utils' It kept failing on Fairseq quite frequently. I even cloned the repo and installed it using !python setup.py build_ext --inplace and !pip install --editable ./

I am curious about what would be the ideal environment to run the package, or am I missing anything in my setup?

Is it possible to run on an M1 mac without the accelerat support?