< Back to IRCAM Forum

Training Rave Model on Macbook Pro M3 GPU

Hi, I’ve installed Rave from the GitHub and created a python environment where I’ve installed all of the dependencies. I installed the Pytorch “Nightly” versions of torch, torchaudio, torchvision

pip3 install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/cpu

I installed ffmpeg with conda, also.

The basic installation runs through the preprocessing stage, using my audio to create a dataset.

However, the training stage is where I’m getting the error that it won’t work on the Apple M3 GPU.

I tried a simple test with pytorch, following this youtube tutorial and provided python notebook, and PyTorch appeared to be using the M3GPU correctly.

Has anyone else had this problem or is there a fix?

Many thanks

Fergal

1 Like