Week2 Lab on Apple Silicon


I downloaded the Week2 Lab and I have been trying tu run it on a local version of Jupiter Notebook on an Apple MacBook Pro with a M1 Max processor.

When I try to train the model with trainer.train(), I get the following error message:

TypeError: Trying to convert BFloat16 to the MPS backend but it does not have support for that dtype.

I have tried removing the argument torch_dtype=torch.bfloat16 when I load the model, tryin to load it to the CPU using the .to(‘cpu’) method but always get some king of error message.

Have anyone been able to make this notebook run on Apple Silicon?



This was solved by installing the library version specified in the notebook. Unfortunately, this older version of the libraries doesn’t support Apple Silicon GPUs.

1 Like