We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Will there be MPS support for the Gemma models? It would enable access to a larger community.
The text was updated successfully, but these errors were encountered:
Took a look, a few things.
Linux / mps support looks to be in progress still pytorch/pytorch#81224 so running in a container isn't ready yet.
MPS has some limitations around complex tensors atm. Since gemma uses RoPE, it uses complex tensors and errors out if you run it locally.
https://github.com/pytorch/pytorch/pull/116764/files#diff-fe061f10677283971d77576718d3a04a00b2225d72c043fd59222a882b92c64bR654
gemma_pytorch/gemma/model.py
Line 426 in 01062c9
Running locally with python scripts/run.py --ckpt gemma-2b-it.ckpt --variant 2b --device mps
python scripts/run.py --ckpt gemma-2b-it.ckpt --variant 2b --device mps
Sorry, something went wrong.
pytorch 2.3 has bf16 and complex tensor support, and dockerised containers now work @lamroger
No branches or pull requests
Will there be MPS support for the Gemma models? It would enable access to a larger community.
The text was updated successfully, but these errors were encountered: