Skip to content
New issue

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

Allow on_host/2 to execute certain operations on host #1031

Open
josevalim opened this issue Dec 17, 2022 · 2 comments
Open

Allow on_host/2 to execute certain operations on host #1031

josevalim opened this issue Dec 17, 2022 · 2 comments
Labels
area:defn Applies to defn area:exla Applies to EXLA

Comments

@josevalim
Copy link
Collaborator

No description provided.

@josevalim
Copy link
Collaborator Author

For CPU offloading during training: https://huggingface.co/docs/transformers/main_classes/deepspeed

@josevalim josevalim added area:exla Applies to EXLA area:defn Applies to defn labels Jun 20, 2023
@josevalim
Copy link
Collaborator Author

Btw, we would implement this with infeed, but infeed is not supported in neither iree or cuda pjrt plugins at the moment. There may be a chance this needs to be implemented in a layer above (such as Axon).

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
area:defn Applies to defn area:exla Applies to EXLA
Projects
None yet
Development

No branches or pull requests

1 participant