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.

Sign up for GitHub

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

Jump to bottom

ScatterMoE Gradient Norm Needs to be Properly Computed When Used With FSDP #109

Open
fabianlim opened this issue Nov 11, 2024 · 0 comments
Open

ScatterMoE Gradient Norm Needs to be Properly Computed When Used With FSDP #109

fabianlim opened this issue Nov 11, 2024 · 0 comments
Labels
bug Something isn't working help wanted Extra attention is needed

Comments

@fabianlim
Copy link
Contributor

When ScatterMoE is used together with FSDP, HF Accelerate will call FSDP's clip_grad_norm, see here, which does not know how to properly compute the grad norms for ScatterMoE shards.

We need to be able to hook in some logic to handle the grad norm properly.
@fabianlim fabianlim added bug Something isn't working help wanted Extra attention is needed labels Nov 11, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
bug Something isn't working help wanted Extra attention is needed
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
1 participant
@fabianlim