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

Bug in use_mem_eff_path with ngroups>1? #401

Open
patronum08 opened this issue Jun 18, 2024 · 0 comments
Open

Bug in use_mem_eff_path with ngroups>1? #401

patronum08 opened this issue Jun 18, 2024 · 0 comments

Comments

@patronum08
Copy link

It seems that there is a bug when using the use_mem_eff_path feature, when ngroups is greater than 1. The loss curve initially decreases but then stabilizes around a constant value and fails to converge further. We are training with 8 H800 GPUs, setting ngroups to 8, N to 128, and not using tensor parallelism.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant