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

better handling of tensorwise float8 recipe in configuration #901

Open
vkuzo opened this issue Feb 27, 2025 · 1 comment
Open

better handling of tensorwise float8 recipe in configuration #901

vkuzo opened this issue Feb 27, 2025 · 1 comment
Assignees
@vkuzo

Comments

@vkuzo
Copy link
Contributor

vkuzo commented Feb 27, 2025

Bug description

We need a follow-up on #808 . If --float8.recipe_name tensorwise is specified, we should handle the FSDP float8 all-gather, scale precompute, etc arguments properly instead of asserting that they aren't supported.

Versions

main branch
@vkuzo vkuzo self-assigned this Feb 27, 2025
@tianyu-l
Copy link
Contributor

Thanks for filing this issue.
That sounds cleaner to me! And we can have recipe_name default to “tensorwise”.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees

@vkuzo vkuzo

Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
2 participants
@vkuzo @tianyu-l