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

Fix FSDP no longer working #35212

Merged
merged 1 commit into from
Dec 13, 2024
Merged

Fix FSDP no longer working #35212

merged 1 commit into from
Dec 13, 2024

Conversation

muellerzr
Copy link
Contributor

What does this PR do?

#34140 introduced a breaking change that users are reporting has many downstream issues:

This PR reverts the breaking logic, while also maintaining the original test the user intended, in such a way that doesn't break downstream code. I'd highly recommend a patch with this fix @ArthurZucker

Fixes huggingface/peft#2205
Fixes axolotl-ai-cloud/axolotl#2149

Before submitting

  • This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
  • Did you read the contributor guideline,
    Pull Request section?
  • Was this discussed/approved via a Github issue or the forum? Please add a link
    to it if that's the case.
  • Did you make sure to update the documentation with your changes? Here are the
    documentation guidelines, and
    here are tips on formatting docstrings.
  • Did you write any new necessary tests?

Who can review?

Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.

@ArthurZucker @winglian @SunMarc @BenjaminBossan

@HuggingFaceDocBuilderDev

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.

Copy link
Member

@SunMarc SunMarc left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM !

Copy link
Member

@BenjaminBossan BenjaminBossan left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Testing this change locally, I can confirm that it enables my testing example to run successfully. There is an exception, namely when flash attention is being used, for some reason (perhaps it's an issue with my setup). But even so, this is an improvement and can be merged from my point of view.

@ArthurZucker ArthurZucker merged commit 7237b3e into main Dec 13, 2024
22 of 25 checks passed
@ArthurZucker ArthurZucker deleted the muellerzr-revert-fp8 branch December 13, 2024 18:20
@ArthurZucker
Copy link
Collaborator

Failures are unrelated!

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

Successfully merging this pull request may close these issues.

Error During Model Saving QLORA + FSDP KeyError: Parameter containing
5 participants