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

[SFT] fix neftune_noise_alpha in SFTTrainer #1841

Merged
merged 11 commits into from
Sep 19, 2024
Merged

Conversation

kashif
Copy link
Collaborator

@kashif kashif commented Jul 17, 2024

fixes #1837 by removing it from the SFTTrainer as the super-class has support for it via the TRL requirements for Transformer version

@kashif kashif requested a review from qgallouedec July 17, 2024 08:06
@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

@jjc10 jjc10 left a comment

Choose a reason for hiding this comment

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

Thanks for picking this up!

trl/trainer/sft_trainer.py Show resolved Hide resolved
trl/trainer/sft_trainer.py Show resolved Hide resolved
@qgallouedec
Copy link
Member

@kashif what is the status of this PR?

@kashif
Copy link
Collaborator Author

kashif commented Sep 17, 2024

@jjc10 can you kindly test with this fix?

Copy link
Member

@lewtun lewtun left a comment

Choose a reason for hiding this comment

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

Thanks for the fix! I would love to see a regression test that reproduces the original error and shows the changes in this PR pass - WDYT?

trl/trainer/sft_trainer.py Outdated Show resolved Hide resolved
@kashif kashif merged commit 9fb871f into huggingface:main Sep 19, 2024
9 checks passed
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

Successfully merging this pull request may close these issues.

Possible bug in SFTTrainer
5 participants