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

[DPO] DPOConfig class #1554

Merged
merged 20 commits into from
Apr 23, 2024
Merged

[DPO] DPOConfig class #1554

merged 20 commits into from
Apr 23, 2024

Conversation

kashif
Copy link
Collaborator

@kashif kashif commented Apr 18, 2024

Add a DPOConfig training argument dataclass together with deprecation warnings to the DPOTrainer

@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.

@kashif kashif requested review from younesbelkada and lvwerra April 18, 2024 17:47
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 refactor and new config @kashif ! Overall LGTM with some nits about moving all args from the trainer init to the config in the script example

As with the SFT trainer, it would be good to have a small integration test that checks we have the same trainer if it is initialised in both ways (via args and via init) to ensure backwards compat

Copy link
Contributor

@younesbelkada younesbelkada left a comment

Choose a reason for hiding this comment

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

Great work thanks @kashif ! LGTM once @lewtun comments are addressed !

@kashif kashif changed the title initial DPOConfig [DPO] DPOConfig class Apr 22, 2024
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 iterating @kashif ! I left one turbo nit and then this can be merged after addressing it :)

@kashif kashif merged commit 24fd8dd into huggingface:main Apr 23, 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.

4 participants