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

[KTO/BCO Trainer] add bos_token_id only if it exists #2279

Merged
merged 1 commit into from
Oct 25, 2024
Merged

[KTO/BCO Trainer] add bos_token_id only if it exists #2279

merged 1 commit into from
Oct 25, 2024

Conversation

seanexp
Copy link
Contributor

@seanexp seanexp commented Oct 25, 2024

What does this PR do?

This PR prepends bos_token_id only if it exists. Previously, KTO/BCO Trainer raise RuntimeError when bos_token_id is None and this hinders training models without bos_token (such as Qwen series models)

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? 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.
  • 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.

@kashif @qgallouedec @claralp

@qgallouedec
Copy link
Member

Thanks for this. Indeed I realized it while working on #2209

@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.

]
batch[f"{kwargs['prefix']}prompt_attention_mask"] = [1] + batch[
f"{kwargs['prefix']}prompt_attention_mask"
]
Copy link
Contributor

Choose a reason for hiding this comment

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

just wondering if it should be:

if len(all_tokens["completion_input_ids"]) == 0 or bos_token_id != all_tokens["completion_input_ids"][0]:

before this

Copy link
Member

Choose a reason for hiding this comment

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

Actually I suspect there is an error here, because we shouldn't add bos to completion (expect for enc-dec). But in a near future, we will refactor as we did in #2209. Open to contributions btw.

I think we can keep it like this for the moment.

@qgallouedec qgallouedec merged commit 110d088 into huggingface:main Oct 25, 2024
4 of 9 checks passed
@seanexp seanexp deleted the fix/add-bos-token-only-if-exists branch November 2, 2024 14:28
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