Skip to content

Adds More Generative tasks #694

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

Merged
merged 6 commits into from
May 16, 2025
Merged

Adds More Generative tasks #694

merged 6 commits into from
May 16, 2025

Conversation

hynky1999
Copy link
Collaborator

@hynky1999 hynky1999 commented Apr 29, 2025

This PR adds more generative tasks and fixes some old ones:

  1. COQA -> Is converted into template + we only use the first question unlike previously when all questions were used which doesn't make sense, because any but first question requires context of previous ones
  2. DROP -> The metric doesn't really make sense + got converted into tempalte
  3. Natural questions -> New task
  4. Squad v2 -> New task
  5. Jopardy -> New task
  6. SImpleQA -> Just added few-shot split

@HuggingFaceDocBuilderDev
Copy link
Collaborator

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.

suite=["lighteval"],
prompt_function=prompt.coqa,
hf_repo="coqa",
coqa_first_question = LightevalTaskConfig(
Copy link
Member

Choose a reason for hiding this comment

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

For all evals in this file, either don't remove the original version (so we keep lighteval + yours) or keep the same name and change the version to 1.

Copy link
Collaborator Author

Choose a reason for hiding this comment

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

The original version didn't make any sense; each consecutive question can depend on each other. So I think this should be really the "first"version. Same for drop, it was outright broken iirc

Copy link
Member

Choose a reason for hiding this comment

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

This is a matter of reproducibility/continuity in the eval suite: if people search for these tasks in the future they will have completely disappeared - if you just say your version is v1, it makes it easy for people to update and see what to replace things by

Copy link
Collaborator Author

@hynky1999 hynky1999 May 16, 2025

Choose a reason for hiding this comment

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

Fixed it. Can I merge @clefourrier ?

Copy link
Member

Choose a reason for hiding this comment

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

I must be blind but I still see the name as coqa_first_question not the original coqa_lighteval

Copy link
Member

Choose a reason for hiding this comment

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

maybe I'm not looking at the correct commit though, so if this is fixed feel free to merge

Copy link
Collaborator Author

Choose a reason for hiding this comment

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

That's just variable name....

Copy link
Member

Choose a reason for hiding this comment

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

yes XD minimal changes in PRs to avoid confusing the users, especially on evals they might be using

Copy link
Member

@clefourrier clefourrier left a comment

Choose a reason for hiding this comment

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

Apart from needed change, lgtm

Copy link
Member

@NathanHB NathanHB 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 modifs !

@clefourrier clefourrier merged commit c6d1231 into main May 16, 2025
5 checks passed
hynky1999 added a commit that referenced this pull request May 22, 2025
* add smolm generative tasks

* add jeopardy

* pretty 🥰

* consistent stop sequences

* add versions  + change names

---------

Co-authored-by: Hynek Kydlicek <kydlicek.hynek@huggingface.co>
Co-authored-by: Clémentine Fourrier <22726840+clefourrier@users.noreply.github.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

4 participants