Datasets:
Update dataset card with paper link, code link, and metadata
#1
by nielsr HF Staff - opened
Hi! I'm Niels, part of the community science team at Hugging Face. I'm opening this PR to improve the documentation of the SupraBench dataset.
This PR updates the dataset card with:
- A link to the research paper on Hugging Face.
- A link to the official GitHub repository.
- Corrected license information (
cc-by-4.0). - Added
text-generationto thetask_categories. - Added a sample usage section demonstrating how to run the benchmark code.
mtybilly changed pull request status to closed
Thank you @nielsr for the improvements! We've incorporated your suggestions directly on main:
- ✅ Added YAML metadata (
license: cc-by-4.0,task_categories,tags,pretty_name) - ✅ Linked the paper (
arXiv:2606.13477, https://arxiv.org/abs/2606.13477) and the GitHub repository - ✅ Added a Sample Usage section
In the same pass we also restructured the four task datasets (bap, tbs, sid, hgd) so the prompting strategy is now encoded as splits (base / cot / fewshot) instead of a single mixed test split, which conflicts with this PR's data-files block. Closing in favor of the updated main. Thanks again!