← all datasets
◆ Evals & benchmarks
SWE-bench Verified
license: see sourceA 500-task subset of the SWE-bench test set, each issue human-validated as solvable and fairly tested — the version most coding-agent results are reported on.
Summary
- Formats
parquet- License
- license: see source
- Added
- 2026-09-10
Get it
From the project’s own instructions where it documents any; otherwise a plain clone.
-
Download from huggingface.co ↗
No single download: pick the files you need from the source page.