← all datasets
◆ Evals & benchmarks
Google MBPP
CC-BY-4.0About 1,000 crowd-sourced entry-level Python problems with three tests each. Old and widely trained on — a smoke test, not a benchmark.
Summary
- Formats
parquet- License
- CC-BY-4.0
- Added
- 2026-09-10
Get it
From the project’s own instructions where it documents any; otherwise a plain clone.
-
Download from huggingface.co ↗
Start from the source page, or use the commands below.
-
$ pip install datasets
-
$ python -c "from datasets import load_dataset; print(load_dataset('mbpp', 'sanitized'))"The hand-verified 427-problem subset; drop 'sanitized' for all 974.