GPQA is a multiple-choice, Q&A dataset of very hard questions written and validated by experts in biology, physics, and chemistry.
Dataset Card
By David Rein, published under cc-by-4.0, revision 633f5ee89ab8.
Dataset Card for GPQA
GPQA is a multiple-choice, Q&A dataset of very hard questions written and validated by experts in biology, physics, and chemistry. When attempting questions out of their own domain (e.g., a physicist answers a chemistry question), these experts get only 34% accuracy, despite spending >30m with full access to Google.
We request that you do not reveal examples from this dataset in plain text or images online, to reduce the risk of leakage into foundation model training corpora.
Dataset Details
Dataset Description
Details
- Repository
- Idavidrein/gpqa
- Publisher
- David Rein
- Task category
- Question answering
- Tags
- open-domain-qa, open-book-qa, multiple-choice-qa
- Size category
- n<1K
- Languages
- en
- Revision
- 633f5ee89ab8ad4522a9f850766b73f62147ffdd
- Last updated
- 2026-03-05
Files
8 files, 8.7 MB in total.
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| gpqa_diamond.csv | Data | 1.4 MB | — |
| gpqa_experts.csv | Data | 26.6 KB | — |
| gpqa_extended.csv | Data | 4.1 MB | — |
| gpqa_main.csv | Data | 3.2 MB | — |
| README.md | Documentation | 3.3 KB | — |
| license.txt | Documentation | 241 B | — |
| eval.yaml | Other | 1.6 KB | — |
| .gitattributes | Repository | 2.3 KB | — |
License and Download
- License
- cc-by-4.0
- Access
- Access requested at publisher
David Rein grants access through its official repository on Hugging Face. Read the license.