QuALITY
General QA · 2021-12-16
QuALITY is a multiple-choice reading-comprehension benchmark over long English passages (stories and articles of roughly 2,000-8,000 words), with questions written and validated so that skimming the passage under time pressure is not enough to answer them. Scores report answer accuracy.
Top models (higher is better)
| Model | Score |
|---|---|
| Claude 1.3 | 84.1 |
| Claude Instant 1.1 | 80.5 |