CLAMBER - Few-shot CoT - Clarification Helpfulness: leaderboard
Metric: Helpfulness (%). Source: arxiv.org. 5 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-3.5 Turbo 16K | 53.29 |
| 2 | Llama 2 13B Chat | 40.75 |
| 3 | Llama 2 70B Chat | 39.99 |
| 4 | vicuna-13B | 31.87 |
Interactive version: theaggregate.ai/benchmark?slug=clamber-few-shot-cot-clarification-helpfulness · How It Works · Data refreshed daily, snapshot 2026-09-19.