HELM CLEVA - Cleva Coreference Resolution: leaderboard
The "Cleva Coreference Resolution" subtask of HELM CLEVA. Stanford HELM's run of CLEVA, the Chinese Language Models Evaluation Platform: Chinese-language scenarios spanning knowledge, reasoning, language understanding and generation, reported as HELM mean win rate across the group's scenarios.
Metric: EM. Source: crfm.stanford.edu. Saturation forecast: Estimated already saturated. 4 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4 (0613) | 91.33 | #496 |
| 2 | LLaMA-65B | 76.67 | #1295 |
| 3 | GPT-3.5 Turbo (0613) | 68.67 | #1123 |
| 4 | vicuna-13B-v1.3 | 65.33 | #1447 |
No result here: #4 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=helm-cleva-cleva-coreference-resolution · How It Works · Data refreshed daily, snapshot 2026-10-24.