Open Japanese LLM - Wiki Coreference SET F1: leaderboard
Metric: Score (%). Source: huggingface.co. 418 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Llama 3.1 Nemotron 70B Instruct | 9.84 |
| 2 | Qwen 2.5 14B Instruct | 9.69 |
| 3 | ABEJA-Qwen2.5-32B-Japanese-v1.0 | 8.53 |
| 4 | Saka-14B | 8.22 |
| 5 | Llama 3.1 70B Instruct | 8.08 |
| 6 | tempmotacilla-cinerea-0308 | 7.04 |
| 7 | lambda-qwen2.5-14B-dpo-test | 6.85 |
| 8 | Llama-3.3-Swallow-70B-Instruct-v0.4 | 6.14 |
| 9 | Awqward2.5-32B-Instruct | 6.09 |
| 10 | Qwen 2.5 Coder 14B Instruct | 5.92 |
| 11 | Llama 3.1 70B | 5.65 |
| 12 | Qwen 2.5 32B Instruct | 5.43 |
| 13 | Qwen2.5-32B-Instruct-CFT | 5.34 |
| 14 | Qwen 2.5 Coder 32B Instruct | 5.29 |
| 15 | oxyge1-33B | 5.27 |
Interactive version: theaggregate.ai/benchmark?slug=open-japanese-llm-wiki-coreference-set-f1 · How It Works · Data refreshed daily, snapshot 2026-09-05.