Reward Model Evaluation on Arena-Hard RU
92.69Best@8 ScoreQwen3-32B-RM
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Qwen3-32B-RMN=82025.12 | 92.69 | 70.48 | 22.21 | |
| Skywork-Reward-V2-Llama-3.1-8BN=82025.12 | 90.49 | 77.31 | 13.18 | |
| Skywork-Reward-Gemma-2-27BN=82025.12 | 89.05 | 74.35 | 14.7 | |
| Llama-3.1-Tulu-3-70B-SFT-RM-RB2N=82025.12 | 87.37 | 78.47 | 8.9 | |
| Llama-3.3-Nemotron-70B-Reward-MultilingualN=82025.12 | 85.93 | 84.91 | 1.02 |