Reward Modeling on RewardBench and RM-Bench Aggregate 2
90.3Average ScoreSkywork
Evaluation Results
| Method | Links | |
|---|---|---|
| Skywork2026.06 | 90.3 | |
| SVR2026.06 | 87.2 | |
| OSS2026.06 | 87 | |
| Self-Rubrics OSSbaseline=self-generated rubrics, model=OSS2026.06 | 78.5 | |
| RM-R12026.06 | 69.6 | |
| Rubric-ARM2026.06 | 68.1 |