Discriminative Accuracy on JudgeBench
77.3Knowledge AccuracyScalar RM
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Scalar RMBackbone=Skywork-RM-V22026.05 | 77.3 | 75.5 | 89.3 | 81 | 79.1 | 80.8 | |
| EVOLM2026.05 | 45.5 | 39.8 | 30.4 | 61.9 | 43.4 | 44.4 | |
| RLCER2026.05 | 43.5 | 40.8 | 48.2 | 64.3 | 46 | 49.2 | |
| RUBRIC-ARMVersion=12026.05 | 43.5 | 42.9 | 62.5 | 57.1 | 48 | 51.5 | |
| GPT-4.1Prompted=true2026.05 | 41.6 | 48 | 30.4 | 45.2 | 42 | 41.3 | |
| SequentialTraining=IQ2026.05 | 37.7 | 55.1 | 44.6 | 57.1 | 46 | 48.6 | |
| Qwen3-8BPrompted=true2026.05 | 34.4 | 38.8 | 30.4 | 42.9 | 36 | 36.6 | |
| SequentialTraining=RC2026.05 | 34.4 | 39.8 | 44.6 | 50 | 39.4 | 42.2 |