Reward Model Controllability on Harmless-helpful
1Kendall's TauMOC
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| MOC2026.04 | 1 | 5 | |
| MORLHF2026.04 | 1 | 1.7 | |
| Rewarded Soups2026.04 | 0.96 | 5.5 | |
| RiC2026.04 | 0.91 | 3 |
| Method | Links | ||
|---|---|---|---|
| MOC2026.04 | 1 | 5 | |
| MORLHF2026.04 | 1 | 1.7 | |
| Rewarded Soups2026.04 | 0.96 | 5.5 | |
| RiC2026.04 | 0.91 | 3 |