Helpful Assistant on HH-RLHF
9.08HV ScoreCAGE
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CAGERegion=Help.–Harm. edge (αhumor=0)2026.05 | 9.08 | 75.9 | |
| GenARMRegion=Help.–Harm. edge (αhumor=0)2026.05 | 7.27 | 67.8 | |
| CAGERegion=All (n=31)2026.05 | 7.26 | 72.3 | |
| GenARMRegion=All (n=31)2026.05 | 7.17 | 68.4 | |
| GenARMRegion=Interior2026.05 | 4.08 | 61.7 | |
| CAGERegion=Interior2026.05 | 3.59 | 63.6 | |
| CAGERegion=Help.–Humor edge (αharm=0)2026.05 | 2.26 | 86.5 | |
| GenARMRegion=Harm.–Humor edge (αhelp=0)2026.05 | 1.81 | 90.7 | |
| GenARMRegion=Help.–Humor edge (αharm=0)2026.05 | 1.6 | 67.8 | |
| CAGERegion=Harm.–Humor edge (αhelp=0)2026.05 | 1.53 | 81 |