Personalized Reward Modeling on Personal-LLM
88.65AccuracyPAFO
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| PAFO2026.06 | 88.65 | 75 | 91.06 | 0.133 | 0.4193 | 0.0703 | |
| Reweight2026.06 | 87.6 | 73.33 | 90.12 | 0.1357 | 0.4672 | 0.0743 | |
| Reg2026.06 | 87.25 | 71.33 | 90.06 | 0.143 | 0.4593 | 0.0776 | |
| P-RLHF2026.06 | 86.45 | 70.33 | 89.29 | 0.1459 | 0.5175 | 0.078 | |
| BTL2026.06 | 83.2 | 56.67 | 87.88 | 0.1924 | 0.7707 | 0.1021 | |
| VPL2026.06 | 81.25 | 50.33 | 86.71 | 0.2166 | 0.7429 | 0.1172 | |
| GPO2026.06 | 80.6 | 34 | 88.82 | 0.2862 | 0.8009 | 0.1426 |