Personalized Reward Modeling on DSP benchmark leaderboard · SOTA2 Research

Personalized Reward Modeling on DSP

78.08AccuracyPAFO

Evaluation Results

Method	Links
PAFO2026.06		78.08	74.47	78.72	0.131	0.1696	0.0728
Reg2026.06		77.5	69.13	78.98	0.1469	0.2472	0.0796
P-RLHF2026.06		77.09	69.67	78.4	0.1506	0.2357	0.0823
Reweight2026.06		75.83	72.73	76.38	0.1373	0.1685	0.0758
BTL2026.06		71.67	31.47	78.76	0.2738	0.678	0.142
GPO2026.06		71.55	33.73	78.22	0.2582	0.6414	0.1332
VPL2026.06		70.24	30.91	77.18	0.2776	0.6622	0.144