Policy Optimization on RLHF benchmark leaderboard · SOTA2 Research

Policy Optimization on RLHF

8.3True ScoreORPO

Evaluation Results

Method	Links
ORPO2026.04		8.3	0.63	-1.84
ORPO2026.04		8.3	0.63	-1.84
Max-Min2026.04		5.38	0.84	-0.1
Max-Min2026.04		5.38	0.84	-0.1
Ensemble2026.04		2.31	1.26	-1.7