Safe Reinforcement Learning on Safety Gymnasium Humanoid (C=25) benchmark leaderboard · SOTA2 Research

Safe Reinforcement Learning on Safety Gymnasium Humanoid (C=25)

6,496.85ReturnCSPO

Evaluation Results

Method	Links
CSPO2026.06		6,496.85	18
CPPOPID2026.06		6,417.05	26.4
PPO-Lag2026.06		6,414.56	25.49
IPO2026.06		6,270.25	30.4
CPO2026.06		6,213.15	18.3
APPO2026.06		6,092.95	25.3
EPO2026.06		5,745.65	13.1
P3O2026.06		5,684.22	22
C-TRPO2026.06		5,450.52	15.9
FOCOPS2026.06		5,295.88	26.7
PCPO2026.06		4,706.07	14.7
CUP2026.06		2,427.39	18.9