Constrained Reinforcement Learning on Safety Gymnasium PointCircle2 (eval)
41.2RewardCUP
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| CUPKnowledge Level=Oracle2025.04 | 41.2 | 31.8 | — | |
| TraCeSKnowledge Level=Zero Knowledge2025.04 | 41.1 | 19.8 | 5,465 | |
| TRPO-SauteKnowledge Level=Oracle2025.04 | 39.5 | 3.7 | — | |
| FOCOPSKnowledge Level=Oracle2025.04 | 39 | 12.9 | — |