PointCircle2 on Safety Gymnasium (val)
41.7RewardPPO-Lagrangian
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| PPO-LagrangianKnowledge Level=Oracle2025.04 | 41.7 | 29.1 | — | |
| CPPO-PIDKnowledge Level=Oracle2025.04 | 41.3 | 22.8 | — | |
| TraCeSKnowledge Level=Zero Knowledge2025.04 | 41.1 | 19.8 | 5,465 | |
| C-T BaselineKnowledge Level=Zero Knowledge2025.04 | 40.2 | 22.7 | 12,000 | |
| RLSFKnowledge Level=Partial Knowledge2025.04 | 39.9 | 2.6 | 3,422 |