CarButton1 on Safety Gymnasium
0.42RewardCPQ
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CPQ2026.04 | 0.42 | 9.66 | |
| CPQcost thresholds=3, evaluation episodes=20, random seeds=32026.04 | 0.42 | 9.66 | |
| CDT + SASBackbone=CDT, Alignment strategy (SAS)=Lyapunov-Guided2026.04 | 0.27 | 0.98 | |
| CDT + SAScost thresholds=3, evaluation episodes=20, random seeds=32026.04 | 0.27 | 0.98 | |
| CDTBackbone=CDT2026.04 | 0.21 | 1.6 | |
| CDTcost thresholds=3, evaluation episodes=20, random seeds=32026.04 | 0.21 | 1.6 | |
| BEAR-Lag2026.04 | 0.18 | 2.72 | |
| BEAR-Lagcost thresholds=3, evaluation episodes=20, random seeds=32026.04 | 0.18 | 2.72 | |
| DT + SASBackbone=DT, Alignment strategy (SAS)=Lyapunov-Guided2026.04 | 0.17 | 1.08 | |
| DT + SAScost thresholds=3, evaluation episodes=20, random seeds=32026.04 | 0.17 | 1.08 | |
| DCRL2026.04 | 0.12 | 0.95 | |
| BC-Safe2026.04 | 0.07 | 0.85 | |
| BC-Safecost thresholds=3, evaluation episodes=20, random seeds=32026.04 | 0.07 | 0.85 | |
| BCQ-Lag2026.04 | 0.04 | 1.63 | |
| BCQ-Lagcost thresholds=3, evaluation episodes=20, random seeds=32026.04 | 0.04 | 1.63 | |
| BC-All2026.04 | 0.03 | 1.38 | |
| BC-Allcost thresholds=3, evaluation episodes=20, random seeds=32026.04 | 0.03 | 1.38 | |
| COptiDICE2026.04 | -0.08 | 1.68 | |
| COptiDICEcost thresholds=3, evaluation episodes=20, random seeds=32026.04 | -0.08 | 1.68 |