CarButton2 on Safety Gymnasium
0.37RewardCPQ
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CPQ2026.04 | 0.37 | 12.51 | |
| CPQcost thresholds=3, evaluation episodes=20, random seeds=32026.04 | 0.37 | 12.51 | |
| CDT + SASBackbone=CDT, Alignment strategy (SAS)=Lyapunov-Guided2026.04 | 0.3 | 1.11 | |
| CDT + SAScost thresholds=3, evaluation episodes=20, random seeds=32026.04 | 0.3 | 1.11 | |
| DT + SASBackbone=DT, Alignment strategy (SAS)=Lyapunov-Guided2026.04 | 0.14 | 0.84 | |
| DT + SAScost thresholds=3, evaluation episodes=20, random seeds=32026.04 | 0.14 | 0.84 | |
| CDTBackbone=CDT2026.04 | 0.13 | 1.58 | |
| CDTcost thresholds=3, evaluation episodes=20, random seeds=32026.04 | 0.13 | 1.58 | |
| DCRL2026.04 | 0.09 | 1.42 | |
| BCQ-Lag2026.04 | 0.06 | 2.13 | |
| BCQ-Lagcost thresholds=3, evaluation episodes=20, random seeds=32026.04 | 0.06 | 2.13 | |
| BC-Safe2026.04 | -0.01 | 0.63 | |
| BEAR-Lag2026.04 | -0.01 | 2.29 | |
| BC-Safecost thresholds=3, evaluation episodes=20, random seeds=32026.04 | -0.01 | 0.63 | |
| BEAR-Lagcost thresholds=3, evaluation episodes=20, random seeds=32026.04 | -0.01 | 2.29 | |
| COptiDICE2026.04 | -0.07 | 1.59 | |
| COptiDICEcost thresholds=3, evaluation episodes=20, random seeds=32026.04 | -0.07 | 1.59 | |
| BC-All2026.04 | -0.13 | 1.24 | |
| BC-Allcost thresholds=3, evaluation episodes=20, random seeds=32026.04 | -0.13 | 1.24 |