PointButton1 on Safety-Gymnasium
69Normalized RewardCPQ
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CPQ2026.04 | 69 | 3.2 | |
| CPQcost thresholds=3, evaluation episodes=20, random seeds=32026.04 | 69 | 3.2 | |
| CDT+SASAlignment=SAS, Base Model=CDT2026.04 | 51 | 1.27 | |
| CDT + SASBackbone=CDT, Alignment strategy (SAS)=Lyapunov-Guided2026.04 | 51 | 1.27 | |
| CDT + SAScost thresholds=3, evaluation episodes=20, random seeds=32026.04 | 51 | 1.27 | |
| CDTBackbone=CDT2026.04 | 50 | 1.68 | |
| CDTcost thresholds=3, evaluation episodes=20, random seeds=32026.04 | 50 | 1.68 | |
| DT + SASBackbone=DT, Alignment strategy (SAS)=Lyapunov-Guided2026.04 | 49 | 1.38 | |
| DT + SAScost thresholds=3, evaluation episodes=20, random seeds=32026.04 | 49 | 1.38 | |
| BCQ-Lag2026.04 | 24 | 1.73 | |
| BCQ-Lagcost thresholds=3, evaluation episodes=20, random seeds=32026.04 | 24 | 1.73 | |
| BEAR-Lag2026.04 | 20 | 1.6 | |
| BEAR-Lagcost thresholds=3, evaluation episodes=20, random seeds=32026.04 | 20 | 1.6 | |
| COptiDICE2026.04 | 13 | 1.4 | |
| COptiDICEcost thresholds=3, evaluation episodes=20, random seeds=32026.04 | 13 | 1.4 | |
| BC-All2026.04 | 10 | 10.5 | |
| BC-Allcost thresholds=3, evaluation episodes=20, random seeds=32026.04 | 10 | 10.5 | |
| BC-Safe2026.04 | 6 | 0.52 | |
| BC-Safecost thresholds=3, evaluation episodes=20, random seeds=32026.04 | 6 | 0.52 | |
| FISOR2026.04 | 1 | 0.58 | |
| DCRL2026.04 | 1 | 0.48 |