Safe Reinforcement Learning on SMARTS Overtaking 512K steps
1,544Unsafe EventsCAL
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| CAL2026.03 | 1,544 | — | — | — | — | |
| SACLag2026.03 | 1,570 | — | — | — | — | |
| COX-Q2026.03 | 1,665 | — | — | — | — | |
| TQC-ORAC2026.03 | 3,215 | — | — | — | — | |
| CPPOPID2026.03 | 3,697 | — | — | — | — | |
| CALsteps=512K, stochastic runs=20002026.03 | — | 186 | 7 | 0 | 1 | |
| COX-Qsteps=512K, stochastic runs=20002026.03 | — | 99 | 4 | 0 | 0 | |
| CPPOPIDsteps=512K, stochastic runs=20002026.03 | — | 331 | 96 | 3 | 0 | |
| SACLagsteps=512K, stochastic runs=20002026.03 | — | 194 | 2 | 0 | 2 | |
| TQC-ORACsteps=512K, stochastic runs=20002026.03 | — | 97 | 3 | 0 | 887 |