List your company

CarCircle on Bullet-Safety-Gym (Normalized Metrics)

0.97Normalized RewardReference policy (πref)
0.15880.36940.580.7906Apr 29, 2026May 2, 2026May 6, 2026May 9, 2026May 13, 2026May 16, 2026May 20, 2026

Evaluation results

MethodLinks
Reference policy (πref)2026.050.974.58
BC2026.050.771.17
SafeDICE2026.050.620.91
PPL2026.050.591.73
PREFINE2026.050.541.23
FISOR2026.040.40.11
CDT+SASAlignment=SAS, Base Model=CDT2026.040.220.75
CPQ2026.050.190.33