Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

CarRun on Bullet-Safety-Gym (Normalized Metrics)

99Normalized Reward

Reference policy (πref)

-22.688.9140.572.09Apr 29, 2026May 2, 2026May 6, 2026May 9, 2026May 13, 2026May 16, 2026May 20, 2026
Updated 13d ago

Evaluation Results

MethodLinks
991.89
2026.05
930.38
2026.05
860.11
2026.05
810.28
2026.04
7314
2026.04
7239
2026.05
710.39
2026.05
-180.06