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

Reward rollout alignment on 10-task benchmark T1: Folding Shorts

0.889Rollout ρ

SARM2

-0.266440.033530.33350.63347Jun 9, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.06
0.889
2026.06
0.833
2026.06
0.667
2026.06
0.611
2026.06
0.556
2026.06
0.222
2026.06
0.167
2026.06
-0.222