List your company
ResearchFormerly WizWand

Harmful score evaluation on Sorrybench

12.95Harmful ScoreSurgery
12.231617.080821.9326.7792Feb 5, 2026
Updated 8mo ago

Evaluation results

MethodLinks
Surgery2026.0212.95
AsFT2026.0222.5
Lisa2026.0222.73
SafeGrad2026.0223.64
DSS2026.0225.68
SPARD2026.0227.73
ConstrainedSFT2026.0228.41
SFT2026.0230.91