List your company

Safety Alignment Evaluation on Harmful (test)

4Harmful Score (HS)AsFT
3.4567.12810.814.472Jun 10, 2025

Evaluation results

MethodLinks
AsFTUtility Task=AGNEWS2025.06484.3
Safe LoRAUtility Task=AGNEWS2025.065.681.2
BEAUtility Task=AGNEWS2025.0616.484.4
Lisa-alignedUtility Task=AGNEWS2025.0616.881.8
SafeInstrUtility Task=AGNEWS2025.0616.884.3
Lisa-baseUtility Task=AGNEWS2025.0617.273.5
SFTUtility Task=AGNEWS2025.0617.684.3