List your company

Alignment Defense against Harmful Fine-tuning on SST2

11.3Harmful Score (HS)Antibody
9.68620.580531.47542.3695Feb 28, 2026

Evaluation results

MethodLinks
AntibodyBackbone=Gemma-2-9B2026.0211.394.27
BoosterBackbone=Gemma-2-9B2026.0228.4794.15
LisaBackbone=Gemma-2-9B2026.0234.6794.04
SFTBackbone=Gemma-2-9B2026.0240.7794.11
VaccineBackbone=Gemma-2-9B2026.0251.6594.5