List your company

Targeted Refusal on Mistral-7B Generation Evaluation Set

97.01CAVanilla
79.704484.197288.6993.1828Nov 16, 2025

Evaluation results

MethodLinks
VanillaAttack=Sleeper Agent2025.1197.0199.01
VanillaAttack=CBA2025.1196.896.4
VanillaAttack=VPI2025.1196.1798.45
OURSAttack=Sleeper Agent2025.1194.18.54
OURSAttack=VPI2025.1193.6115.14
GracCeFulAttack=Sleeper Agent2025.1192.8115.41
OURSAttack=CBA2025.1192.1711.3
CleanGenAttack=VPI2025.1192.0425.12
CleanGenAttack=Sleeper Agent2025.1191.5919.54
GracCeFulAttack=VPI2025.1191.2817.35
CleanGenAttack=CBA2025.1190.4524.81
GracCeFulAttack=CBA2025.1189.8716.54
MuScleLoRAAttack=Sleeper Agent2025.1184.813.7
MuScleLoRAAttack=CBA2025.1181.5214.83
MuScleLoRAAttack=VPI2025.1180.3716.48