List your company

Jailbreak Detection on AdvBench Mistral-7B

99AUROCMTK
33.16850.25967.3584.441Jun 5, 2026

Evaluation results

MethodLinks
MTKAttack=AutoDAN2026.0699
MTKAttack=Avg.2026.0695.3
SelfDefendAttack=AutoDAN2026.0694.5
HiddenDetectAttack=AutoDAN2026.0685.8
SelfDefendAttack=Avg.2026.0684.8
GradCuffAttack=AutoDAN2026.0682
HSFAttack=Avg.2026.0680.8
HiddenDetectAttack=Avg.2026.0678.9
HSFAttack=AutoDAN2026.0671.7
SaPAttack=Avg.2026.0663.5
GradCuffAttack=Avg.2026.0660
SmoothLLMAttack=Avg.2026.0653.7
SaPAttack=AutoDAN2026.0650.5
GradSafeAttack=Avg.2026.0647.3
GradSafeAttack=AutoDAN2026.0643.1
SmoothLLMAttack=AutoDAN2026.0635.7