List your company

Jailbreaking on MaliciousInstruct

4.8BERT ScoreLINT
1.06642.03573.0053.9743May 17, 2025

Evaluation results

MethodLinks
LINTAPI Compatibility=false2025.054.82.360.4
FLIPAPI Compatibility=true2025.054.422.030.46
NaiveAPI Compatibility=true2025.054.071.540.17
JULITarget Model=Llama2-7B-CHAT, Defense=DEEPALIGN, Evaluation Setting=open-source setting2025.054.013.843.19
GCGAPI Compatibility=false2025.053.631.740.37
EDAPI Compatibility=false2025.053.553.662.74
JULIAPI Compatibility=false2025.053.392.521.72
JULI (API)API Compatibility=true2025.052.861.950.71
WTSAPI Compatibility=false2025.052.041.330.21
BaseAPI Compatibility=true2025.051.981.160.04
NaiveTarget Model=Llama2-7B-CHAT, Defense=DEEPALIGN, Evaluation Setting=open-source setting2025.051.2110