List your company

Jailbreak Defense Evaluation on SB

3.258Strong-Reject Score (SR)vanilla
1.013681.596342.1792.76166Feb 14, 2026

Evaluation results

MethodLinks
vanillaBase Model=Mistral-7b-U2026.023.258
SCANSBase Model=Mistral-7b-U2026.023.008
vanillaBase Model=Llama2-7b-U2026.022.892
SAGEBase Model=Llama2-7b-U2026.022.642
SAGEBase Model=Mistral-7b-U2026.022.533
ICDBase Model=Llama2-7b-U2026.022.175
SCANSBase Model=Llama2-7b-U2026.022.167
vanillaBase Model=Llama3.1-8b-D2026.022.158
SAGEBase Model=Llama3.1-8b-D2026.021.958
SCANSBase Model=Llama3.1-8b-D2026.021.933
InstructBase Model=Mistral-7b-U2026.021.892
SelfDefenDBase Model=Llama2-7b-U2026.021.825
AISABase Model=Llama2-7b-U2026.021.558
SelfDefenDBase Model=Llama3.1-8b-D2026.021.492
ICDBase Model=Mistral-7b-U2026.021.375
AISABase Model=Llama3.1-8b-D2026.021.175
SelfDefenDBase Model=Mistral-7b-U2026.021.117
InstructBase Model=Llama2-7b-U2026.021.117
InstructBase Model=Llama3.1-8b-D2026.021.117
AISABase Model=Mistral-7b-U2026.021.108
ICDBase Model=Llama3.1-8b-D2026.021.1