Jailbreak Defense Evaluation on SB
3.258Strong-Reject Score (SR)vanilla
Evaluation Results
| Method | Links | |
|---|---|---|
| vanillaBase Model=Mistral-7b-U2026.02 | 3.258 | |
| SCANSBase Model=Mistral-7b-U2026.02 | 3.008 | |
| vanillaBase Model=Llama2-7b-U2026.02 | 2.892 | |
| SAGEBase Model=Llama2-7b-U2026.02 | 2.642 | |
| SAGEBase Model=Mistral-7b-U2026.02 | 2.533 | |
| ICDBase Model=Llama2-7b-U2026.02 | 2.175 | |
| SCANSBase Model=Llama2-7b-U2026.02 | 2.167 | |
| vanillaBase Model=Llama3.1-8b-D2026.02 | 2.158 | |
| SAGEBase Model=Llama3.1-8b-D2026.02 | 1.958 | |
| SCANSBase Model=Llama3.1-8b-D2026.02 | 1.933 | |
| InstructBase Model=Mistral-7b-U2026.02 | 1.892 | |
| SelfDefenDBase Model=Llama2-7b-U2026.02 | 1.825 | |
| AISABase Model=Llama2-7b-U2026.02 | 1.558 | |
| SelfDefenDBase Model=Llama3.1-8b-D2026.02 | 1.492 | |
| ICDBase Model=Mistral-7b-U2026.02 | 1.375 | |
| AISABase Model=Llama3.1-8b-D2026.02 | 1.175 | |
| SelfDefenDBase Model=Mistral-7b-U2026.02 | 1.117 | |
| InstructBase Model=Llama2-7b-U2026.02 | 1.117 | |
| InstructBase Model=Llama3.1-8b-D2026.02 | 1.117 | |
| AISABase Model=Mistral-7b-U2026.02 | 1.108 | |
| ICDBase Model=Llama3.1-8b-D2026.02 | 1.1 |