Jailbreak Defense Evaluation on ALL-4
2.664SR Scorevanilla
Evaluation Results
| Method | Links | |
|---|---|---|
| vanillaBase Model=Mistral-7b-U2026.02 | 2.664 | |
| SCANSBase Model=Mistral-7b-U2026.02 | 2.643 | |
| vanillaBase Model=Llama3.1-8b-D2026.02 | 2.579 | |
| SCANSBase Model=Llama3.1-8b-D2026.02 | 2.549 | |
| vanillaBase Model=Llama2-7b-U2026.02 | 2.511 | |
| SelfDefenDBase Model=Llama3.1-8b-D2026.02 | 2.496 | |
| SAGEBase Model=Mistral-7b-U2026.02 | 2.473 | |
| SCANSBase Model=Llama2-7b-U2026.02 | 2.47 | |
| ICDBase Model=Llama3.1-8b-D2026.02 | 2.464 | |
| InstructBase Model=Mistral-7b-U2026.02 | 2.448 | |
| SAGEBase Model=Llama3.1-8b-D2026.02 | 2.417 | |
| SelfDefenDBase Model=Llama2-7b-U2026.02 | 2.403 | |
| SAGEBase Model=Llama2-7b-U2026.02 | 2.37 | |
| ICDBase Model=Llama2-7b-U2026.02 | 2.305 | |
| ICDBase Model=Mistral-7b-U2026.02 | 2.264 | |
| SelfDefenDBase Model=Mistral-7b-U2026.02 | 2.251 | |
| AISABase Model=Llama2-7b-U2026.02 | 1.993 | |
| InstructBase Model=Llama3.1-8b-D2026.02 | 1.84 | |
| AISABase Model=Llama3.1-8b-D2026.02 | 1.586 | |
| AISABase Model=Mistral-7b-U2026.02 | 1.535 | |
| InstructBase Model=Llama2-7b-U2026.02 | 1.227 |