Jailbreak Defense Evaluation on HB
3.065Strong-Reject Score (SR)vanilla
Evaluation Results
| Method | Links | |
|---|---|---|
| vanillaBase Model=Mistral-7b-U2026.02 | 3.065 | |
| vanillaBase Model=Llama2-7b-U2026.02 | 2.75 | |
| vanillaBase Model=Llama3.1-8b-D2026.02 | 2.425 | |
| SCANSBase Model=Mistral-7b-U2026.02 | 2.315 | |
| SAGEBase Model=Mistral-7b-U2026.02 | 2.25 | |
| SAGEBase Model=Llama2-7b-U2026.02 | 2.245 | |
| ICDBase Model=Llama2-7b-U2026.02 | 2.165 | |
| SelfDefenDBase Model=Llama2-7b-U2026.02 | 1.89 | |
| SAGEBase Model=Llama3.1-8b-D2026.02 | 1.88 | |
| InstructBase Model=Mistral-7b-U2026.02 | 1.74 | |
| SelfDefenDBase Model=Llama3.1-8b-D2026.02 | 1.69 | |
| SCANSBase Model=Llama2-7b-U2026.02 | 1.615 | |
| SCANSBase Model=Llama3.1-8b-D2026.02 | 1.575 | |
| InstructBase Model=Llama3.1-8b-D2026.02 | 1.335 | |
| SelfDefenDBase Model=Mistral-7b-U2026.02 | 1.33 | |
| AISABase Model=Llama2-7b-U2026.02 | 1.23 | |
| ICDBase Model=Mistral-7b-U2026.02 | 1.15 | |
| ICDBase Model=Llama3.1-8b-D2026.02 | 1.136 | |
| AISABase Model=Mistral-7b-U2026.02 | 1.05 | |
| AISABase Model=Llama3.1-8b-D2026.02 | 1.045 | |
| InstructBase Model=Llama2-7b-U2026.02 | 1.015 |