Jailbreak Defense on AdvBench Mistral-7B v0.2
59.2ASRNone
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| NoneAttack Protocol=Adaptive Attack2026.05 | 59.2 | 6.03 | |
| IBProtectorAttack Protocol=Adaptive Attack2026.05 | 44.2 | 8.31 | |
| Semantic SmoothAttack Protocol=Adaptive Attack2026.05 | 35.8 | 10.79 | |
| Self DefenseAttack Protocol=Adaptive Attack2026.05 | 31.7 | 9.97 | |
| SmoothLLMAttack Protocol=Adaptive Attack2026.05 | 19.2 | 10.8 | |
| RA-LLMAttack Protocol=Adaptive Attack2026.05 | 0.8 | 18.83 | |
| EvoDefenseAttack Protocol=Adaptive Attack2026.05 | 0.7 | 18.04 |