Jailbreak Defense on JailbreakBench and AdvBench
0.1ASR
Certified Semantic Smoothing
Updated 4d ago
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
2026.02 | 0.1 | 97.1 | 22.5 | 68.2 | 2.15 | |
2026.02 | 0.1 | 97.4 | 22.9 | 68.2 | 2.15 | |
2026.02 | 0.3 | 94.8 | 18.9 | 86.5 | 2.05 | |
2026.02 | 0.5 | 94.2 | 18.4 | 86.5 | 2.05 | |
2026.02 | 0.6 | 93.5 | 18.1 | 86.5 | 2.05 | |
2026.02 | 0.8 | 90.1 | 15.2 | 94.1 | 1.95 | |
2026.02 | 1.2 | 89.4 | 14.6 | 94.1 | 1.95 | |
2026.02 | 1.5 | 88.7 | 14.3 | 94.1 | 1.95 | |
2026.02 | 3.5 | 62.1 | 4.2 | 74.3 | 2.1 | |
2026.02 | 4.1 | 61.8 | 4.1 | 74.3 | 2.1 | |
2026.02 | 8.2 | 58.4 | 3.8 | 74.3 | 2.1 | |
2026.02 | 12.4 | - | - | 88.5 | 4.2 | |
2026.02 | 14.1 | - | - | 88.5 | 4.2 | |
2026.02 | 18.9 | 24.5 | 1.2 | 89.2 | 3.45 | |
2026.02 | 22.1 | 23.8 | 1.1 | 89.2 | 3.45 | |
2026.02 | 38.4 | - | - | 91.2 | 0.92 | |
2026.02 | 41.2 | - | - | 91.2 | 0.92 | |
2026.02 | 45.3 | - | - | 91.2 | 0.92 | |
2026.02 | 68.9 | 0 | 0 | 96.5 | 0.85 | |
2026.02 | 72.1 | 0 | 0 | 96.5 | 0.85 | |
2026.02 | 84.2 | 0 | 0 | 96.5 | 0.85 |