Harmful Request Defense on SORRY-Bench
13ASR
Self-Guard
Updated 4d ago
Evaluation Results
| Method | Links | ||
|---|---|---|---|
2026.01 | 13 | - | |
2026.01 | 14.3 | - | |
2026.01 | 15.7 | - | |
2026.01 | 16.1 | - | |
2026.01 | 16.6 | - | |
2026.01 | 18.2 | - | |
2026.01 | 18.4 | - | |
2026.01 | 18.9 | - | |
2026.01 | 18.9 | - | |
2026.01 | 20.7 | - | |
2026.01 | 22.5 | - | |
2026.01 | 22.7 | - | |
2026.01 | 22.7 | - | |
2026.01 | 22.7 | - | |
2026.01 | 27.2 | - | |
2026.01 | 43 | - | |
2026.01 | 45.9 | - | |
2026.01 | 47.5 | - | |
2026.01 | 50.9 | - | |
2026.01 | 52.3 | - | |
2026.01 | 52.7 | - | |
2026.01 | 53 | - | |
2026.01 | 53.2 | - | |
2026.01 | 53.9 | - |