Concept Unlearning Robustness on Ring-A-Bell adversarial prompts K38
1.05ASR (T=0.3)
ACE
Updated 1mo ago
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
2026.03 | 1.05 | 1.05 | 0 | 0 | |
2026.03 | 3.16 | 0 | 0 | 0 | |
2026.03 | 32.63 | 24.21 | 9.47 | 4.21 | |
2026.03 | 44.21 | 34.74 | 18.95 | 6.32 | |
2026.03 | 45.26 | 36.84 | 17.89 | 3.16 | |
2026.03 | 51.58 | 40 | 18.95 | 4.21 | |
2026.03 | 52.63 | 36.84 | 23.16 | 8.42 | |
2026.03 | 52.63 | 41.05 | 20 | 4.21 | |
2026.03 | 55.79 | 48.42 | 31.58 | 13.68 | |
2026.03 | 72.63 | 66.32 | 46.32 | 17.89 |