Universal Targeted Adversarial Attack on Seen Samples (Used for Optimization) (train)
5.6KMRaAnyAttack
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| AnyAttackTarget Model=Claude2026.01 | 5.6 | 2.3 | 0.3 | 5 | 0.03 | |
| AnyAttackTarget Model=GPT-4o2026.01 | 7.9 | 3.2 | 0.3 | 6.2 | 0.04 | |
| AnyAttackTarget Model=Gemini-2.02026.01 | 9.2 | 4.1 | 0.3 | 6.4 | 0.04 | |
| UnivIntruderTarget Model=Claude2026.01 | 13.4 | 7.3 | 1.1 | 9.2 | 0.05 | |
| UAPTarget Model=Claude2026.01 | 14.9 | 8.3 | 1.6 | 13.8 | 0.07 | |
| M-AttackTarget Model=Claude2026.01 | 15 | 8.5 | 1.2 | 14.5 | 0.08 | |
| MCRMO-AttackTarget Model=Claude2026.01 | 17.8 | 11.4 | 2.9 | 25 | 0.13 | |
| FOA-AttackTarget Model=Claude2026.01 | 18.6 | 10.5 | 2.1 | 18 | 0.1 | |
| UnivIntruderTarget Model=GPT-4o2026.01 | 22.9 | 12.8 | 2.4 | 15 | 0.07 | |
| UnivIntruderTarget Model=Gemini-2.02026.01 | 27.3 | 16.5 | 3.2 | 15 | 0.07 | |
| UAPTarget Model=Gemini-2.02026.01 | 60.2 | 40.7 | 11.2 | 61.3 | 0.29 | |
| UAPTarget Model=GPT-4o2026.01 | 62.6 | 41.1 | 10.5 | 66.7 | 0.32 | |
| MCRMO-AttackTarget Model=Gemini-2.02026.01 | 72.7 | 51.4 | 11.3 | 75.5 | 0.36 | |
| MCRMO-AttackTarget Model=GPT-4o2026.01 | 73.5 | 50.5 | 14 | 85.5 | 0.39 | |
| M-AttackTarget Model=Gemini-2.02026.01 | 74.1 | 52.9 | 14.2 | 80.3 | 0.44 | |
| FOA-AttackTarget Model=Gemini-2.02026.01 | 80 | 57.1 | 17.5 | 85.4 | 0.48 | |
| M-AttackTarget Model=GPT-4o2026.01 | 81.7 | 59.5 | 18.6 | 91.2 | 0.53 | |
| FOA-AttackTarget Model=GPT-4o2026.01 | 84.1 | 60.9 | 20.1 | 93 | 0.57 |