Targeted Attack on GPT closed-source standard MLLMs 5.4
3.8ASRFRA-Attack
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| FRA-AttackThreshold=0.92026.05 | 3.8 | 41.1 | |
| M-Attack-V2Threshold=0.92026.05 | 1.5 | 36.9 | |
| FOA-AttackThreshold=0.92026.05 | 0.6 | 30 | |
| M-AttackThreshold=0.92026.05 | 0.3 | 26.8 | |
| AttackVLMThreshold=0.92026.05 | 0 | 4.4 | |
| AdvDiffVLMThreshold=0.92026.05 | 0 | 3.6 | |
| SSA-CWAThreshold=0.92026.05 | 0 | 3.3 | |
| AnyAttackThreshold=0.92026.05 | 0 | 5.7 |