Multimodal Jailbreaking on SafeBench FigStep (ID)
92.3ASRVanilla
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| VanillaModel=LLaVA-1.5-7B2026.03 | 92.3 | 4.38 | |
| VanillaModel=Qwen3-VL-8B2026.03 | 30.8 | 1.54 | |
| Static SFTModel=Qwen3-VL-8B2026.03 | 0 | 0 | |
| CEMMAModel=Qwen3-VL-8B2026.03 | 0 | 0 | |
| Static SFTModel=LLaVA-1.5-7B2026.03 | 0 | 0 | |
| CEMMAModel=LLaVA-1.5-7B2026.03 | 0 | 0 |