Response Safety on MM-SafetyBench (avg)
99MS-RDACO
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| DACOTarget Model=Qwen2.5-VL-7B-Instruct [5], Steering Method=DACO (Ours)2026.04 | 99 | 98.4 | |
| DACOTarget Model=InternVL3.5-8B-Instruct [108], Steering Method=DACO (Ours)2026.04 | 98.1 | 98.7 | |
| DACOTarget Model=LLaVA1.6-Mistral-7B [48], Steering Method=DACO (Ours)2026.04 | 96.1 | 99.5 | |
| MOP [56]Target Model=InternVL3.5-8B-Instruct [108], Steering Method=MOP [56]2026.04 | 95.6 | 98.3 | |
| ActAdd [50, 101, 134]Target Model=LLaVA1.6-Mistral-7B [48], Steering Method=ActAdd [50, 101, 134]2026.04 | 94.4 | 98.8 | |
| MOP [56]Target Model=LLaVA1.6-Mistral-7B [48], Steering Method=MOP [56]2026.04 | 93.8 | 95.4 | |
| ActAdd [50, 101, 134]Target Model=InternVL3.5-8B-Instruct [108], Steering Method=ActAdd [50, 101, 134]2026.04 | 80.9 | 92.9 | |
| Prompting [52]Target Model=LLaVA1.6-Mistral-7B [48], Steering Method=Prompting [52]2026.04 | 78.3 | 89.1 | |
| MOP [56]Target Model=Qwen2.5-VL-7B-Instruct [5], Steering Method=MOP [56]2026.04 | 77.1 | 84 | |
| Prompting [52]Target Model=InternVL3.5-8B-Instruct [108], Steering Method=Prompting [52]2026.04 | 72.4 | 89.7 | |
| No SteeringTarget Model=LLaVA1.6-Mistral-7B [48], Steering Method=No Steering2026.04 | 71.5 | 66.9 | |
| ActAdd [50, 101, 134]Target Model=Qwen2.5-VL-7B-Instruct [5], Steering Method=ActAdd [50, 101, 134]2026.04 | 65.3 | 73.5 | |
| Prompting [52]Target Model=Qwen2.5-VL-7B-Instruct [5], Steering Method=Prompting [52]2026.04 | 60.7 | 71.1 | |
| No SteeringTarget Model=InternVL3.5-8B-Instruct [108], Steering Method=No Steering2026.04 | 47.2 | 71.9 | |
| No SteeringTarget Model=Qwen2.5-VL-7B-Instruct [5], Steering Method=No Steering2026.04 | 44.2 | 65.2 |