Safety Evaluation on ViSU safe 24
0.1Refusal RateZero-shot
Evaluation Results
| Method | Links | |
|---|---|---|
| Zero-shotModel=GEMMA32026.06 | 0.1 | |
| Zero-shotModel=MOLMO22026.06 | 0.1 | |
| Zero-shotModel=QWEN3-VL2026.06 | 0.1 | |
| Zero-shotModel=QWEN3.52026.06 | 0.1 | |
| ECSOModel=GEMMA32026.06 | 0.1 | |
| ECSOModel=QWEN3-VL2026.06 | 0.1 | |
| ECSOModel=QWEN3.52026.06 | 0.1 | |
| AdaSteerModel=GEMMA32026.06 | 0.1 | |
| AdaSteerModel=MOLMO22026.06 | 0.1 | |
| AdaSteerModel=QWEN3-VL2026.06 | 0.1 | |
| AdaSteerModel=QWEN3.52026.06 | 0.1 | |
| Zero-shotModel=INTERNVL 3.52026.06 | 0.5 | |
| ECSOModel=MOLMO22026.06 | 0.5 | |
| ECSOModel=INTERNVL 3.52026.06 | 0.5 | |
| MARSModel=QWEN3.52026.06 | 0.6 | |
| AdaSteerModel=INTERNVL 3.52026.06 | 0.8 | |
| MARSModel=GEMMA32026.06 | 1.3 | |
| MARSModel=QWEN3-VL2026.06 | 2.2 | |
| SASAModel=QWEN3-VL2026.06 | 2.5 | |
| SASAModel=INTERNVL 3.52026.06 | 4.9 | |
| MARSModel=MOLMO22026.06 | 7.3 | |
| MARSModel=INTERNVL 3.52026.06 | 10.3 | |
| SASAModel=QWEN3.52026.06 | 13.7 | |
| SASAModel=MOLMO22026.06 | 15.2 | |
| SASAModel=GEMMA32026.06 | 71.1 |