Safety Evaluation on CATQA FigStep
100Safety ScorePSI
Evaluation Results
| Method | Links | |
|---|---|---|
| PSIBackbone Model=HuatuoGPT-Vision-7B2025.12 | 100 | |
| RESTABackbone Model=Lingshu-7B2025.12 | 99.82 | |
| RESTABackbone Model=HuatuoGPT-Vision-7B2025.12 | 99.27 | |
| PSIBackbone Model=Lingshu-7B2025.12 | 93.45 | |
| ModelMergeBackbone Model=Lingshu-7B2025.12 | 22 | |
| OriginalBackbone Model=Lingshu-7B2025.12 | 21.27 | |
| ModelMergeBackbone Model=HuatuoGPT-Vision-7B2025.12 | 17.45 | |
| OriginalBackbone Model=HuatuoGPT-Vision-7B2025.12 | 12.36 |