Response Harmfulness Detection on SPA-VL-Eval
74.73F1 ScoreGuardReasoner-Omni 2B
Evaluation results
| Method | Links | |
|---|---|---|
| GuardReasoner-Omni 2BModel Category=VLM Guard Model2026.02 | 74.73 | |
| GuardReasoner-VL 7BModel Category=VLM Guard Model2026.02 | 72.62 | |
| GuardReasoner-Omni 4BModel Category=VLM Guard Model2026.02 | 72.13 | |
| GuardReasoner-VL 3BModel Category=VLM Guard Model2026.02 | 72.01 | |
| LLaMA Guard 4 12BModel Category=VLM Guard Model2026.02 | 52.57 |