Safety Moderation on PolyGuard
85.8Prompt F1YuFeng-XGuard
Evaluation results
| Method | Links | ||
|---|---|---|---|
| YuFeng-XGuardSize=8B, Task=s-cls2026.05 | 85.8 | 79.2 | |
| GPT-OSS-SafeGuardSize=20B, Task=s-cls2026.05 | 82.9 | 74.8 | |
| WildGuardSize=7B, Task=s-cls2026.05 | 74.7 | 66.1 | |
| GLiGuard OmniSize=209M, Task=s-cls+IE2026.05 | 71.7 | 58.9 | |
| GLiGuard bi-encSize=145M, Task=s-cls+IE2026.05 | 69.5 | 49.9 | |
| Llama3GuardSize=8B, Task=s-cls2026.05 | 68.5 | 66.3 | |
| GLiGuard uni-encSize=147M, Task=s-cls+IE2026.05 | 66.2 | 48.5 | |
| Llama4GuardSize=12B, Task=s-cls2026.05 | 62.4 | 54.7 | |
| GLiNER2 Multi v1Size=209M, Task=cls+IE2026.05 | 60.3 | 29.9 | |
| GLiClass Base v1.0Size=150M, Task=cls2026.05 | 57.6 | 37.1 | |
| NemotronGuardV2Size=8B, Task=s-cls2026.05 | 57.5 | 64.4 | |
| ShieldGemmaSize=9B, Task=s-cls2026.05 | 47.2 | 41 | |
| PromptGuard 2*Size=86M, Task=s-cls2026.05 | 44.3 | 49.3 | |
| DeBERTa-v3-base-PISize=184M, Task=p-inj2026.05 | 43.5 | 49.8 | |
| Longformer-harmful-roSize=149M, Task=s-cls2026.05 | 32.5 | 23.2 |