Prompt Harmfulness Classification on OAI
87.4F1 ScoreHaloGuard 1.0
Evaluation Results
| Method | Links | |
|---|---|---|
| HaloGuard 1.0Size=4B2026.07 | 87.4 | |
| HaloGuard 1.0Size=0.8B2026.07 | 83.7 | |
| NemoGuardSize=8B2026.07 | 81 | |
| ShieldGemmaSize=27B2026.07 | 80.5 | |
| PolyGuard-QwenSize=7B2026.07 | 74.1 | |
| LlamaGuard4Size=12B2026.07 | 73.5 | |
| WildGuardSize=7B2026.07 | 72.1 | |
| Qwen3Guard-GenSize=8B2026.07 | 68.8 | |
| Qwen3Guard-GenSize=4B2026.07 | 68.3 | |
| Qwen3Guard-GenSize=0.6B2026.07 | 66.5 |