Prompt Classification on ToxiC
84.5F1 ScoreHaloGuard 1.0
Evaluation Results
| Method | Links | |
|---|---|---|
| HaloGuard 1.0Size=4B2026.07 | 84.5 | |
| HaloGuard 1.0Size=0.8B2026.07 | 83.5 | |
| Qwen3Guard-GenModel Size=4B, Evaluation protocol=loose2026.01 | 81.9 | |
| Qwen3Guard-GenModel Size=8B, Evaluation protocol=loose2026.01 | 81.3 | |
| Qwen3Guard-GenModel Size=0.6B, Evaluation protocol=loose2026.01 | 76.5 | |
| NemoGuardSize=8B2026.07 | 75.6 | |
| YuFeng-XGuardModel Size=0.6B2026.01 | 74.4 | |
| NemotronGuardV2Model Size=8B2026.01 | 73 | |
| ShieldGemmaSize=27B2026.07 | 72.9 | |
| NemotronReasoningModel Size=4B2026.01 | 72.4 | |
| ShieldGemmaModel Size=9B2026.01 | 72 | |
| PolyGuard-QwenSize=7B2026.07 | 71.5 | |
| YuFeng-XGuardModel Size=8B2026.01 | 71.1 | |
| GPT-OSS-SafeGuardModel Size=20B2026.01 | 71 | |
| WildGuardSize=7B2026.07 | 70.8 | |
| Qwen3Guard-GenSize=4B2026.07 | 69.5 | |
| Qwen3Guard-GenSize=8B2026.07 | 68.9 | |
| PolyGuardModel Size=7B2026.01 | 68.8 | |
| Qwen3Guard-GenSize=0.6B2026.07 | 65.1 | |
| WildGuardModel Size=7B2026.01 | 64.1 | |
| Qwen3Guard-GenModel Size=8B, Evaluation protocol=strict2026.01 | 63.7 | |
| Qwen3Guard-GenModel Size=4B, Evaluation protocol=strict2026.01 | 63.4 | |
| Qwen3Guard-GenModel Size=0.6B, Evaluation protocol=strict2026.01 | 58.9 | |
| LlamaGuard4Size=12B2026.07 | 51.3 | |
| Llama3GuardModel Size=8B2026.01 | 48.4 | |
| Llama4GuardModel Size=12B2026.01 | 42.6 |