Prompt Harmfulness Classification on WildG
96.2F1 ScoreHaloGuard 1.0
Evaluation Results
| Method | Links | |
|---|---|---|
| HaloGuard 1.0Size=4B2026.07 | 96.2 | |
| HaloGuard 1.0Size=0.8B2026.07 | 95.9 | |
| Qwen3Guard-8B-GenParams=8B2026.07 | 89.1 | |
| WildGuardSize=7B2026.07 | 88.9 | |
| Qwen3Guard-GenSize=8B2026.07 | 88.9 | |
| Stage1-SFT-v4Params=4B2026.07 | 88.8 | |
| Stage3-DPOParams=4B2026.07 | 88.8 | |
| Qwen3Guard-4B-GenParams=4B2026.07 | 88.7 | |
| Stage2-SFTParams=4B2026.07 | 88.4 | |
| PolyGuard-QwenSize=7B2026.07 | 88.1 | |
| YuFeng-XGuard-Reason-0.6BParams=0.6B2026.07 | 87.9 | |
| Qwen3Guard-0.8B-GenParams=0.8B2026.07 | 87.8 | |
| Qwen3Guard-GenSize=0.6B2026.07 | 87.7 | |
| YuFeng-XGuard-Reason-8BParams=8B2026.07 | 87.7 | |
| Stage1-SFT-v0Params=4B2026.07 | 87 | |
| Stage1-SFT-v1Params=4B2026.07 | 87 | |
| Stage1-SFT-v3Params=4B2026.07 | 86.4 | |
| Qwen3Guard-GenSize=4B2026.07 | 85.6 | |
| Stage1-SFT-v2Params=4B2026.07 | 82.9 | |
| NemoGuardSize=8B2026.07 | 81.6 | |
| LlamaGuard4Size=12B2026.07 | 73 | |
| ShieldGemmaSize=27B2026.07 | 54.3 |