Safety Classification on MultiJail
0.9335F1 ScoreCREST-BASE
Evaluation Results
| Method | Links | |
|---|---|---|
| CREST-BASEModel Variant=Base, Backbone=XLM-RoBERTa-Base2025.12 | 0.9335 | |
| CREST-LARGEModel Variant=Large, Backbone=XLM-RoBERTa-Large2025.12 | 0.9329 | |
| NemotronParameters=8B2026.05 | 0.85 | |
| ML-GUARDParameters=7B2026.05 | 0.85 | |
| ML-GUARDParameters=1.5B2026.05 | 0.83 | |
| gpt-oss-safeguardParameters=20B2026.05 | 0.8 | |
| PolyGuard-Qwen2026.05 | 0.78 | |
| Qwen3Guard-GenParameters=8B2026.05 | 0.77 | |
| Qwen3Guard-GenParameters=4B2026.05 | 0.74 | |
| Qwen3Guard-GenParameters=0.6B2026.05 | 0.72 | |
| Omni-moderation2026.05 | 0.66 | |
| Llama-Guard-3Parameters=8B2026.05 | 0.63 | |
| DuoGuardParameters=1.5B2026.05 | 0.58 | |
| Llama-Guard-4Parameters=12B2026.05 | 0.57 | |
| Llama-Guard-3Parameters=1B2026.05 | 0.45 |