Safety Detection on Polyguard Social Media
96.73AUROCT3
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| T3classifier=GMM2026.02 | 96.73 | 12.08 | |
| T3classifier=OCSVM2026.02 | 96.2 | 14.85 | |
| WildGuard2026.02 | 82.52 | 100 | |
| LlamaGuard3parameters=1B, output_format=logits2026.02 | 82.49 | 65.76 | |
| CIDER2026.02 | 76.45 | 93.56 | |
| LlamaGuard3parameters=1B2026.02 | 74.82 | 100 | |
| MDJudge2026.02 | 74.45 | 87.82 | |
| DuoGuard2026.02 | 72.24 | 94.46 | |
| Polyguard2026.02 | 72.24 | 68.08 | |
| LlamaGuard4parameters=12B2026.02 | 71.51 | 100 | |
| FDBD2026.02 | 67.62 | 94.8 | |
| GMM2026.02 | 63.29 | 98.58 | |
| RMD2026.02 | 61.85 | 97.64 | |
| OPENMAX2026.02 | 61.26 | 99.96 | |
| VIM2026.02 | 46.06 | 99.96 | |
| NNGUIDE2026.02 | 42.27 | 99.95 | |
| REACT2026.02 | 32.05 | 99.87 | |
| ADASCALE2026.02 | 27.19 | 99.97 |