Toxicity Detection on Prompt Overflow K=4
100Bypass RateGranite Guardian 125M
Evaluation Results
| Method | Links | |
|---|---|---|
| Granite Guardian 125MFiller type=Blank\, Layout strategy=Tail, Density (K)=42026.05 | 100 | |
| Granite Guardian 125MFiller type=Blank\, Layout strategy=Interleave, Density (K)=42026.05 | 100 | |
| Granite Guardian 125MFiller type=Novel Scripts, Layout strategy=Interleave, Density (K)=42026.05 | 100 |