| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Safety Dialogue Evaluation | SafeDialBench | Score8.42 | 33 | |
| Safety Detection | SafeDialBench (full) | Recall99 | 12 | |
| Safety dialogue evaluation | SafeDialBench | Normalized Score61.33 | 5 | |
| Unsafe-input detection | SafeDialBench EN | Recall99.07 | 2 |