Content Safety Detection on CyberSecEval
92.2F1 Scoreprompt–response verification framework
Evaluation results
| Method | Links | ||
|---|---|---|---|
| prompt–response verification framework2026.06 | 92.2 | 5.7 | |
| Single-Agent+CoTType=Strongest Applicable Baseline2026.06 | 84.7 | — |
| Method | Links | ||
|---|---|---|---|
| prompt–response verification framework2026.06 | 92.2 | 5.7 | |
| Single-Agent+CoTType=Strongest Applicable Baseline2026.06 | 84.7 | — |