Pre-load security auditing on SkillGuardBench Ext.-eco
99.66Overall Exact Match RateSKILLGUARD-ROBUST
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| SKILLGUARD-ROBUST2026.04 | 99.66 | 100 | 99.26 | 100 | 99.27 | 100 | 100 | |
| SKILLGUARD-CALIBRATE2026.04 | 98.45 | 100 | 97.62 | 100 | 97.67 | 100 | 99.83 | |
| SKILLGUARD-VERIFY2026.04 | 97.03 | 100 | 95.62 | 100 | 95.67 | 100 | 99.33 | |
| BUNDLEJUDGE2026.04 | 93.72 | 99 | 89.68 | 88.3 | 92.13 | 91.24 | 98.5 | |
| Qwen2.5-14BPackage Mode=full-package2026.04 | 89.28 | 98 | 84.48 | 83 | 87.33 | 86.1 | 97 |