Attack Detection on Attack Detection Evaluation Set
100Safety (Th=0.3)traceguard
Evaluation results
| Method | Links | ||
|---|---|---|---|
| traceguardScoring Model=Trusted only2026.04 | 100 | — | |
| collusion_resistantScoring Model=Untrusted only2026.04 | 100 | 95 | |
| separation_of_dutiesScoring Model=Mixed (trusted + untrusted)2026.04 | 100 | 86 | |
| simple_trustedScoring Model=Trusted only (single-score)2026.04 | 0 | 0 | |
| simple_untrustedScoring Model=Untrusted only (single-score)2026.04 | 0 | 0 |