List your company

Group Collusive Attack Detection on HumanEval

96Detection AccuracyGroupGuard
45.66458.73271.884.868Mar 14, 2026

Evaluation results

MethodLinks
GroupGuardAttack Type=Group Exclusion, Model=Deepseek-V3.12026.0396
GroupGuardAttack Type=Group Exclusion, Model=Qwen3-235B-A22B2026.0393
G-safeguardAttack Type=Group Exclusion, Model=Deepseek-V3.12026.0391.2
GroupGuardAttack Type=False Consensus, Model=Qwen3-32B2026.0389
GroupGuardAttack Type=False Consensus, Model=Deepseek-V3.12026.0387
G-safeguardAttack Type=Group Exclusion, Model=Qwen3-235B-A22B2026.0386.4
G-safeguardAttack Type=False Consensus, Model=Deepseek-V3.12026.0386.2
GroupGuardAttack Type=False Consensus, Model=Qwen3-235B-A22B2026.0386
AutoDefenseAttack Type=Group Exclusion, Model=Qwen3-235B-A22B2026.0381.6
G-safeguardAttack Type=False Consensus, Model=Qwen3-32B2026.0380.8
AutoDefenseAttack Type=Group Exclusion, Model=Deepseek-V3.12026.0380.6
G-safeguardAttack Type=False Consensus, Model=Qwen3-235B-A22B2026.0380
AutoDefenseAttack Type=False Consensus, Model=Qwen3-235B-A22B2026.0377.4
GroupGuardAttack Type=Group Exclusion, Model=Qwen3-32B2026.0377
G-safeguardAttack Type=Group Exclusion, Model=Qwen3-32B2026.0375
G-safeguardAttack Type=Targeted Suppression, Model=Deepseek-V3.12026.0374.7
AutoDefenseAttack Type=Targeted Suppression, Model=Deepseek-V3.12026.0374.4
GroupGuardAttack Type=Targeted Suppression, Model=Qwen3-235B-A22B2026.0374
AutoDefenseAttack Type=False Consensus, Model=Qwen3-32B2026.0372.6
G-safeguardAttack Type=Targeted Suppression, Model=Qwen3-235B-A22B2026.0370.8
GroupGuardAttack Type=Targeted Suppression, Model=Deepseek-V3.12026.0370
AutoDefenseAttack Type=Group Exclusion, Model=Qwen3-32B2026.0369.2
G-safeguardAttack Type=Targeted Suppression, Model=Qwen3-32B2026.0367.8
AutoDefenseAttack Type=Targeted Suppression, Model=Qwen3-235B-A22B2026.0367.2
GroupGuardAttack Type=Targeted Suppression, Model=Qwen3-32B2026.0365
AutoDefenseAttack Type=False Consensus, Model=Deepseek-V3.12026.0353.6
AutoDefenseAttack Type=Targeted Suppression, Model=Qwen3-32B2026.0347.6