List your company

Group Collusive Attack Detection on GSM8K

92Detection AccuracyGroupGuard
52.06462.43272.883.168Mar 14, 2026

Evaluation results

MethodLinks
GroupGuardAttack Type=Group Exclusion, Model=Deepseek-V3.12026.0392
G-safeguardAttack Type=Targeted Suppression, Model=Deepseek-V3.12026.0391.46
GroupGuardAttack Type=Targeted Suppression, Model=Deepseek-V3.12026.0391
G-safeguardAttack Type=Group Exclusion, Model=Deepseek-V3.12026.0389.75
GroupGuardAttack Type=Group Exclusion, Model=Qwen3-235B-A22B2026.0389
AutoDefenseAttack Type=Group Exclusion, Model=Deepseek-V3.12026.0387.8
GroupGuardAttack Type=False Consensus, Model=Qwen3-32B2026.0387
GroupGuardAttack Type=Group Exclusion, Model=Qwen3-32B2026.0386
G-safeguardAttack Type=False Consensus, Model=Qwen3-32B2026.0385.8
GroupGuardAttack Type=False Consensus, Model=Deepseek-V3.12026.0384
G-safeguardAttack Type=False Consensus, Model=Deepseek-V3.12026.0383.2
G-safeguardAttack Type=Group Exclusion, Model=Qwen3-235B-A22B2026.0382
G-safeguardAttack Type=Group Exclusion, Model=Qwen3-32B2026.0381
AutoDefenseAttack Type=Group Exclusion, Model=Qwen3-32B2026.0377.4
AutoDefenseAttack Type=Targeted Suppression, Model=Deepseek-V3.12026.0376.6
AutoDefenseAttack Type=False Consensus, Model=Deepseek-V3.12026.0370.8
AutoDefenseAttack Type=Targeted Suppression, Model=Qwen3-235B-A22B2026.0369.4
AutoDefenseAttack Type=False Consensus, Model=Qwen3-32B2026.0367.8
AutoDefenseAttack Type=False Consensus, Model=Qwen3-235B-A22B2026.0367.2
G-safeguardAttack Type=False Consensus, Model=Qwen3-235B-A22B2026.0367
GroupGuardAttack Type=False Consensus, Model=Qwen3-235B-A22B2026.0366
AutoDefenseAttack Type=Group Exclusion, Model=Qwen3-235B-A22B2026.0363.6
GroupGuardAttack Type=Targeted Suppression, Model=Qwen3-32B2026.0363
GroupGuardAttack Type=Targeted Suppression, Model=Qwen3-235B-A22B2026.0362
G-safeguardAttack Type=Targeted Suppression, Model=Qwen3-235B-A22B2026.0361.8
G-safeguardAttack Type=Targeted Suppression, Model=Qwen3-32B2026.0361.2
AutoDefenseAttack Type=Targeted Suppression, Model=Qwen3-32B2026.0353.6