Group Collusive Attack Detection on GSM8K
92Detection AccuracyGroupGuard
Evaluation Results
| Method | Links | |
|---|---|---|
| GroupGuardAttack Type=Group Exclusion, Model=Deepseek-V3.12026.03 | 92 | |
| G-safeguardAttack Type=Targeted Suppression, Model=Deepseek-V3.12026.03 | 91.46 | |
| GroupGuardAttack Type=Targeted Suppression, Model=Deepseek-V3.12026.03 | 91 | |
| G-safeguardAttack Type=Group Exclusion, Model=Deepseek-V3.12026.03 | 89.75 | |
| GroupGuardAttack Type=Group Exclusion, Model=Qwen3-235B-A22B2026.03 | 89 | |
| AutoDefenseAttack Type=Group Exclusion, Model=Deepseek-V3.12026.03 | 87.8 | |
| GroupGuardAttack Type=False Consensus, Model=Qwen3-32B2026.03 | 87 | |
| GroupGuardAttack Type=Group Exclusion, Model=Qwen3-32B2026.03 | 86 | |
| G-safeguardAttack Type=False Consensus, Model=Qwen3-32B2026.03 | 85.8 | |
| GroupGuardAttack Type=False Consensus, Model=Deepseek-V3.12026.03 | 84 | |
| G-safeguardAttack Type=False Consensus, Model=Deepseek-V3.12026.03 | 83.2 | |
| G-safeguardAttack Type=Group Exclusion, Model=Qwen3-235B-A22B2026.03 | 82 | |
| G-safeguardAttack Type=Group Exclusion, Model=Qwen3-32B2026.03 | 81 | |
| AutoDefenseAttack Type=Group Exclusion, Model=Qwen3-32B2026.03 | 77.4 | |
| AutoDefenseAttack Type=Targeted Suppression, Model=Deepseek-V3.12026.03 | 76.6 | |
| AutoDefenseAttack Type=False Consensus, Model=Deepseek-V3.12026.03 | 70.8 | |
| AutoDefenseAttack Type=Targeted Suppression, Model=Qwen3-235B-A22B2026.03 | 69.4 | |
| AutoDefenseAttack Type=False Consensus, Model=Qwen3-32B2026.03 | 67.8 | |
| AutoDefenseAttack Type=False Consensus, Model=Qwen3-235B-A22B2026.03 | 67.2 | |
| G-safeguardAttack Type=False Consensus, Model=Qwen3-235B-A22B2026.03 | 67 | |
| GroupGuardAttack Type=False Consensus, Model=Qwen3-235B-A22B2026.03 | 66 | |
| AutoDefenseAttack Type=Group Exclusion, Model=Qwen3-235B-A22B2026.03 | 63.6 | |
| GroupGuardAttack Type=Targeted Suppression, Model=Qwen3-32B2026.03 | 63 | |
| GroupGuardAttack Type=Targeted Suppression, Model=Qwen3-235B-A22B2026.03 | 62 | |
| G-safeguardAttack Type=Targeted Suppression, Model=Qwen3-235B-A22B2026.03 | 61.8 | |
| G-safeguardAttack Type=Targeted Suppression, Model=Qwen3-32B2026.03 | 61.2 | |
| AutoDefenseAttack Type=Targeted Suppression, Model=Qwen3-32B2026.03 | 53.6 |