Code Generation on HumanEval (Attack/Defense Accuracy)
100Accuracy (Attack)Reporting-and-penalty mechanism
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Reporting-and-penalty mechanismAttack Type=All Attack Types, LLM Model=All Above Models2026.04 | 100 | 100 | |
| GroupGuardAttack Type=False Consensus, LLM Model=Qwen3-32B2026.04 | 67 | 78 | |
| GroupGuardAttack Type=False Consensus, LLM Model=Deepseek-V3.12026.04 | 59 | 100 | |
| GroupGuardAttack Type=False Consensus, LLM Model=Qwen3-235B2026.04 | 47 | 81 |