Distributed Misuse Detection on DM-Cyber 20x
84.2Trace-level APNaive Agent
Evaluation Results
| Method | Links | |
|---|---|---|
| Naive AgentModel=GPT-4o2026.04 | 84.2 | |
| MeerkatModel=Qwen2.5-72B2026.04 | 83.7 | |
| Naive AgentModel=Qwen2.5-72B2026.04 | 63.2 | |
| MeerkatModel=GPT-4o2026.04 | 44.5 | |
| BayesianModel=GPT-4o2026.04 | 33.2 | |
| BayesianModel=Qwen2.5-72B2026.04 | 24.3 | |
| BufferModel=GPT-4o2026.04 | 22.9 | |
| MonitorModel=GPT-4o2026.04 | 19.9 | |
| BufferModel=Qwen2.5-72B2026.04 | 19.1 | |
| MonitorModel=Qwen2.5-72B2026.04 | 12.5 |