Multi-stage Attack Detection on CSTM-Bench (dilution shard)
100Detection RatePer-session judge
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Per-session judgeReader=Per-session judge2026.04 | 100 | 100 | 72.7 | 80.9 | 21.4 | 64.3 | |
| Coreset Memory ReaderReader=Coreset Memory Reader, Model=Claude Sonnet 4.6, Coreset size=K=502026.04 | 96.2 | 96.2 | 92.6 | 77.5 | 0 | 21.4 | |
| Full-Log CorrelatorReader=Full-Log Correlator, Model=Claude Sonnet 4.62026.04 | 88.5 | 88.5 | 81.2 | 86.8 | 0 | 50 |