Malicious Code Generation on Malicious Code Generation Dataset (test)
19.38ASR (Claude3.5)SPELL
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| SPELL2025.12 | 19.38 | 83.75 | 68.13 | 57.09 | |
| Redcode2025.12 | 18.13 | 91.88 | 10.63 | 40.21 | |
| Random Selection2025.12 | 16.88 | 53.13 | 60.63 | 43.54 | |
| CL-GSO2025.12 | 7.5 | 5.63 | 5.63 | 6.25 | |
| RL-breaker2025.12 | 0 | 0 | 27.5 | 8.33 | |
| DRA2025.12 | 0 | 0 | 0 | 0 | |
| CodeAttack2025.12 | 0 | 0 | 0 | 0 |