Vulnerability Detection On Vulscriber
Metrics
F1 Score
Results
Performance results of various models on this benchmark
| Paper Title | Code | ||
|---|---|---|---|
| Reveal Model - Tested on Reveal (Training on Devign + VulScribeR 20K + Extra Cleans) | 26.18 | VulScribeR: Exploring RAG-based Vulnerability Augmentation with LLMs | |
| Devign Model - Tested on Reveal (Training on Devign + VulScribeR 20K + Extra Cleans) | 24.99 | - | |
| Reveal Model - Tested on Bigvul (Training on Devign + VulScribeR 20K + Extra Cleans) | 18.98 | - | |
| Devign Model - Tested on Bigvul (Training on Devign + VulScribeR 20K + Extra Cleans) | 18.51 | - | |
| LineVul - Tested on Reveal (Training on Devign + VulScribeR 20K + Extra Cleans) | 17.38 | - | |
| LineVul - Tested on BigVul (Training on Devign + VulScribeR 20K+ Extra Cleans) | 16.23 | - |
0 of 6 row(s) selected.