VulnScout-C: A Lightweight Transformer for C Code Vulnerability Detection
Aymen Lassoued 1, Nacef Mbarek 1, Bechir Dardouri 1, Bassem Ouni 2, Qing Li 3, Fakhri Karray 4 · 2026
Read VulnScout-C: A Lightweight Transformer for C Code Vulnerability Detection by Aymen Lassoued 1, Nacef Mbarek 1, Bechir Dardouri 1, Bassem Ouni 2, Qing Li 3, Fakhri Karray 4 (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-28309.txt.
Description
Lassoued and five co-authors build a compact transformer that hunts security flaws in C code. Giant language models detect flaws well but cost too much for constant use in coding workflows. Their VulnScout-C has 693 million parameters, with only 353 million active per run. They also release a verified dataset of 33,565 samples that covers rare weakness types. On a standard benchmark, the small model beat all rivals, including big reasoning models and commercial scanners.
Catalog record
| CREATOR | Aymen Lassoued 1, Nacef Mbarek 1, Bechir Dardouri 1, Bassem Ouni 2, Qing Li 3, Fakhri Karray 4 |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 77145 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.28309