Small Language Models for Smart Contract Vulnerability Detection: A Fine-Grained Analysis on Capabilities, Localization, and Failure Modes
Published in The 32nd IEEE International Conference on Parallel and Distributed Systems (ICPADS 2026), 2026
This paper presents a fine-grained analysis of small language models for smart-contract vulnerability detection, with particular attention to model capabilities, vulnerability localization, and common failure modes.
Authors: Doan Minh Trung, Hoang Cao Phong, Luc Vinh Kiet, Ho Minh Tri, Nguyen Nhat Dong, Phan The Duy, Tu-Anh Nguyen-Hoang, and Van-Hau Pham
Status: Accepted full paper
Conference: The 32nd IEEE International Conference on Parallel and Distributed Systems (ICPADS 2026)
Location: Tokyo, Japan
Ranking: CORE Rank B
Paper and supplementary links will be added when they become publicly available.
Recommended citation: Doan Minh Trung, Hoang Cao Phong, Luc Vinh Kiet, Ho Minh Tri, Nguyen Nhat Dong, Phan The Duy, Tu-Anh Nguyen-Hoang, and Van-Hau Pham. (2026). "Small Language Models for Smart Contract Vulnerability Detection: A Fine-Grained Analysis on Capabilities, Localization, and Failure Modes." The 32nd IEEE International Conference on Parallel and Distributed Systems (ICPADS 2026). Accepted full paper.