1 paper
Yujian Zhuang, Dehong Gao, Qichao Zhang +4
Large language models (LLMs) are increasingly used in binary decompilation to refine the C-like pseudocode produced by traditional rule-based decompilers. While this pseudocode is…