1. Introduction
In recent years, applying machine learning (ML) in the domain of big code (ML4Code) [1] has gained significant attention. ML4Code leverages the power of ML, especially deep learning (DL), to extract patterns from large code corpora. The remarkable performance of ML4Code in various downstream code tasks, such as clone detection [2], bug detection [3], and problem classification [4], demonstrate its immense potential in facilitating developers in daily activity.