HQCMD - The dataset of the paper: \IEEF: Enhancing Commit Message Generation by Segmenting Key Information in High-quality Data\

Inspired by ensemble learning, we took the intersection of the results from LLM and a deep-learning-based classifier as the final output, thereby enhancing the reliability of the data. As a result, we successfully constructed a high-quality commit message dataset with 300,000 samples from GitHub (called the HQCMD).