CommitChronicle
CommitChronicle is a large-scale, diverse multilingual dataset for commit message generation and completion: more than 10 million carefully filtered commits. We used GitHub Search tool and official GitHub API to select relevant repositories with permissive licenses (Apache, BSD 3-clause, MIT). On February 9th, 2023, we collected all commits made since 2017 from these repositories via PyDriller. Next, we extensively cleaned the data, including filtering outliers, dropping commits from bot authors, and dropping duplicates. Note: to avoid disclosing personal information, we replaced the commit authors' names and emails with unique identifiers. Data Paper Paper pre-print 