JetBrains-Research/commit-msg-edits

该数据集是通过Commit Message Editing应用收集的专家标注的提交信息编辑集合。标注者被展示了由GPT-4生成的15个提交信息，并被要求手动编辑这些信息，使其质量足够好以提交到版本控制系统。数据集的结构包括每个数据点的字段，如仓库名称、提交哈希、会话标识符、初始提交信息、最终提交信息、提交信息编辑历史等。原始数据点也以类似的结构存储在`raw`文件夹中。

This dataset contains a collection of expert-labeled commit message edits contributed via the Commit Message Editing app. Labelers were presented with GPT-4 generated messages for 15 commits from the CMG benchmark and asked to manually edit them to be of good enough quality to submit to VCS. The dataset structure includes fields such as repository name, commit hash, session identifier, initial commit message, final commit message, a list of character-level edits made by the labeler, and timestamps. Additionally, raw datapoints collected through the app are also provided.