kraalfar/CoEdPilot-generator-processed

该数据集是[code-philia/CoEdPilot-generator](https://huggingface.co/datasets/code-philia/CoEdPilot-generator)的预处理版本，用于下一个编辑预测任务。数据集包含代码修改前后的信息、编辑历史、下一个编辑、标签窗口、指令、提交URL和文件名等字段。数据集分为训练集、验证集和测试集，分别包含2633966、376282和752565个示例。

This is a processed version of the dataset specifically for the next edit prediction task. The dataset includes fields such as code_before (initial code to be modified), edits (previous edits), next_edit (next edit to be done), code_after (code after the edit), label_window (labels for each line in the code before, including keep, add, and replace), instruction (presumably commit message), html_url (commit URL), and file_name (changed file). The dataset is divided into training, validation, and test sets, containing 2,633,966, 376,282, and 752,565 samples respectively.