ellamind/summary_texts_new_extract

这是一个经过分层抽样的文本摘要数据集子集，它保证了不同长度文本的平衡代表性，特别关注了较长文本的覆盖。该子集从原始数据集ellamind/summary_texts_new的_train部分中提取，并保持了所有原始字段。

This is a stratified subset of the summary texts dataset, ensuring balanced representation across different text lengths with a focus on coverage of longer texts. The subset is extracted from the train split of the original dataset ellamind/summary_texts_new and maintains all original fields.