bh2821/LightNovel5000

这是一个包含约1500部中文轻小说的数据集，从未禁止爬虫的公共网站爬取，包括带插图的PDF和txt文本文件。这些轻小说可以用于训练具有风格化特征的大型语言模型。数据集的作者已经对文本进行了部分清理，但不保证文本完全无杂质。数据集的文件结构包括加密的PDF和txt文件、合并文本文件、索引-标题映射文件和索引-标题-描述映射文件。

This dataset contains approximately 1500 Chinese light novels crawled from public websites that do not prohibit crawlers, including illustrated PDFs and txt text files. These light novels can be used to train large language models with stylized features. The datasets author has partially cleaned the text, but does not guarantee that the text is completely free of impurities. The file structure of the dataset includes encrypted PDF and txt files, combined text files, index-title mapping files, and index-title-description mapping files.