Dans-DiscountModels/gutenberg-dpo-v0.1

Gutenberg DPO数据集是一个用于增强大型语言模型小说创作能力的集合，它使用了来自Project Gutenberg的公共领域书籍。数据集通过将书籍解析为章节，并使用LLM生成与这些章节相对应的合成提示来构建。每个章节都会生成一个摘要，以便在生成下一章节的提示时提供上下文。该数据集包含了人类编写的章节和LLM基于提示和前章节摘要创作的章节。

The Gutenberg DPO dataset is a collection designed to enhance the novel writing capabilities of large language models by using public domain books from Project Gutenberg. The dataset is constructed by parsing books into chapters and using LLMs to generate synthetic prompts corresponding to those chapters. Each chapter is also provided with a summary for context when prompting the next chapter. The dataset includes both human-written chapters and chapters written by LLMs based on the prompts and summaries of the previous chapters.