QuasarResearch/apollo-preview-v0.2

Apollo是一个RP/ERP数据集，通过结合角色扮演、创意写作和指令跟随数据生成。此版本为预览版，未来版本计划包含自有数据生成管道。数据集使用了Claude 3 Opus、Claude 3.5 Sonnet、GPT-4o、GPT-4等语言模型生成，并经过去重和无审查处理。数据来源包括roleplay4fun/aesir-v1.1、grimulkan/LimaRP-augmented等多个公开数据集。

Apollo is a RP/ERP dataset, which has been generated by amalgamating several data sources with roleplay, creative writing, and general instruction following data. The dataset includes features such as conversations (with sender and content), source, and index. It is divided into a training set with 15426 samples. The dataset is generated using various language models like Claude 3 Opus, Claude 3.5 Sonnet, GPT-4o, etc. The dataset is deduplicated and uncensored.