manikineko/humanizer

该数据集旨在创建具有自然、类似人类对话能力的 LLM，避免过于正式或由 AI 生成的语气。数据来源包括个人 Discord 数据、公共领域文本、加密货币相关平台的公共聊天记录以及数据集创建者编写的网站内容。This dataset is designed to create LLMs with natural, human-like conversational abilities, avoiding overly formal or AI-generated tones. The sources include personal Discord data, public domain texts, public chatlogs from cryptocurrency-related platforms, and website content authored by the dataset creator.