theprint/RolePlaying

RolePlaying数据集是一个用于文本生成的合成数据集，通过DataMix和LMDataTools工具生成，包含22,677个JSON格式的条目。该数据集通过采样和组合Hugging Face上的多个来源（如andrijdavid/roleplay-conversation、NewEden-Forge/Gryphe-Sonnet3.5-Charcard-Roleplay-unfiltered和chimbiwide/RolePlay-NPC）创建，专注于角色扮演和AI聊天机器人训练。每个条目包含多轮对话数据（如human/gpt或system/human/gpt格式），涉及黑暗奇幻、浪漫/NSFW等主题，例如残酷的女王Kaori和迷信的Olivia Tezuka等具体角色场景。数据集适用于开发者和研究人员，用于训练和评估角色驱动交互式虚构和创意角色扮演模型。

The RolePlaying dataset is a synthetic dataset for text generation, generated using DataMix and LMDataTools, containing 22,677 entries in JSON format. It is created by sampling and combining multiple sources from Hugging Face, such as andrijdavid/roleplay-conversation, NewEden-Forge/Gryphe-Sonnet3.5-Charcard-Roleplay-unfiltered, and chimbiwide/RolePlay-NPC, with a focus on roleplaying and AI chatbot training. Each entry includes multi-turn conversational data (e.g., in human/gpt or system/human/gpt formats), covering topics like dark fantasy and romantic/NSFW themes, with specific character scenarios such as a cruel queen named Kaori and a superstitious woman named Olivia Tezuka. The dataset is most useful for developers and researchers working on training and evaluating models for character-driven interactive fiction and creative, uncensored roleplay.