ketchup123/tulu_pref_filter_only

这是一个包含会话ID、图鲁ID、来源、提示、选择内容、角色、拒绝内容、意图、知识、难度、输入质量、质量说明、质量生成器、任务类别、其他任务类别、任务类别生成器、llama_guard_2、选择指令奖励、拒绝指令奖励、奖励模型、语言等字段的数据集。它包含一个训练集，大小为1,518,248,923.9122126字节，共有210,635个示例。

This is a dataset containing fields such as conversation_id, tulu_id, source, prompt, chosen content, role, rejected content, intent, knowledge, difficulty, input quality, quality explanation, quality generator, task category, other task category, task category generator, llama_guard_2, chosen instruct reward, rejected instruct reward, reward model, language. It includes a training set that is 1,518,248,923.9122126 bytes in size and has 210,635 examples.