five

davanstrien/cosmochat

收藏
Hugging Face2024-05-10 更新2024-06-15 收录
下载链接:
https://hf-mirror.com/datasets/davanstrien/cosmochat
下载链接
链接失效反馈
官方服务:
资源简介:
CosmoChat数据集是一个使用distilabel库创建的合成数据集,旨在从Cosmopedia生成多轮对话数据。Cosmopedia是一个包含教科书、博客文章、故事和WikiHow文章的大型合成数据集。该数据集的目标是创建具有教学价值的对话数据,以帮助提高语言模型在对话场景中的表现。数据集包括cosmopedia_text、audience、dialogue和final_question_prompt等特征,重点关注反映现实世界对话的多轮交互。该数据集旨在帮助模型保持上下文、解决歧义并进行复杂的交互。

The CosmoChat dataset aims to create pedagogically valuable multi-turn synthetic datasets from Cosmopedia. The dataset includes features such as cosmopedia_text, audience, dialogue, and final_question_prompt. The dialogue feature contains sub-features content and role. The dataset is structured with a train split containing 200 examples. The dataset is synthetic, created using the distilabel library, and aims to improve the performance of language models in real-world conversational scenarios by focusing on multi-turn data.
提供机构:
davanstrien
原始信息汇总

数据集概述

数据集基本信息

  • 数据集名称: CosmoChat
  • 大小类别: n<1K
  • 语言: 英语
  • 许可证: CC0-1.0
  • 标签: synthetic, distilabel, rlaif

数据集结构

  • 特征:
    • cosmopedia_text: 字符串类型
    • audience: 字符串类型
    • dialogue: 列表类型,包含以下子特征:
      • content: 字符串类型
      • role: 字符串类型
    • final_question_prompt: 字符串类型
  • 分割:
    • train: 包含200个样本,占用3638063字节
  • 下载大小: 1577349字节
  • 数据集大小: 3638063字节

配置信息

  • 配置名称: default
  • 数据文件:
    • 分割: train
    • 路径: data/train-*

数据集示例

以下是一个数据集示例的结构: json { "audience": "grade school student", "cosmopedia_text": " Imagine you have a balloon shaped like a perfect sphere...", "dialogue": [ { "content": "How do you find the area of a curvy surface when its hard to measure its length and width?", "role": "user" }, { "content": "Finding the area of a curvy surface can be a bit more challenging...", "role": "system" }, { "content": "Can you explain more about how to use the sector area formula, like how to find the central angle and radius if Im given a picture of a circle with a pie-shaped region shaded in?", "role": "user" }, { "content": "The sector area formula is a great way to find the area of a pie-shaped region...", "role": "system" }, { "content": "How can you use the sector area formula to find the area of a sector that is not a complete pie-shaped region, but rather a sector with a curved edge that doesnt follow a perfect circle?", "role": "user" }, { "content": "The sector area formula I provided earlier, A = (θ/360) × π × r^2, assumes that the sector is a perfect pie-shaped region...", "role": "system" } ], "final_question_prompt": "Based on the conversation below, generate a thoughtful and engaging final question that a user could ask to further the discussion..." }

5,000+
优质数据集
54 个
任务类型
进入经典数据集
二维码
社区交流群

面向社区/商业的数据集话题

二维码
科研交流群

面向高校/科研机构的开源数据集话题

数据驱动未来

携手共赢发展

商业合作