heegyu/Magpie-Pro-DPO-200K-Filtered

该数据集是从[Magpie-Align/Magpie-Pro-DPO-200K](https://huggingface.co/datasets/Magpie-Align/Magpie-Pro-DPO-200K)中筛选出chosen_score >= 4的数据。数据集包含训练集和测试集，特征包括uuid、task_category、prompt、chosen、rejected、messages、score_chosen、score_rejected和generator。其中，chosen和rejected是列表类型，包含content和role两个字段。数据集的总下载大小为246519433字节，总数据集大小为394627164.83175字节。

The dataset includes multiple configurations, with the default configuration containing train and test data files. The features of the dataset include UUID, task category, prompt, chosen content and role, rejected content and role, messages content and role, chosen and rejected scores, and generator. The dataset is divided into train and test sets, containing 49770 and 7339 examples respectively. The download size of the dataset is 246519433 bytes, and the total size is 394627164.83175 bytes. The dataset is filtered from Magpie-Align/Magpie-Pro-DPO-200K with the condition that the chosen score is greater than or equal to 4.

原始信息汇总

数据集概述

数据集配置

• 配置名称: default
  • 数据文件:
    • 训练集: data/train-*
    • 测试集: data/test-*

数据集信息

• 特征:
  • uuid: 字符串类型
  • task_category: 字符串类型
  • prompt: 字符串类型
  • chosen:
    • content: 字符串类型
    • role: 字符串类型
  • rejected:
    • content: 字符串类型
    • role: 字符串类型
  • messages:
    • content: 字符串类型
    • role: 字符串类型
  • score_chosen: 浮点数类型
  • score_rejected: 浮点数类型
  • generator: 字符串类型

数据集分割

• 训练集:
  • 字节数: 343544331.83175
  • 样本数: 49770
• 测试集:
  • 字节数: 51082833
  • 样本数: 7339

数据集大小

• 下载大小: 246519433
• 数据集总大小: 394627164.83175

数据集来源

• 过滤条件: chosen_score >= 4
• 来源数据集: Magpie-Align/Magpie-Pro-DPO-200K