anthj/dpo_mw_new10
收藏Hugging Face2024-07-20 更新2024-07-22 收录
下载链接:
https://hf-mirror.com/datasets/anthj/dpo_mw_new10
下载链接
链接失效反馈官方服务:
资源简介:
该数据集包含两个主要分割:训练集和评估集。训练集包含19534个样本,评估集包含2171个样本。每个样本包含三个主要特征:问题(question)、选择的答案(chosen)和被拒绝的答案(rejected)。这些特征均为字符串类型,可能用于问答系统的训练和评估,特别是涉及答案选择或排序的场景。
The dataset includes two main splits: a training set and an evaluation set. The training set contains 19,534 samples, and the evaluation set contains 2,171 samples. Each sample consists of three main features: a question, a chosen answer, and a rejected answer, all of which are of string type. This dataset is likely used for training and evaluating question-answering systems, particularly in scenarios involving answer selection or ranking.
提供机构:
anthj
原始信息汇总
数据集概述
数据集特征
- question: 类型为字符串 (string)
- chosen: 类型为字符串 (string)
- rejected: 类型为字符串 (string)
- index_level_0: 类型为整数 (int64)
数据集分割
- train:
- 字节数: 19363936
- 样本数: 19534
- eval:
- 字节数: 2147796
- 样本数: 2171
数据集大小
- 下载大小: 6279857 字节
- 数据集总大小: 21511732 字节
配置
- 配置名称: default
- 数据文件:
- train: data/train-*
- eval: data/eval-*
- 数据文件:



