datalama/question_pair
收藏资源简介：
该数据集是基于songys公开的Question_pair数据集，经过一些数据修改后上传的。修改内容包括将`is_duplicate`字段重命名为`label`，将测试集中的`test_id`重命名为`id`，并将标签值从原来的{같은 질문: 0, 다른 질문: 1}反转为{같은 질문: 1, 다른 질문: 0}。最终数据集包含id, question1, question2, label四个字段。数据集分为训练集和测试集，训练集包含6888个样本，测试集包含688个样本。
This dataset is uploaded after applying several modifications to the publicly available Question_pair dataset released by songys. The modifications include renaming the `is_duplicate` field to `label`, renaming the `test_id` field in the test split to `id`, and reversing the original label mapping from {같은 질문: 0, 다른 질문: 1} to {같은 질문: 1, 다른 질문: 0}. The final dataset contains four fields: id, question1, question2, and label. The dataset is split into a training set and a test set, with 6888 samples in the training set and 688 samples in the test set.
数据集概述
数据集特征
- id: 整数类型 (int64)
- question1: 字符串类型 (string)
- question2: 字符串类型 (string)
- label: 分类标签，包含两个类别：
- 0: 다른 질문 (不同问题)
- 1: 같은 질문 (相同问题)
数据集划分
- 训练集 (train):
- 样本数量: 6888
- 数据大小: 634021 字节
- 测试集 (test):
- 样本数量: 688
- 数据大小: 62628 字节
数据集大小
- 下载大小: 403049 字节
- 数据集总大小: 696649 字节
