manu/fquad2_test

该数据集是FrenchBench项目的一部分，用于法语语言模型的评估。数据集包含标题、上下文、问题、答案和是否不可能回答等特征。数据集分为测试集和验证集，每个集又分为有答案和无答案的子集。适用于抽取式问答、二元分类或信息检索评估。

This dataset is part of the FrenchBench project, designed for the evaluation of French language models. It comprises features such as title, context, question, answer, and a flag indicating whether the question is impossible to answer. The dataset is split into a test set and a validation set, each of which is further divided into subsets with valid answers and those without valid answers. This dataset is suitable for extractive question answering, binary classification, or information retrieval evaluation.

原始信息汇总

数据集概述

数据集信息

特征

• title: 字符串类型
• context: 字符串类型
• question: 字符串类型
• answers: 结构体类型，包含以下字段：
  • answers_start: 整数序列
  • text: 字符串序列
• is_impossible: 布尔类型

数据分割

• test: 865505字节，800个样本
• valid: 217746字节，200个样本
• test_hasAns: 458114字节，400个样本
• valid_hasAns: 113725字节，100个样本

大小

• 下载大小: 785547字节
• 数据集大小: 1655090字节

许可

• license: apache-2.0

任务类别

• question-answering
• feature-extraction
• sentence-similarity

语言

• fr

大小类别

• n<1K