mteb/JQaRARerankingLite

JQaRARerankingLite是一个用于日语问答与检索增强（Japanese Question Answering with Retrieval Augmentation）的数据集，属于MTEB（Massive Text Embedding Benchmark）的一部分。该数据集包含来自JAQKET的问题和日本维基百科的语料库，是一个轻量级版本，语料库经过缩减（172,897篇文档），并使用了5个高性能模型的硬负样本构建。数据集的任务类别包括文本排序、多项选择问答和问答，语言为日语，许可证为cc-by-sa-4.0。数据集包含四个配置：corpus、qrels、queries和top_ranked，每个配置都有详细的特征和分割信息。数据集来源于sbintuitions/JMTEB-lite。

JQaRARerankingLite is a reranking dataset for Japanese Question Answering with Retrieval Augmentation (JQaRA), part of the Massive Text Embedding Benchmark (MTEB). It consists of questions from JAQKET and corpus from Japanese Wikipedia. This is the lightweight version with a reduced corpus (172,897 documents) constructed using hard negatives from 5 high-performance models. The datasets task categories include text-ranking, multiple-choice-qa, and question-answering. The language is Japanese, and the license is cc-by-sa-4.0. The dataset includes four configurations: corpus, qrels, queries, and top_ranked, each with detailed features and split information. The dataset is derived from sbintuitions/JMTEB-lite.