if001/auto-wiki-select-qa

该数据集是一个日语问答数据集，基于cl-nagoya/auto-wiki-qa-nemotron数据集构建，使用Qwen/Qwen3-32B模型自动生成了三个错误答案。每个数据点包含唯一标识符row_id、查询问题query、正确答案answer以及三个错误答案wrong_answer_1、wrong_answer_2和wrong_answer_3。数据集适用于问答系统训练、错误答案检测或模型评估任务，特别针对日语语言处理。数据格式为文本，训练集包含19,864个示例，总大小约4.5 MB，采用CC BY-SA 4.0许可证。

This dataset is a Japanese question-answering dataset based on cl-nagoya/auto-wiki-qa-nemotron, with three wrong answers automatically generated using the Qwen/Qwen3-32B model. Each data point includes a unique identifier row_id, a query question, a correct answer, and three wrong answers (wrong_answer_1, wrong_answer_2, wrong_answer_3). It is suitable for QA system training, wrong answer detection, or model evaluation tasks, specifically for Japanese language processing. The data is in text format, with a training set of 19,864 examples, total size approximately 4.5 MB, and licensed under CC BY-SA 4.0.