jaeyong2/ja-rag-cot
收藏Hugging Face2024-11-01 更新2024-12-14 收录
下载链接：
https://hf-mirror.com/datasets/jaeyong2/ja-rag-cot
下载链接
链接失效反馈官方服务：
资源简介：
该数据集是一个日语问答数据集，包含上下文、问题、原始真实答案、思考过程和最终答案等特征。数据集来源于日语维基百科，并使用Qwen/Qwen2-72B-Instruct模型生成带有COT（Chain of Thought）的答案。数据集包含209,496个训练样本，总大小为1,642,812,120字节，下载大小为936,747,874字节。
This dataset contains Japanese text, primarily used for training and evaluating context-based question answering models. The dataset features include context, question, RAW Ground Truth, thinking process, and final answer. The dataset is divided into a training set containing 209496 samples. During the development process, the dataset sourced from range3/wikipedia-ja-20230101 and used the Qwen/Qwen2-72B-Instruct model to generate answers with a thinking process.
提供机构：
jaeyong2
