reasonir/reasonir-data
收藏Hugging Face2025-05-07 更新2025-07-05 收录
下载链接：
https://hf-mirror.com/datasets/reasonir/reasonir-data
下载链接
链接失效反馈官方服务：
资源简介：
ReasonIR数据集包含用于训练ReasonIR-8B模型的合成示例。数据集分为两种配置：HQ（困难查询）和VL（可变长度）。HQ数据包含查询、正文档标识符和负文档，而VL数据包含查询、正文档和负文档的完整数据。由于法律原因，HQ数据无法重新托管原始的正文档，因此提供了一个数据处理脚本来下载和合并它们。数据集根据cc-by-nc-4.0许可证授权。
The ReasonIR dataset contains synthetic examples used to train the ReasonIR-8B model. It comes in two configurations: HQ (hard-query) and VL (varied-length). The HQ data includes queries, positive document identifiers, and negative documents, while the VL data provides the full set with queries, positive documents, and negative documents. Due to legal reasons, the original positive documents for the HQ data cannot be rehosted and a data processing script is provided to download and merge them. The dataset is licensed under cc-by-nc-4.0.
提供机构：
reasonir
