NewSHead
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/google-research-datasets/NewSHead
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含了新闻文章的标题及其人工标注的描述，旨在用于评估混合编码器在更广泛场景下的性能表现。该数据集规模较大，其任务是从大规模语料库中进行语义接近目标的选择。
This dataset consists of news article titles and their manually annotated descriptions, which is intended to evaluate the performance of hybrid encoders across a broader range of scenarios. It is a large-scale dataset, and its core task is to conduct semantic similarity-based target selection from a large-scale corpus.
提供机构：
Google
