HeSum

该数据集名为HeSum，包含了来自希伯来语新闻网站的1万个文章摘要对，专门为现代希伯来语的抽象文本摘要而设计。HeSum数据集展现出高度的抽象性、低冗余度以及具有挑战性的压缩比，使其成为评估希伯来语生成语言技术的宝贵基准。该数据集的规模为1万个文章摘要对，针对的任务是抽象文本摘要。

This dataset, named HeSum, contains 10,000 article-summary pairs sourced from Hebrew news websites, and is specifically developed for abstractive text summarization in modern Hebrew. The HeSum dataset features high abstractiveness, low redundancy, and challenging compression ratios, making it a valuable benchmark for evaluating generative language technologies for the Hebrew language. It has a total of 10,000 article-summary pairs and targets the abstractive text summarization task.