NaiveNeuron/SMESum

SMESum是一个斯洛伐克新闻摘要语料库，包含来自SME新闻门户网站的文章。每个实例都提供完整的文章（document），以及两个简短的摘要字段（title和introduction），这两个字段可以连接起来形成黄金摘要。该语料库被分割成训练/验证/测试分区，大小分别为64,001/8,001/8,001，使用每个文件名的盐化SHA-256哈希来保证可重现性。支持的任务包括斯洛伐克新闻文章的抽象或提取式摘要，以及内容的分类。

SMESum is a Slovak news summarization corpus containing articles from the SME news portal. Each example includes the full article (document) along with two short abstract fields (title and introduction) that can be concatenated to form the golden summary. The corpus is split into train/validation/test partitions of sizes 64,001/8,001/8,001 using a salted SHA-256 hash of each filename to ensure reproducibility. Supported tasks include abstractive or extractive summarization of Slovak news articles and content classification.