aisingapore/NLG-Abstractive-Summarization

SEA Abstractive Summarization是一个多语言抽象总结评估数据集，用于评估模型在阅读文档、识别关键点并将其总结成连贯流畅的文本的同时进行释义的能力。数据集包括印尼语、泰米尔语、泰语和越南语四种语言，从XL-Sum数据集中采样得到，并包含额外的少量示例分割。数据集的结构包括文档ID、标签、提示文本、提示模板和元数据等字段。

SEA Abstractive Summarization is a multilingual abstractive summarization evaluation dataset designed to assess a models ability to read a document, identify the key points, and summarize them into a coherent and fluent text while paraphrasing. It includes four languages: Indonesian, Tamil, Thai, and Vietnamese, sampled from the XL-Sum dataset, and contains additional few-shot example splits. The dataset structure includes fields for document ID, label, prompt text, prompt templates, and metadata.