Tele-AI/TeleTableBench
收藏资源简介:
T2RBench是一个面向真实工业场景的中英双语“表格生成报告”基准数据集。它涵盖了汽车、能源、金融、政务、财务、科技、教育、银行等垂直领域,包含6个一级大类和19个二级业务小类。该数据集是目前覆盖工业级表格类别最全的数据集,包括单表多sheet、多表多sheet、复杂结构表、超大宽表等复杂表格场景。数据集旨在服务于表格生成报告任务,并提供了人工校对的高质量问题和报告关键点标注。
T2RBench is a bilingual (Chinese and English) benchmark dataset for Table-to-Report generation in real-world industrial scenarios. It covers vertical domains such as automotive, energy, finance, government affairs, finance, technology, education, and banking, with 6 first-level categories and 19 second-level business subcategories. This dataset is currently the most comprehensive in terms of industrial table categories, including complex scenarios like single-table multi-sheet, multi-table multi-sheet, complex structured tables, and super wide tables. The dataset is designed for the table-to-report generation task and provides manually verified high-quality questions and key point annotations for reports.




