rootsautomation/MultiHiertt

MultiHiertt数据集是一个用于在包含多个层次化财务表格的文档上进行数值推理的数据集。它基于FinTabNet构建，从S&P 500年度报告（1999-2019年）中提取了4,791份多页文档。每个文档包含2-6个层次化HTML表格和周围文本。问题需要跨多个表格和/或文本段落进行推理，答案通过算术程序或直接跨度选择得出。数据集包含训练集（7,830个示例）和验证集（1,044个示例），并提供了丰富的字段，如唯一标识符、段落、表格、问题、答案、程序、文本证据和表格证据等。

MultiHiertt is a dataset for numerical reasoning over documents containing multiple hierarchical financial tables. It is built on FinTabNet, extracting 4,791 multi-page documents from S&P 500 annual reports (1999-2019). Each document contains 2-6 hierarchical HTML tables and surrounding text. Questions require reasoning across multiple tables and/or text passages, with answers derived via arithmetic programs or direct span selection. The dataset includes a training set (7,830 examples) and a validation set (1,044 examples), with fields such as unique ID, paragraphs, tables, question, answer, program, text evidence, and table evidence.