matrixorigin/parsing_bench

ParsingBench是一个文档解析基准测试数据集，用于评估模型在图像文本到文本任务上的性能。它涵盖了多种数据源类型、页面布局、语言类型、表格属性和文本属性，以全面评估模型在文档解析方面的能力。

ParsingBench is a document parsing benchmark dataset designed to evaluate the performance of models on image-text-to-text tasks. It includes various data source types, page layouts, language types, table attributes, and text attributes to comprehensively assess the models capabilities in document parsing.