nanonets/small_sparse_structured_table

这是一个合成的表格数据集，专为表格问题回答任务设计，包含稀疏的单元格（40%至70%的空单元格），清晰的行列分隔，以及小型表格（4至10行，2至6列）。数据集支持英文，并且与光学字符识别（OCR）、表格和智能文档处理（IDP）相关。数据集规模小于1000个样本。This is a synthetic table dataset designed for table question answering tasks, containing sparse cells (40% to 70% empty cells), clear row and column separators, and small tables (4 to 10 rows, 2 to 6 columns). The dataset supports English and is related to Optical Character Recognition (OCR), Tables, and Intelligent Document Processing (IDP). The dataset size is less than 1000 samples.