WikiTableEdit

WikiTableEdit是由北京大学创建的一个高质量表格编辑数据集，包含超过200,000个实例，用于评估大型语言模型在表格编辑任务中的表现。该数据集利用WikiSQL数据集中的26,531个表格，通过自动生成自然语言指令和相应的操作结果来构建。数据集涵盖了六种基本的表格编辑操作，包括添加、删除、交换、重新排序、合并和拆分，适用于常规和非常规表格的编辑。WikiTableEdit旨在通过自然语言指令简化表格编辑过程，特别是对于非专业用户，无需编写代码即可直接编辑表格。此数据集的应用领域广泛，可用于提升语言模型在处理复杂表格数据时的性能，解决实际场景中表格编辑的难题。

WikiTableEdit is a high-quality table editing dataset developed by Peking University, which contains more than 200,000 instances for evaluating the performance of large language models (LLMs) on table editing tasks. This dataset is constructed using 26,531 tables from the WikiSQL dataset by automatically generating natural language instructions and their corresponding operation results. It covers six fundamental table editing operations, namely add, delete, swap, reorder, merge, and split, and supports editing of both conventional and unconventional tables. WikiTableEdit aims to simplify the table editing workflow through natural language instructions, enabling non-expert users to directly edit tables without writing any code. This dataset has broad application prospects, which can be used to enhance the performance of language models when handling complex tabular data and address the challenges of table editing in real-world scenarios.