EIC-Math

该数据集旨在评估数学推理中的错误识别与纠正任务，包含了错误解答、错误步骤、错误类型以及正确答案。该数据集涵盖了九种常见错误类型，并构建自现有的GSM8K和MathQA数据集。规模上，总共包含了1,800个案例，用于错误识别与纠正任务，每种错误类型均有100个案例。该数据集的任务是错误识别与修正。

This dataset is designed to evaluate the error identification and correction task in mathematical reasoning. It contains incorrect solutions, erroneous reasoning steps, error types, and the correct answers. Covering nine common error types, this dataset is constructed based on the existing GSM8K and MathQA datasets. In terms of scale, it includes a total of 1,800 cases for the error identification and correction task, with 100 cases for each error type. The targeted task of this dataset is error identification and correction.