Mathador-LM

该数据集是一个动态的基准，旨在评估大型语言模型在数学推理能力方面的表现，其灵感来源于Mathador游戏，目标是通过使用基本算术运算来达到一个给定的目标数字。为了缓解对测试集泄露的担忧，该数据集是动态生成的，并且根据难度分为简单、中等和困难的问题集。该数据集的任务是数学推理和问题解决。

This dataset is a dynamic benchmark designed to assess the mathematical reasoning capabilities of large language models (LLMs), inspired by the Mathador game. Its core objective is to reach a specified target number using fundamental arithmetic operations. To mitigate concerns over test set leakage, this dataset is dynamically generated, and the problems are categorized into three difficulty tiers: easy, medium, and hard. The task of this dataset centers on mathematical reasoning and problem-solving.