MathArena/kangaroo_2025

Kangaroo 2025数据集包含了2025年MathArena数学竞赛的问题，用于评估大型语言模型在未受污染的数学竞赛中的表现。数据集包括问题的索引、以图片形式呈现的问题陈述、问题的答案以及竞赛的名称。数据集根据竞赛题目来源进行了提取、翻译和截图，并分为训练集。数据集遵循知识共享署名-非商业性使用-相同方式共享4.0国际许可。

The Kangaroo 2025 dataset from 2025 includes questions from the MathArena mathematical competition used to evaluate the performance of large language models in uncontaminated mathematical competitions. The dataset consists of problem indices, problem statements as images, answers to the questions, and competition names. The dataset is based on competition questions that have been extracted, translated, and screenshot, and is split into a training set. The dataset follows the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International license.