MSVAMP
收藏arXiv2025-09-30 收录
下载链接：
https://mathoctopus.github.io
下载链接
链接失效反馈官方服务：
资源简介：
该数据集名为mSVAMP，是一个用于评估多语言数学推理能力的域外测试集。此外，mSVAMP也用于评估SLAM的泛化能力，证明了其在多种语言中均展现出稳健的性能。该数据集所涉及的任务是跨语言数学推理。
This dataset, named mSVAMP, is an out-of-distribution test set for evaluating multilingual mathematical reasoning capabilities. Additionally, mSVAMP is also utilized to assess the generalization ability of SLAM, demonstrating its robust performance across multiple languages. The task involved in this dataset is cross-lingual mathematical reasoning.
