Culturally Adapted GSM8K
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/akarim23131/Lost_in_Cultural_Translation
下载链接
链接失效反馈官方服务：
资源简介：
该数据集由六个合成数据集组成，这些数据集基于GSM8K基准生成，通过替换数学问题中的文化实体，同时保留了原始的数学逻辑。该数据集是通过在原始GSM8K数据集中系统地替换文化实体而创建的，确保了问题的数学完整性得到保持。该数据集包含了1,319个经过文化适应的题目，其任务是评估语言模型在文化适应情境下的数学推理能力。
This dataset consists of six synthetic datasets, each of which is generated from the GSM8K benchmark by replacing cultural entities within their mathematical problems while preserving the original mathematical logic. It is created by systematically replacing cultural entities in the original GSM8K dataset, ensuring that the mathematical integrity of the problems is maintained. This dataset contains 1,319 culturally adapted questions, which are designed to evaluate the mathematical reasoning capabilities of language models in culturally adapted scenarios.
