McGill-NLP/mgsm-pro

该数据集名为MGSM-Pro，是一个多语言小学数学数据集，涵盖9种语言（阿姆哈拉语、中文、英语、法语、伊博语、日语、斯瓦希里语、特威语和约鲁巴语）。数据集包含两个版本：symbolic和ic。两者都用10种独特的变体替换了原始名称和数字。此外，IC配置在每个问题中添加了一条无关的上下文句子。每个语言版本的数据集都有2250个示例。

This dataset is named MGSM-Pro and is a multilingual grade school math dataset covering 9 languages (Amharic, Chinese, English, French, Igbo, Japanese, Swahili, Twi, and Yoruba). The dataset consists of two versions: symbolic and ic. Both substitute the original names and numbers with 10 unique variations. Additionally, the IC configuration adds a piece of irrelevant context sentence to each problem. Each language version of the dataset contains 2250 examples.