MCLM
收藏资源简介:
MCLM(多语言竞赛级别数学)是一个设计用于评估多语言环境下高级数学推理能力的基准。该数据集包含了55种语言的竞赛级别数学问题,不仅包括标准的文字问题,还挑战了最先进的大型语言模型。数据集由两部分构成:机器翻译推理问题和人工注释推理问题。机器翻译推理问题是从MATH-500和AIME 2024基准中衍生出来的,而人工注释推理问题则包括IMO(国际数学奥林匹克)问题的官方翻译和国内及地区数学奥林匹克问题。
MCLM (Multilingual Competition-Level Mathematics) is a benchmark designed to evaluate advanced mathematical reasoning capabilities in multilingual settings. This dataset includes competition-level mathematical problems across 55 languages, covering both standard word problems and posing challenging tests for state-of-the-art large language models. The dataset is composed of two parts: machine-translated reasoning problems and manually annotated reasoning problems. The machine-translated reasoning problems are derived from the MATH-500 and AIME 2024 benchmarks, while the manually annotated reasoning problems consist of official translations of International Mathematical Olympiad (IMO) problems and domestic and regional mathematical Olympiad problems.




