IrokoBench

IrokoBench是一个由Masakhane NLP创建的基准数据集，专门设计用于评估大型语言模型在16种非洲低资源语言上的表现。该数据集涵盖自然语言推理、数学推理和多选知识问答三个复杂任务。数据集通过专业翻译人员将英语评估数据集翻译成16种非洲语言，确保了数据的质量和适用性。IrokoBench的应用领域广泛，旨在解决非洲语言在人工智能领域中的代表性不足问题，推动这些语言的数字化和智能化进程。IrokoBench is a benchmark dataset created by Masakhane NLP, specifically designed to evaluate the performance of large language models (LLMs) across 16 low-resource African languages. This dataset encompasses three complex tasks: natural language inference, mathematical reasoning, and multiple-choice knowledge question answering. The dataset was developed by having professional translators translate English evaluation datasets into the 16 target African languages, ensuring the quality and applicability of the data. IrokoBench has a wide range of application scenarios, aiming to address the underrepresentation of African languages in the field of artificial intelligence and promote the digitalization and intelligentization of these languages.