skilledu/MaLLM-Bench
收藏资源简介:
马来西亚考试基准数据集是一个来自马来西亚国家考试的题目集合,旨在评估大语言模型在理解马来语及其文化细微差别方面的表现。该数据集通过融入文化相关和上下文丰富的内容,填补了现有基准(如MMLU及其翻译)的空白。数据集包括小学和中学考试题目,如马来西亚教育文凭(SPM)、初中评估考试(PMR)、初中三评估(PT3)和小学成就考试(UPSR)。学科涵盖多个领域,如历史、地理和伊斯兰研究,使其成为评估LLM在需要语言理解和文化相关性任务上的全面资源。
The Malaysian Exam Benchmark Dataset is a collection of questions derived from Malaysia’s national examinations, designed to evaluate the performance of Large Language Models (LLMs) in understanding the Malay language and its cultural nuances. This dataset addresses gaps in existing benchmarks like MMLU and its translations by incorporating culturally relevant and contextually rich content. The dataset includes questions from primary and secondary school exams such as Sijil Pelajaran Malaysia (SPM), Penilaian Menengah Rendah (PMR), Pentaksiran Tingkatan Tiga (PT3), Ujian Pencapaian Sekolah Rendah (UPSR). The subjects span various areas of study, such as history, geography, and Islamic study, making it a well-rounded resource for evaluating LLMs on tasks that require both language understanding and cultural relevance.




