ArabicMMLU: Assessing Massive Multitask Language Understanding in Arabic
This addresses the problem of limited Arabic evaluation datasets for researchers and developers, though it is incremental as it adapts an existing benchmark format to a new language.
The paper tackles the lack of evaluation benchmarks for Arabic language models by introducing ArabicMMLU, a multi-task benchmark with 40 tasks and 14,575 questions, and finds that top models like BLOOMZ and LLaMA2 score below 50%, with the best Arabic-centric model achieving only 62.3%.
The focus of language model evaluation has transitioned towards reasoning and knowledge-intensive tasks, driven by advancements in pretraining large models. While state-of-the-art models are partially trained on large Arabic texts, evaluating their performance in Arabic remains challenging due to the limited availability of relevant datasets. To bridge this gap, we present \datasetname{}, the first multi-task language understanding benchmark for the Arabic language, sourced from school exams across diverse educational levels in different countries spanning North Africa, the Levant, and the Gulf regions. Our data comprises 40 tasks and 14,575 multiple-choice questions in Modern Standard Arabic (MSA) and is carefully constructed by collaborating with native speakers in the region. Our comprehensive evaluations of 35 models reveal substantial room for improvement, particularly among the best open-source models. Notably, BLOOMZ, mT0, LLaMA2, and Falcon struggle to achieve a score of 50%, while even the top-performing Arabic-centric model only achieves a score of 62.3%.