Measuring Massive Multitask Language Understanding (MMLU)

 a popular benchmark for evaluating the capabilities of large language models.