social sciences

CMMLU is an evaluation benchmark designed for the Chinese context, covering 67 topics to comprehensively test language models' knowledge and reasoning abilities, with a particular emphasis on China-specific knowledge areas.