CoMMET: A Psychologically Grounded Benchmark for Evaluating Theory of Mind in Multimodal LLMs
A new multimodal benchmark dataset, CoMMET, a comprehensive mental states and moral evaluation task inspired by the Theory of Mind Booklet Task is proposed, which is the first psychology-grounded benchmark to evaluate MLLMs across multiple mental states in a multimodal, open-ended, and multi-turn setting.
Rui-Rui Chen, Wei-Feng Jiang, Cheng-Wei Qin et al.
· 0 citations