MeetingToM: Evaluating Multimodal LLMs on Theory-of-Mind Reasoning in Multi-Party Meetings
Paper ·
GitHub ·
Project Page
MeetingToM is a multimodal benchmark for evaluating Theory-of-Mind (ToM) reasoning in multi-party meetings. It studies social reasoning at three complementary levels: individual mental states, interpersonal relations, and group-level consensus.
The released benchmark contains 900 source bundles, expanded into 1,200 evaluation records with 1,800… See the full description on the dataset page:
https://huggingface.co/datasets/OliviaWang1101/MeetingToM.
