Social Caption: Evaluating Social Understanding in Multimodal Models 事件

Social Caption: Evaluating Social Understanding in Multimodal Models arXiv:2601.14569v2 Announce Type: replace Abstract: Social understanding abilities are crucial for multimodal large language models (MLLMs) to interpret human social interactions. We introduce SOCIAL CAPTION, a framework grounded in interaction theory to evaluate social understanding abilities of MLLMs along three dimensions: Social Inference (SI), the ability to make accurate inferences about interactions; Holistic Social Ana