1 paper · 1 filter
Jonghyun Lee, Dojun Park, Jiwoo Lee +2
This study investigated whether multimodal large language models can achieve human-like sensory grounding by examining their ability to capture perceptual strength ratings across s…