1 paper · 1 filter
Samuele Punzo, Niccolò Caselli, Ippokratis Pantelidis +3
We study whether pretrained video foundation models encode intuitive-physics information in their frozen representations, and how this information varies across model families, lay…