Abstract page for arXiv paper 2608.20473: Aggregating Visual Information with Optimal Transport for VideoLM Token Compression