1 paper · 1 filter
Ali Asgarov, Kaushik Narasimhan, Najibul Haque Sarker +6
Recent progress in vision-language models has enabled processing of increasingly long video sequences, but handling extended token streams does not translate to understanding compl…