1 paper
Arsha Nagrani, Mingda Zhang, Ramin Mehran +10
We introduce Neptune, a benchmark for long video understanding that requires reasoning over long time horizons and across different modalities. Many existing video datasets and mod…