VITATECS: A Diagnostic Dataset for Temporal Concept Understanding of Video-Language ModelsShare on Twitter Facebook LinkedIn Previous Next