【论文&模型讲解】VideoBERT: A Joint Model for Video and Language Representation Learning