Unicoder-VL: A Universal Encoder for Vision and Language by Cross-Modal Pre-TrainingGen Li, Daxin Jiang|Proceedings of the AAAI Conference on Artificial Intelligence|2020Cited by 744