Image as a Foreign Language: BEiT Pretraining for All Vision and Vision-Language Tasks - 42Papers