#efficient_models #foundation_models #generation #optimization #vision_transformer #CLS #token #Imagen #ViT #Vision #Transformer #attention #image #classification #image #patch #embedding #inductive #bias #large-scale #pre-training #position #embedding