1K resolution vision transformers pretrained on 1B human images. - View it on GitHub