Implementation of Vision Transformer in PyTorch, a new model to achieve SOTA in vision classification with using transformer style encoders. Associated blog article.