A giant DINOv2 model with a single layer head trained on ImageNet-1k for self-supervised visual feature extraction.