Fetching the paper…

Self-supervised Pre-training with Hard Examples Improves Visual Representations · Around