You Don’t Need Domain-Specific Data Augmentations When Scaling Self-Supervised Learning | Read Paper on Bytez