Self-supervised Image-text Pre-training With Mixed Data In Chest X-rays | Read Paper on Bytez