VILA-M3: Enhancing Vision-Language Models with Medical Expert Knowledge | Read Paper on Bytez