A Self-Correcting Vision-Language-Action Model for Fast and Slow System Manipulation | Read Paper on Bytez