Optimization Inspired Few-Shot Adaptation for Large Language Models | Read Paper on Bytez