Recent studies have shown that large language models exhibit formidable capabilities through online contextual learning, enabling them to learn from human feedback to write robot code. Research teams have successfully enhanced the efficiency of LLMs in writing robot code through the LMPC framework, thereby accelerating the robot learning process. Experiments have demonstrated that LMPC significantly improves the success rate of unseen tasks, providing robust support for adaptive robot learning. This research marks a new breakthrough in the field of robot learning, promoting the ability of robots to quickly adapt to human inputs.