Abstract: This study explored methods to improve the success rate of reward function generation, a crucial task in robotic reinforcement learning, by applying In-Context Learning (ICL) and ...