<ul><li>Reinforcement learning often faces challenges with reward misalignment, where agents optimize for given rewards but fail to exhibit the desired behaviors.</li><li>To address these issues, the proposed approach utilizes zero-shot, off-the-shelf large language models (LLMs) for reward shaping in continuous control tasks.</li><li>The LLM-HFBF framework is introduced to identify and correct biases in human feedback while incorporating it into the reward shaping process.</li><li>Empirical experiments demonstrate that the proposed approach reduces reliance on potentially biased human guidance and maintains high reinforcement learning performance.</li></ul>

Zero-Shot LLMs in Human-in-the-Loop RL: Replacing Human Feedback for Reward Shaping

Discover more