Memory-Based Advantage Shaping for LLM-Guided Reinforcement Learning
This addresses scalability and reliability issues in LLM-guided RL for environments with sparse rewards, though it is incremental.
The paper tackles the high sample complexity in sparse-reward reinforcement learning by using a memory graph to encode subgoals from LLM guidance and agent rollouts, shaping the advantage function to improve sample efficiency and early learning speed, with final returns matching methods needing frequent LLM calls.
In environments with sparse or delayed rewards, reinforcement learning (RL) incurs high sample complexity due to the large number of interactions needed for learning. This limitation has motivated the use of large language models (LLMs) for subgoal discovery and trajectory guidance. While LLMs can support exploration, frequent reliance on LLM calls raises concerns about scalability and reliability. We address these challenges by constructing a memory graph that encodes subgoals and trajectories from both LLM guidance and the agent's own successful rollouts. From this graph, we derive a utility function that evaluates how closely the agent's trajectories align with prior successful strategies. This utility shapes the advantage function, providing the critic with additional guidance without altering the reward. Our method relies primarily on offline input and only occasional online queries, avoiding dependence on continuous LLM supervision. Preliminary experiments in benchmark environments show improved sample efficiency and faster early learning compared to baseline RL methods, with final returns comparable to methods that require frequent LLM interaction.