Grounding Large Language Models in Interactive Environments with Online Reinforcement Learning
This work addresses the lack of grounding in LLMs for interactive decision-making, which is an incremental improvement in reinforcement learning methods for AI agents.
The paper tackles the problem of aligning Large Language Models (LLMs) with interactive environments to improve decision-making by proposing GLAM, an approach that uses online reinforcement learning to update LLM policies during interaction, resulting in enhanced sample efficiency and generalization in spatial and navigation tasks.
Recent works successfully leveraged Large Language Models' (LLM) abilities to capture abstract knowledge about world's physics to solve decision-making problems. Yet, the alignment between LLMs' knowledge and the environment can be wrong and limit functional competence due to lack of grounding. In this paper, we study an approach (named GLAM) to achieve this alignment through functional grounding: we consider an agent using an LLM as a policy that is progressively updated as the agent interacts with the environment, leveraging online Reinforcement Learning to improve its performance to solve goals. Using an interactive textual environment designed to study higher-level forms of functional grounding, and a set of spatial and navigation tasks, we study several scientific questions: 1) Can LLMs boost sample efficiency for online learning of various RL tasks? 2) How can it boost different forms of generalization? 3) What is the impact of online learning? We study these questions by functionally grounding several variants (size, architecture) of FLAN-T5.