Exploratory Memory-Augmented LLM Agent via Hybrid On- and Off-Policy Optimization

TL;DR

This paper introduces EMPO$^2$, a hybrid reinforcement learning framework that enhances large language model agents' exploration and adaptability by leveraging memory and combined update strategies, leading to significant performance improvements.

Contribution

EMPO$^2$ is a novel hybrid RL approach that integrates memory with on- and off-policy updates to improve exploration and robustness in LLM agents.

Findings

Achieves 128.6% and 11.3% improvements over GRPO on ScienceWorld and WebShop.

Demonstrates superior out-of-distribution adaptability with minimal trials and no parameter updates.

Enhances exploration and generalization in LLM-based agents.

Abstract

Exploration remains the key bottleneck for large language model agents trained with reinforcement learning. While prior methods exploit pretrained knowledge, they fail in environments requiring the discovery of novel states. We propose Exploratory Memory-Augmented On- and Off-Policy Optimization (EMPO$^2$), a hybrid RL framework that leverages memory for exploration and combines on- and off-policy updates to make LLMs perform well with memory while also ensuring robustness without it. On ScienceWorld and WebShop, EMPO$^2$ achieves 128.6% and 11.3% improvements over GRPO, respectively. Moreover, in out-of-distribution tests, EMPO$^2$ demonstrates superior adaptability to new tasks, requiring only a few trials with memory and no parameter updates. These results highlight EMPO$^2$ as a promising framework for building more exploratory and generalizable LLM-based agents.