Search papers, labs, and topics across Lattice.
This paper presents a unified framework for categorizing and analyzing memory methods used in LLM-based agents, focusing on their ability to handle long-horizon complex tasks. Through extensive comparative experiments on established benchmarks, the authors evaluate the effectiveness of representative memory architectures. They also introduce a novel memory method derived from existing modules, achieving state-of-the-art performance and highlighting promising research directions.
LLM agent memory architectures matter, and a systematic comparison reveals a new SOTA method by combining the best of existing approaches.
Memory emerges as the core module in the large language model (LLM)-based agents for long-horizon complex tasks (e.g., multi-turn dialogue, game playing, scientific discovery), where memory can enable knowledge accumulation, iterative reasoning and self-evolution. A number of memory methods have been proposed in the literature. However, these methods have not been systematically and comprehensively compared under the same experimental settings. In this paper, we first summarize a unified framework that incorporates all the existing agent memory methods from a high-level perspective. We then extensively compare representative agent memory methods on two well-known benchmarks and examine the effectiveness of all methods, providing a thorough analysis of those methods. As a byproduct of our experimental analysis, we also design a new memory method by exploiting modules in the existing methods, which outperforms the state-of-the-art methods. Finally, based on these findings, we offer promising future research opportunities. We believe that a deeper understanding of the behavior of existing methods can provide valuable new insights for future research.