OMAC: A Holistic Optimization Framework for LLM-Based Multi-Agent Collaboration

TL;DR

This paper introduces OMAC, a comprehensive framework for optimizing large language model-based multi-agent systems across multiple dimensions, improving their performance on complex tasks through systematic design and joint optimization strategies.

Contribution

OMAC is the first holistic framework that systematically optimizes LLM-based multi-agent systems across five key dimensions, enhancing their collaborative capabilities.

Findings

OMAC outperforms recent approaches on diverse tasks.

The framework effectively optimizes both individual and multiple dimensions.

Experiments validate the superiority of OMAC in complex applications.

Abstract

Agents powered by advanced large language models (LLMs) have demonstrated impressive capabilities across diverse complex applications. Recently, Multi-Agent Systems (MAS), wherein multiple agents collaborate and communicate with each other, have exhibited enhanced capabilities in complex tasks, such as high-quality code generation and arithmetic reasoning. However, the development of such systems often relies on handcrafted methods, and the literature on systematic design and optimization of LLM-based MAS remains limited. In this work, we introduce \textbf{OMAC}, a general framework designed for holistic optimization of LLM-based MAS. Specifically, we identify five key optimization dimensions for MAS, encompassing both agent functionality and collaboration structure. Building upon these dimensions, we first propose a general algorithm, utilizing two actors termed the Semantic Initializer and the Contrastive Comparator, to optimize any single dimension. Then, we present an algorithm for joint optimization across multiple dimensions. Extensive experiments demonstrate the superior performance of OMAC on diverse tasks against recent approaches.