AgentSquare: Automatic LLM Agent Search in Modular Design Space

2024-10-08

arXiv.org

AI Summary

To address the poor generalizability and task adaptability of manually designed LLM agents, this paper proposes Modular LLM Agent Search (MoLAS), a novel paradigm. First, it defines a unified modular design space comprising four composable interfaces: planning, reasoning, tool invocation, and memory. Second, it introduces AgentSquare—an automated search framework integrating module-level evolutionary optimization, structural recombination, and an in-context surrogate model for efficient performance prediction. The method unifies input-output abstraction, evolutionary optimization, and modular architecture search. Evaluated across six benchmarks, MoLAS achieves an average 17.2% improvement over human-designed optimal agents and enables interpretable architectural analysis. The implementation is publicly available.

Technology Category

• Cognitive Modeling & Cognitive Systems: Agent Architectures
• Search and Optimization: Learning to Search
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

Recent advancements in Large Language Models (LLMs) have led to a rapid growth of agentic systems capable of handling a wide range of complex tasks. However, current research largely relies on manual, task-specific design, limiting their adaptability to novel tasks. In this paper, we introduce a new research problem: Modularized LLM Agent Search (MoLAS). We propose a modular design space that abstracts existing LLM agent designs into four fundamental modules with uniform IO interface: Planning, Reasoning, Tool Use, and Memory. Building on this design space, we present a novel LLM agent search framework called AgentSquare, which introduces two core mechanisms, i.e., module evolution and recombination, to efficiently search for optimized LLM agents. To further accelerate the process, we design a performance predictor that uses in-context surrogate models to skip unpromising agent designs. Extensive experiments across six benchmarks, covering the diverse scenarios of web, embodied, tool use and game applications, show that AgentSquare substantially outperforms hand-crafted agents, achieving an average performance gain of 17.2% against best-known human designs. Moreover, AgentSquare can generate interpretable design insights, enabling a deeper understanding of agentic architecture and its impact on task performance. We believe that the modular design space and AgentSquare search framework offer a platform for fully exploiting the potential of prior successful designs and consolidating the collective efforts of research community. Code repo is available at https://github.com/tsinghua-fib-lab/AgentSquare.

Problem

• Automates LLM agent design
• Enhances task adaptability
• Optimizes agent performance

Innovation

• Modularized LLM Agent Search
• Module evolution and recombination
• In-context surrogate models