Talk Structurally, Act Hierarchically: A Collaborative Framework for LLM Multi-Agent Systems
This work addresses the problem of improving collaboration and output quality in multi-agent AI systems for researchers and practitioners, representing a novel method rather than an incremental improvement.
The paper tackles the challenge of managing communication and refinement in LLM-based multi-agent systems by proposing the TalkHier framework, which introduces structured communication and hierarchical refinement to address issues like incorrect outputs and biases, achieving superior performance over state-of-the-art methods across diverse tasks such as open-domain question answering and advertisement text generation.
Recent advancements in LLM-based multi-agent (LLM-MA) systems have shown promise, yet significant challenges remain in managing communication and refinement when agents collaborate on complex tasks. In this paper, we propose \textit{Talk Structurally, Act Hierarchically (TalkHier)}, a novel framework that introduces a structured communication protocol for context-rich exchanges and a hierarchical refinement system to address issues such as incorrect outputs, falsehoods, and biases. \textit{TalkHier} surpasses various types of SoTA, including inference scaling model (OpenAI-o1), open-source multi-agent models (e.g., AgentVerse), and majority voting strategies on current LLM and single-agent baselines (e.g., ReAct, GPT4o), across diverse tasks, including open-domain question answering, domain-specific selective questioning, and practical advertisement text generation. These results highlight its potential to set a new standard for LLM-MA systems, paving the way for more effective, adaptable, and collaborative multi-agent frameworks. The code is available https://github.com/sony/talkhier.