TopoDIM: One-shot Topology Generation of Diverse Interaction Modes for Multi-Agent Systems

TL;DR

TopoDIM tackles the inefficiency of multi-round, spatio-temporal communication in LLM-based MAS by proposing a one-shot, heterogeneous topology generation framework. It models agents and interactions as a directed heterogeneous graph and learns a policy to generate diverse edge types—Conditioned, Feedback, and Debate—within a decentralized, privacy-aware architecture. The approach couples a semantics-aware relational encoder with an autoregressive edge-sampling decoder and optimizes topology via reinforcement learning with a diversity-aware reward, while distilling global decisions into lightweight local policies for agents. Empirical results show improved task performance and substantial token efficiency across heterogeneous agent setups and reasoning benchmarks, with strong robustness to perturbations and clear advantages over existing intra-, inter-, and hybrid-dialogue baselines. Overall, TopoDIM offers a scalable, adaptive, and cost-efficient framework for decentralized MAS coordination that can better exploit diverse cognitive interactions in large-scale AI systems.

Abstract

Optimizing communication topology in LLM-based multi-agent system is critical for enabling collective intelligence. Existing methods mainly rely on spatio-temporal interaction paradigms, where the sequential execution of multi-round dialogues incurs high latency and computation. Motivated by the recent insights that evaluation and debate mechanisms can improve problem-solving in multi-agent systems, we propose TopoDIM, a framework for one-shot Topology generation with Diverse Interaction Modes. Designed for decentralized execution to enhance adaptability and privacy, TopoDIM enables agents to autonomously construct heterogeneous communication without iterative coordination, achieving token efficiency and improved task performance. Experiments demonstrate that TopoDIM reduces total token consumption by 46.41% while improving average performance by 1.50% over state-of-the-art methods. Moreover, the framework exhibits strong adaptability in organizing communication among heterogeneous agents. Code is available at: https://anonymous.4open.science/r/TopoDIM-8D35/

Figures (13)

• Figure 1: Illustration of hybrid intra/inter-round dialogue method versus TopoDIM. TopoDIM models complex interactions via one-shot topology generation, efficiently cutting potential token overhead.
• Figure 2: The framework of TopoDIM, comprising 4 components: ❶ Materials: defining agents, roles, interaction modes, and plugins; ❷ Topology: illustrating the heterogeneous topology design; ❸ Optimization: detailing topology optimization strategies, and ❹ Decentralization: describing the decentralized agent decision-making.
• Figure 3: Interaction modes of TopoDIM. TopoDIM selects three effective interaction modes aiming to optimize the execution sequence among the agents.
• Figure 4: Performance-cost trade-off between TopoDIM and state-of-the-art methods. The bubble size is proportional to token consumption.
• Figure 5: Dependence on hyperparameter and architecture designs. (a) Edge diversity vs. accuracy/pass@1, (b) edge sparsity vs. accuracy/pass@1, (c) training sample size vs. accuracy/pass@1, and (d) decentralized sample size vs. accuracy/pass@1.