ST-EVO: Towards Generative Spatio-Temporal Evolution of Multi-Agent Communication Topologies

TL;DR

ST-EVO tackles the challenge of coordinating multi-agent LLM systems by introducing a Spatio-Temporal Evolving MAS that schedules communication topologies across dialogue iterations. It combines a compact Flow-Matching based Scheduler with entropy and experience driven self-feedback to adapt topology dynamically, producing a sequence of graphs $G_t$ and execution orders $S_t$ guided by a query $\mathcal{Q}$. The approach achieves state-of-the-art performance across nine benchmarks, with significant accuracy gains and improved efficiency and robustness, demonstrating the advantages of joint spatial and temporal scheduling. This method offers a practical path toward robust, flexible, and scalable cooperative AI systems capable of adapting to diverse tasks with limited additional supervision.

Abstract

LLM-powered Multi-Agent Systems (MAS) have emerged as an effective approach towards collaborative intelligence, and have attracted wide research interests. Among them, ``self-evolving'' MAS, treated as a more flexible and powerful technical route, can construct task-adaptive workflows or communication topologies, instead of relying on a predefined static structue template. Current self-evolving MAS mainly focus on Spatial Evolving or Temporal Evolving paradigm, which only considers the single dimension of evolution and does not fully incentivize LLMs' collaborative capability. In this work, we start from a novel Spatio-Temporal perspective by proposing ST-EVO, which supports dialogue-wise communication scheduling with a compact yet powerful flow-matching based Scheduler. To make precise Spatio-Temporal scheduling, ST-EVO can also perceive the uncertainty of MAS, and possesses self-feedback ability to learn from accumulated experience. Extensive experiments on nine benchmarks demonstrate the state-of-the-art performance of ST-EVO, achieving about 5%--25% accuracy improvement.

Figures (10)

• Figure 1: Existing multi-agent systems. Our proposed ST-EVO pioneers the exploration of Spatio-Temporal Evolving multi-agent systems.
• Figure 2: The overview of ST-EVO.
• Figure 3: Visualization of performance metrics and token consumption of different multi-agent topologies across MMLU, HumanEval, GSM8K, SVAMP.
• Figure 4: We utilize the accuracy (%) and $e^{\mathcal{R}_{\text{sta}}}\in [0,1]$ to visualize the stability of ST-EVO. Higher values indicate better stabilities.
• Figure 5: We compare the accuracy (%) of multiple multi-agent systems before and after prompt attacks on all benchmarks, and report the average results.