Stable communication methods enable agent model swapping in AI systems
LatBridge: Efficient and Stable Model Replacement in Latent-Communication Multi-Agent Systems
Artificial Intelligence
Summary
Multi-agent AI systems need smooth communication to work well, especially when the agents themselves change. The authors found that existing ways to share hidden messages between agents break when one agent is switched out for another. They created LatBridge, a method that keeps communication working even after swapping agents by reusing parts of the messaging system and training only what’s needed. Tests showed LatBridge maintains good task accuracy and efficient communication after agent replacement.
What this means in practice
- •For robotics developers: Maintain reliable communication between autonomous robots when individual robot models are updated or replaced without retraining the entire system.
- •For game ai programmers: Swap AI character behaviors in games while keeping their communication intact, ensuring smooth collaboration without full retraining.
Authors
Jiajun Zhu, Haobo Li, Zixin Chen, Huamin Qu, Wenshuo Zhang
Abstract
Multi-agent systems rely on communication to complete tasks, and latent communication passes continuous representations to reduce text generation and re-encoding. However, multi-agent systems must reconfigure agent models, whereas existing latent-communication methods are built for fixed model combinations, so their interfaces may fail after replacement. Our pilot study confirms this interface failure and identifies the requirement for interface reuse after model replacement. We present LatBridge, which connects agent models through a shared Bridge and separate interfaces under a shared latent-communication protocol. After agent model replacement, LatBridge reuses unchanged interfaces, selects matching low-rank interfaces, and combines them through the Bridge, training only missing ones with all models frozen. The Bridge processes source representations in the basic mode and constructs target input without autoregressive generation in the accelerated mode. Experiments support effective communication after model replacement, with mean basic-mode accuracies of 14.8% on MATH and 27.3% on ALFWorld next-action prediction across eight small-model configurations. Analyses examine the factors supporting task accuracy and communication efficiency.