Context Learning for Multi-Agent Discussion

TL;DR

This paper introduces M2CL, a method that improves multi-agent discussions by dynamically generating coherent contexts for each agent, significantly enhancing consensus accuracy across various tasks.

Contribution

M2CL is a novel context learning approach that controls discussion coherence and reduces noise, leading to better multi-agent collaboration in complex problem-solving.

Findings

M2CL outperforms existing methods by 20-50% in accuracy.

It demonstrates improved transferability across tasks.

It maintains computational efficiency while enhancing discussion quality.

Abstract

Multi-Agent Discussion (MAD) has garnered increasing attention very recently, where multiple LLM instances collaboratively solve problems via structured discussion. However, we find that current MAD methods easily suffer from discussion inconsistency, LLMs fail to reach a coherent solution, due to the misalignment between their individual contexts.In this paper, we introduce a multi-LLM context learning method (M2CL) that learns a context generator for each agent, capable of dynamically generating context instructions per discussion round via automatic information organization and refinement. Specifically, inspired by our theoretical insights on the context instruction, M2CL train the generators to control context coherence and output discrepancies via a carefully crafted self-adaptive mechanism.It enables LLMs to avoid premature convergence on majority noise and progressively reach the correct consensus. We evaluate M2CL on challenging tasks, including academic reasoning, embodied tasks, and mobile control. The results show that the performance of M2CL significantly surpasses existing methods by 20%--50%, while enjoying favorable transferability and computational efficiency.