Context Learning for Multi-Agent Discussion

2026-02-02

AI Summary

This work addresses the inefficiency and difficulty in reaching consensus in multi-agent discussions, often caused by inconsistent individual contexts. To this end, the paper proposes Multi-LLM Context Learning (M2CL), a method that dynamically generates refined contextual instructions for each agent per discussion round through an adaptive mechanism, thereby enabling effective information organization and consistency control. The approach is theoretically grounded in balancing contextual consistency with output diversity, preventing premature convergence to noisy conclusions and fostering progressive convergence toward correct consensus. Evaluated across diverse tasks—including academic reasoning, embodied tasks, and mobile control—M2CL outperforms existing methods by 20%–50%, while demonstrating strong transferability and computational efficiency.

Technology Category

• Multiagent Systems: Multiagent Learning
• Machine Learning: Large Multimodal Models (LMMs)
• Search and Optimization: Learning to Search

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work
• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information

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.

Problem

• Multi-Agent Discussion
• Context Inconsistency
• LLM Coordination
• Coherent Consensus
• Context Alignment

Innovation

• Multi-Agent Discussion
• Context Learning
• LLM Coordination
• Self-Adaptive Mechanism
• Consensus Formation