Enabling Flexible Multi-LLM Integration for Scalable Knowledge Aggregation

TL;DR

This paper introduces a flexible framework for integrating multiple large language models by adaptively selecting and aggregating their knowledge, improving scalability and reducing interference compared to traditional methods.

Contribution

It proposes an adaptive selection and dynamic fusion strategy for multi-LLM integration, addressing limitations of existing ensemble and weight merging techniques.

Findings

Reduces knowledge interference by up to 50%.

Enables more stable and scalable LLM knowledge aggregation.

Demonstrates improved performance over existing methods.

Abstract

Large language models (LLMs) have shown remarkable promise but remain challenging to continually improve through traditional finetuning, particularly when integrating capabilities from other specialized LLMs. Popular methods like ensemble and weight merging require substantial memory and struggle to adapt to changing data environments. Recent efforts have transferred knowledge from multiple LLMs into a single target model; however, they suffer from interference and degraded performance among tasks, largely due to limited flexibility in candidate selection and training pipelines. To address these issues, we propose a framework that adaptively selects and aggregates knowledge from diverse LLMs to build a single, stronger model, avoiding the high memory overhead of ensemble and inflexible weight merging. Specifically, we design an adaptive selection network that identifies the most relevant source LLMs based on their scores, thereby reducing knowledge interference. We further propose a dynamic weighted fusion strategy that accounts for the inherent strengths of candidate LLMs, along with a feedback-driven loss function that prevents the selector from converging on a single subset of sources. Experimental results demonstrate that our method can enable a more stable and scalable knowledge aggregation process while reducing knowledge interference by up to 50% compared to existing approaches. Code is avaliable at https://github.com/ZLKong/LLM_Integration