The Single-Multi Evolution Loop for Self-Improving Model Collaboration Systems

2026-02-05

AI Summary

This work addresses the high inference cost and deployment complexity of multi-model collaboration despite its performance benefits. To overcome these limitations, the authors propose a single-to-multi evolutionary cycle mechanism that leverages knowledge distillation to transfer the output of a collaborative system into a single model, enabling it to emulate collaborative behavior while maintaining low inference overhead. By establishing an iterative closed loop—collaboration, distillation, and re-collaboration—the approach facilitates continuous self-evolution of the model. The method is architecture-agnostic and compatible with diverse collaboration strategies and distillation techniques. Evaluated across 15 tasks, it improves single-model performance by 8.0% on average and boosts overall collaborative system performance by 14.9%, significantly outperforming existing evolutionary AI approaches.

Technology Category

• Multiagent Systems: Coordination and Collaboration
• Machine Learning: Evolutionary Learning
• Computer Vision: Multi-modal Vision

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG

Abstract

Model collaboration -- systems where multiple language models (LMs) collaborate -- combines the strengths of diverse models with cost in loading multiple LMs. We improve efficiency while preserving the strengths of collaboration by distilling collaborative patterns into a single model, where the model is trained on the outputs of the model collaboration system. At inference time, only the distilled model is employed: it imitates the collaboration while only incurring the cost of a single model. Furthermore, we propose the single-multi evolution loop: multiple LMs collaborate, each distills from the collaborative outputs, and these post-distillation improved LMs collaborate again, forming a collective evolution ecosystem where models evolve and self-improve by interacting with an environment of other models. Extensive experiments with 7 collaboration strategies and 15 tasks (QA, reasoning, factuality, etc.) demonstrate that: 1) individual models improve by 8.0% on average, absorbing the strengths of collaboration while reducing the cost to a single model; 2) the collaboration also benefits from the stronger and more synergistic LMs after distillation, improving over initial systems without evolution by 14.9% on average. Analysis reveals that the single-multi evolution loop outperforms various existing evolutionary AI methods, is compatible with diverse model/collaboration/distillation settings, and helps solve problems where the initial model/system struggles to.

Problem

• model collaboration
• efficiency
• self-improvement
• distillation
• evolutionary AI

Innovation

• model collaboration
• knowledge distillation
• evolutionary AI
• single-multi evolution loop
• self-improving systems