通过筛选关键神经元实现模型融合，缓解参数冲突问题。
MIN-Merging: Merge the Important Neurons for Model Merging
- 基于路由机制选择重要神经元进行融合
- 在CV和NLP任务上均提升领域内性能
- 适合需要融合多个模型又保持泛化能力的场景
深度学习的快速发展催生了众多开源模型。模型融合虽能整合各模型优势，但现有方法常因参数冲突导致特定领域任务性能下降。本文提出MIN-Merging，一种基于路由的框架，仅合并最重要的神经元以减少冲突。在计算机视觉（CV）与自然语言处理（NLP）基准上的大量实验表明，MIN-Merging在领域内任务上持续提升性能，同时保留预训练模型在域外任务上的泛化能力。结果验证了其作为解决模型融合中参数冲突问题的有效实用方案。
原文摘要 · Abstract (English)
Recent advances in deep learning have led to a surge of open-source models across diverse domains. While model merging offers a promising way to combine their strengths, existing approaches often suffer from parameter conflicts that degrade performance on domain-specific tasks. We propose MIN-Merging, a router-based framework that selectively merges the most important neurons to reduce such conflicts. Extensive experiments on Computer Vision(CV) and Natural Language Processing(NLP) benchmarks show that MIN-Merging achieves consistent gains on in-domain tasks while retaining the generalization ability of pretrained models on out-of-domain tasks. These results highlight its effectiveness as a practical solution to the parameter conflict problem in model merging.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。