Evolutionary model merging provides a powerful framework for the automated, training-free composition of LLMs through parameter-space search. However, existing methods predominantly rely on stochastic, hand-crafted operators that overlook the underlying performance landscape of the coefficient space. We propose Evolutionary Generative Merging (EvoGM), a framework that transcends manual heuristics by employing learnable generative modeling to optimize merging coefficients. Specifically, EvoGM features a dual-generator architecture with cycle-consistent learning to adaptively sample and refine promising merging candidates. By constructing winner-loser pairs from historical search trajectories, our framework effectively captures high-performance parameter distributions and maximizes data efficiency. This generative process is seamlessly integrated into a multi-round evolutionary pipeline, where elite merged models iteratively serve as new expert foundations. Extensive experiments across diverse benchmarks demonstrate that EvoGM significantly outperforms state-of-the-art baselines, exhibiting robust performance on both seen and unseen tasks. Code and data are available at https://github.com/JiangTao97/evogm.

翻译：进化模型合并为无需训练的自动化大语言模型参数空间组合提供了强大框架。然而，现有方法主要依赖随机的手工设计算子，忽略了系数空间中的潜在性能景观。本文提出进化生成合并（EvoGM）框架，通过采用可学习的生成建模来优化合并系数，超越人工启发式方法。具体而言，EvoGM采用具有循环一致性学习的双生成器架构，能够自适应采样并优化有前景的合并候选方案。通过从历史搜索轨迹中构建胜者-败者对，该框架有效捕捉高性能参数分布，并最大化数据利用效率。这一生成过程无缝融入多轮进化流水线，使精英合并模型迭代成为新的专家基础。跨多个基准的大量实验表明，EvoGM显著优于当前最优基线，在已见与未见任务上均展现出稳健性能。代码与数据见https://github.com/JiangTao97/evogm。