通过对齐参数空间，实现大模型推理能力高效迁移。
Leveraging Parameter Space Symmetries for Reasoning Skill Transfer in LLMs
- 利用Transformer的对称性对齐参数空间，减少技能迁移干扰。
- 在复杂推理基准上优于传统任务算术，提升迁移效果。
- 适合需要快速融合专家能力的模型优化场景。
任务算术是一种强大的大型语言模型（LLM）技能迁移技术，但在模型训练路径分化后常遭遇负面干扰。本文提出先对齐模型参数空间，利用Transformer架构固有的排列、旋转和缩放对称性来缓解此问题。方法适配现代分组查询注意力（GQA）和SwiGLU层，探索基于权重与激活的对齐方式。采用该对齐优先策略，成功将高级推理能力迁移到非推理型模型中。在多个挑战性推理基准上的实验表明，本方法始终优于标准任务算术。该工作为跨演进语言模型家族的技能融合与迁移提供了有效方案，减少冗余微调，增强模型适应性。
原文摘要 · Abstract (English)
Task arithmetic is a powerful technique for transferring skills between Large Language Models (LLMs), but it often suffers from negative interference when models have diverged during training. We address this limitation by first aligning the models' parameter spaces, leveraging the inherent permutation, rotation, and scaling symmetries of Transformer architectures. We adapt parameter space alignment for modern Grouped-Query Attention (GQA) and SwiGLU layers, exploring both weight-based and activation-based approaches. Using this alignment-first strategy, we successfully transfer advanced reasoning skills to a non-reasoning model. Experiments on challenging reasoning benchmarks show that our method consistently outperforms standard task arithmetic. This work provides an effective approach for merging and transferring specialized skills across evolving LLM families, reducing redundant fine-tuning and enhancing model adaptability.
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。