arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2603.21705cs.LG
收藏

无需数据的层自适应融合:通过信息论实现长到短推理LLM

Data-Free Layer-Adaptive Merging via Fisher Information for Long-to-Short Reasoning LLMs

Tian Xia

AI总结:

本文提出FIM-Merging方法,通过计算Fisher信息矩阵来分配每层融合系数,提升长到短推理LLM的性能,减少响应长度,且无需校准数据。

摘要:

模型融合已成为一种实用方法,用于在不额外训练的情况下结合专用大型语言模型(LLM)的能力。在长到短(L2S)场景中,将基础模型与长链推理模型融合旨在保留推理准确性的同时减少输出长度。现有方法依赖于任务算术及其变种,隐含假设模型输出与融合系数线性变化——我们证明在L2S设置中这一假设系统性地被违反。我们提供了首个理论依据:证明融合误差被每层Hessian范数成比例的项所限制(命题1),并通过局部极值点处的Fisher-Hessian等价性,建立Fisher信息矩阵(FIM)作为此界限的可行代理。基于此理论,我们提出FIM-Merging,仅使用随机标记输入计算对角FIM(无需领域特定校准数据),并利用其分配每层融合系数。在7B L2S基准上,FIM-TIES在六个评估基准中的五个中取得最佳性能,包括在MATH500上比ACM-TIES(90.2 vs. 84.0)多6.2个点,同时无需校准数据。在1.5B基准上,FIM-TIES的平均准确率为47.3,超过之前的最佳ACM-TIES(43.3)3.9个点,同时将响应长度减少91.9%相对长CoT模型。我们的框架还提供了现有层自适应方法如ACM为何在经验上优于均匀融合的统一理论解释。

英文摘要:

Model merging has emerged as a practical approach to combine capabilities of specialized large language models (LLMs) without additional training. In the Long-to-Short (L2S) scenario, merging a base model with a long-chain-of-thought reasoning model aims to preserve reasoning accuracy while reducing output length. Existing methods rely on Task Arithmetic and its variants, which implicitly assume that model outputs vary linearly with the merging coefficient -- an assumption we show is systematically violated in L2S settings. We provide the first theoretical justification for layer-adaptive merging: we prove that merging error is bounded by a term proportional to the per-layer Hessian norm (Proposition~1), and establish that the Fisher Information Matrix (FIM) is a principled, computable proxy for this bound via the Fisher-Hessian equivalence at local optima. Building on this theory, we propose \textbf{FIM-Merging}, which computes diagonal FIM using only random token inputs (no domain-specific calibration data required) and uses it to assign per-layer merging coefficients. On the 7B L2S benchmark, FIM-TIES achieves state-of-the-art performance on five out of six evaluation benchmarks, including a \textbf{+6.2} point gain on MATH500 over ACM-TIES (90.2 vs.\ 84.0), while requiring no calibration data. On the 1.5B benchmark, FIM-TIES achieves an average accuracy of \textbf{47.3}, surpassing the previous best ACM-TIES (43.3) by \textbf{+3.9} points, while reducing average response length by \textbf{91.9\%} relative to the long-CoT model. Our framework also provides a unified theoretical explanation for why existing layer-adaptive methods such as ACM empirically outperform uniform merging.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑