arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2602.12952cs.LGcs.AIcs.CV
收藏

在不同架构间传输任务向量而无需训练

Transporting Task Vectors across Different Architectures without Training

AImageLab, University of Modena and Reggio Emilia(AImageLab,Modena和雷吉奥艾米利亚大学)

Filippo Rinaldi, Aniello Panariello, Giacomo Salici, Angelo Porrello, Simone Calderara

AI总结:

本文提出Theseus方法,通过功能匹配在不同宽度模型间传输任务更新,无需训练或反向传播,展示了在视觉和语言模型上的改进效果。

摘要:

适应大型预训练模型以完成下游任务时,通常会产生针对特定任务的参数更新,这些更新对于每个模型变体重新学习都很昂贵。尽管最近的研究表明,这些更新可以在具有相同架构的模型之间转移,但跨不同宽度的模型转移仍鲜有探索。在本文中,我们引入Theseus,一种无需训练的方法,用于在异构宽度模型间传输任务更新。与其匹配参数,我们通过其在中间表示上诱导的功能效应来表征任务更新。我们正式将任务向量传输定义为在观察到的激活上进行的功能匹配问题,并显示在通过正交Procrustes分析对齐表示空间后,它允许一个稳定的闭式解,该解保留了更新的几何结构。我们在不同宽度的视觉和语言模型上评估Theseus,显示在不进行额外训练或反向传播的情况下,相对于基线有持续的改进。我们的结果表明,当任务身份通过功能而非参数定义时,任务更新可以有意义地在不同架构间转移。代码可在https://github.com/apanariello4/merge-and-rebase获取。

英文摘要:

Adapting large pre-trained models to downstream tasks often produces task-specific parameter updates that are expensive to relearn for every model variant. While recent work has shown that such updates can be transferred between models with identical architectures, transferring them across models of different widths remains unexplored. In this work, we introduce Theseus, a training-free method for transporting task updates across heterogeneous-width models. Rather than matching parameters, we characterize a task update by the functional effect it induces on intermediate representations. We formalize task-vector transport as a functional matching problem on observed activations and show that, after aligning representation spaces via orthogonal Procrustes analysis, it admits a stable closed-form solution that preserves the geometry of the update. We evaluate Theseus on vision and language models across different widths, showing consistent improvements over baselines without additional training or backpropagation. Our results show that task updates can be meaningfully transferred across architectures when task identity is defined functionally rather than parametrically. Code is available at https://github.com/apanariello4/merge-and-rebase.

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。

↑