发表机构
Appier AI Research; National Taiwan University（Appier AI Research; 国立台湾大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本文发现模型合并中系数搜索引入的隐式正则化限制了权重空间，去除该正则化直接优化权重可显著提升多任务性能，甚至超越现有合并方法，呼吁重新审视合并流程。
AI 中文摘要
模型合并旨在通过组合各个任务特定模型的权重，廉价地构建一个多任务模型。为了在多个任务上表现良好，大多数现有的合并方法使用额外的数据集来寻找任务特定权重更新的最佳线性组合系数。然而，我们在这项标准实践中发现了一种隐式正则化：对系数的搜索将候选模型限制在由任务特定权重更新所张成的子空间内。在这项工作中，我们研究了这种正则化是否确实有用。令人惊讶的是，实证结果表明，在没有这种正则化的情况下直接优化合并后的模型权重，能显著提升常见合并方法在多种架构、领域乃至极端数据受限场景（每类仅有一个实例可用）下的性能。此外，直接优化预训练模型权重甚至优于一些现有的合并方法。分析表明，在子空间之外存在更好的多任务权重，并且可以通过多种方法找到它们。我们研究了使用额外数据集的不同策略，讨论了它们的实际用途及对模型合并的启示。总体而言，这项工作呼吁重新审视现有的模型合并流程，激励对权重空间进行更广泛的探索，并重新考虑任务算术引发的隐式正则化。
英文摘要
Model merging aims to build a multi-task model cheaply by combining the weights of individual task-specific models. To perform well across multiple tasks, most existing merging methods use an additional dataset to find the coefficients for the best linear combination of task-specific weight updates. However, we identify an implicit regularization in this standard practice: searching over coefficients restricts the candidate models to a subspace spanned by task-specific weight updates. In this work, we investigate whether this regularization is actually useful. Surprisingly, empirical results show that optimizing merged-model weights without this regularization significantly boosts the performance of common merging methods across multiple architectures, domains, and even in an extremely data-limited scenario where only one instance is available per class. Moreover, directly optimizing the pretrained model weights even outperforms some existing merging methods. Analysis shows that better multi-task weights exist outside the subspace and can be found using multiple methods. We study different strategies for using the additional dataset, discussing their practical use and implications for model merging. Overall, this work calls for revisiting the existing model-merging pipeline, motivating a broader exploration of the weight space and a reconsideration of the implicit regularization induced by task arithmetic.
CommentsPreprint
