arXivDaily arXiv每日学术速递 周一至周五更新
北美退伍军人节(Veterans Day),12 日无更新
arXiv 2302.04863cs.LGcs.AIcs.CL
收藏

知识是微调语言模型权重空间中的一个区域

Knowledge is a Region in Weight Space for Fine-tuned Language Models

Technion - IIT(以色列理工学院);IBM Research(IBM研究院);UNC Chapel Hill(北卡罗来纳大学教堂山分校)

Almog Gueta, Elad Venezian, Colin Raffel, Noam Slonim, Yoav Katz, Leshem Choshen

AI总结:

本研究揭示微调语言模型在权重空间中形成区域,区域内的模型性能优越,且区域中心可作为高效微调的更好起点,在多数数据集上提升准确率。

摘要:

神经网络的研究一直聚焦于理解在单一数据集上训练的单一模型。然而,关于不同模型之间的关系,尤其是那些在不同数据集上训练或测试的模型,我们知之甚少。我们通过研究不同模型的权重空间和底层损失景观如何相互关联来解决这一问题。具体而言,我们证明了为高性能而优化的微调模型位于权重空间中定义明确的区域,反之亦然——任何位于这些区域内的模型也表现出高性能。值得注意的是,我们展示了在同一数据集上微调的语言模型在权重空间中形成紧密的簇,而在同一底层任务的不同数据集上微调的模型则形成较松散的簇。此外,在模型之间的区域周围遍历会产生新的模型,其性能与通过微调获得的模型相当甚至更好,即使在原始模型未微调的任务上也是如此。我们的发现提供了对模型之间关系的洞察,表明位于两个相似模型之间的模型可以获取两者的知识。我们利用这一点,设计了一种选择更好模型以进行高效微调的方法。具体来说,我们表明在12个数据集中有11个,从区域中心开始与使用预训练模型一样有效,甚至更有效,平均准确率提高了3.06。

英文摘要:

Research on neural networks has focused on understanding a single model trained on a single dataset. However, relatively little is known about the relationships between different models, particularly those trained or tested on different datasets. We address this by studying how the weight space and the underlying loss landscape of different models are interconnected. Specifically, we demonstrate that finetuned models that were optimized for high performance, reside in well-defined regions in weight space, and vice versa -- that any model that resides anywhere in those regions also exhibits high performance. Notably, we show that language models that have been finetuned on the same dataset form a tight cluster in the weight space, while models finetuned on different datasets from the same underlying task form a looser cluster. Moreover, traversing around the region between the models leads to new models that perform comparably or even better than models obtained via finetuning, even on tasks that the original models were not finetuned on. Our findings provide insight into the relationships between models, demonstrating that a model positioned between two similar models can acquire the knowledge of both. We leverage this and design a method for selecting a better model for efficient finetuning. Specifically, we show that starting from the center of the region is as effective, if not more, than using the pretrained model in 11 out of 12 datasets, resulting in an average accuracy improvement of 3.06.

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。

↑