arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2510.04548cond-mat.dis-nncs.LGstat.ML
收藏

在上下文中学习低秩任务的线性回归

Learning Linear Regression with Low-Rank Tasks in-Context

The University of Tokyo(东京大学);RIKEN AIP(理化学研究所AIP)

Kaito Takanami, Takashi Takahashi, Yoshiyuki Kabashima

AI总结:

本文研究了在上下文中学习如何在高维极限下精确刻画预测分布和泛化误差,并发现有限预训练数据中的统计波动诱导了隐式正则化,揭示了任务结构对泛化误差的相变。

摘要:

在上下文中学习(ICL)是现代大语言模型的关键构建块,但其理论机制仍不明确。本文通过分析一个在低秩回归任务上训练的线性注意力模型,探讨了ICL在现实应用中如何操作。在该设定下,我们精确刻画了预测分布和高维极限下的泛化误差。此外,我们发现有限预训练数据中的统计波动诱导了隐式正则化。最后,我们识别了由任务结构决定的泛化误差的尖锐相变。这些结果为理解transformers如何学习任务结构提供了框架。

英文摘要:

In-context learning (ICL) is a key building block of modern large language models, yet its theoretical mechanisms remain poorly understood. It is particularly mysterious how ICL operates in real-world applications where tasks have a common structure. In this work, we address this problem by analyzing a linear attention model trained on low-rank regression tasks. Within this setting, we precisely characterize the distribution of predictions and the generalization error in the high-dimensional limit. Moreover, we find that statistical fluctuations in finite pre-training data induce an implicit regularization. Finally, we identify a sharp phase transition of the generalization error governed by task structure. These results provide a framework for understanding how transformers learn to learn the task structure.

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。

↑