arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2106.02205cs.LGcs.AIquant-ph
收藏

基于矩阵乘积算子实现预训练语言模型压缩的轻量级微调

Enabling Lightweight Fine-tuning for Pre-trained Language Model Compression based on Matrix Product Operators

Gaoling School of Artificial Intelligence, Renmin University of China(中国人民大学高瓴人工智能学院);Renmin University of China(中国人民大学);Beijing Academy of Artificial Intelligence(北京智源人工智能研究院);School of Information, Renmin University of China(中国人民大学信息学院)

Peiyu Liu, Ze-Feng Gao, Wayne Xin Zhao, Z. Y. Xie, Zhong-Yi Lu, Ji-Rong Wen

摘要:

提出了一种基于量子多体物理矩阵乘积算子的预训练语言模型压缩方法。它可以将原始矩阵分解为中心张量(包含核心信息)和辅助张量(只有一小部分参数)。在此基础上,提出了一种仅通过辅助张量更新参数的微调策略,并设计了一种基于MPO的堆叠网络结构优化算法.我们的方法可以应用于原始或压缩PLM中的一般方式,从而得到一个更轻的网络,并显着减少了参数进行微调。大量的实验证明了该方法在模型压缩中的有效性,特别是微调参数的减少(平均减少91%)。

英文摘要:

This paper presents a novel pre-trained language models (PLM) compression approach based on the matrix product operator (short as MPO) from quantum many-body physics. It can decompose an original matrix into central tensors (containing the core information) and auxiliary tensors (with only a small proportion of parameters). With the decomposed MPO structure, we propose a novel fine-tuning strategy by only updating the parameters from the auxiliary tensors, and design an optimization algorithm for MPO-based approximation over stacked network architectures. Our approach can be applied to the original or the compressed PLMs in a general way, which derives a lighter network and significantly reduces the parameters to be fine-tuned. Extensive experiments have demonstrated the effectiveness of the proposed approach in model compression, especially the reduction in finetuning parameters (91% reduction on average).

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。

↑