arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2209.01667cs.LGcs.CL
收藏

深度学习中的稀疏专家模型综述

A Review of Sparse Expert Models in Deep Learning

Google Brain(谷歌大脑);Google Research(谷歌研究院)

William Fedus, Jeff Dean, Barret Zoph

摘要:

稀疏专家模型(Sparse Expert Models)是一个有着30年历史的概念,现在作为深度学习中的流行架构重新出现。这类体系结构包括专家混合、交换机Transformers、路由网络、基础层等,所有这些都具有统一的思想,即每个示例都由参数的子集作用。通过这样做,稀疏度使每个示例的计算中的参数计数成倍增加,从而允许非常大但有效的模型。由此产生的模型在不同领域(如自然语言处理、计算机视觉和语音识别)都有了显著的改进。我们回顾了稀疏专家模型的概念,提供了常见算法的基本描述,介绍了深度学习时代的进展,并通过强调未来工作的领域得出结论。

英文摘要:

Sparse expert models are a thirty-year old concept re-emerging as a popular architecture in deep learning. This class of architecture encompasses Mixture-of-Experts, Switch Transformers, Routing Networks, BASE layers, and others, all with the unifying idea that each example is acted on by a subset of the parameters. By doing so, the degree of sparsity decouples the parameter count from the compute per example allowing for extremely large, but efficient models. The resulting models have demonstrated significant improvements across diverse domains such as natural language processing, computer vision, and speech recognition. We review the concept of sparse expert models, provide a basic description of the common algorithms, contextualize the advances in the deep learning era, and conclude by highlighting areas for future work.

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。

↑