当前位置:网站首页>A survey on model compression for natural language processing (NLP model compression overview)
A survey on model compression for natural language processing (NLP model compression overview)
2022-06-24 16:32:00 【Zhiyuan community】
author :Canwen Xu, Julian McAuley
brief introduction : With Transformer And pre training technology , natural language processing (NLP) Great progress has been made in the application of . However ,Transformer High energy consumption and long reasoning delay hinder NLP Into a broader scene , Including edge and mobile computing . Effective NLP The purpose of the study is to comprehensively consider the calculation , The whole life cycle of time and carbon emissions NLP, Including data preparation , Model training and reasoning . In this review , The author focuses on the reasoning stage , And review NLP Current situation of model compression , Including benchmark 、 Indicators and methods , The last author also The current obstacles and future research directions are summarized .



Paper download :https://arxiv.org/pdf/2202.07105
边栏推荐
- Abnormal dockgeddon causes CPU 100%
- Where is the most formal and safe account opening for speculation futures? How to open a futures account?
- 2021-05-01: given an ordered array arr, it represents the points located on the X axis. Given a positive number k
- How to open a futures account safely? Which futures companies are more reliable?
- Istio FAQ: failed to resolve after enabling smart DNS
- Pageadmin CMS solution for redundant attachments in website construction
- A troubleshooting of golang memory leak
- Global and Chinese market of training dance clothes 2022-2028: Research Report on technology, participants, trends, market size and share
- 2021-05-03: given a non negative integer num, how to avoid circular statements,
- C. K-th Not Divisible by n(数学+思维) Codeforces Round #640 (Div. 4)
猜你喜欢

B. Ternary Sequence(思维+贪心)Codeforces Round #665 (Div. 2)

My network relationship with "apifox"

C. Three displays(动态规划)Codeforces Round #485 (Div. 2)

There are potential safety hazards Land Rover recalls some hybrid vehicles
![[cloud native | kubernetes chapter] Introduction to kubernetes Foundation (III)](/img/21/503ed54a2fa14fbfd67f75a55ec286.png)
[cloud native | kubernetes chapter] Introduction to kubernetes Foundation (III)
![[go] concurrent programming channel](/img/6a/d62678467bbc6dfb6a50ae42bacc96.jpg)
[go] concurrent programming channel
MySQL Advanced Series: locks - locks in InnoDB

Problems encountered in the work of product manager

Ui- first lesson

Siggraph 2022 | truly restore the hand muscles. This time, the digital human hands have bones, muscles and skin
随机推荐
[download attached] installation and simple use of Chinese version of awvs
Cause analysis of the failure of web page live broadcast on demand RTMP streaming platform easydss streaming live broadcast
对深度可分离卷积、分组卷积、扩张卷积、转置卷积(反卷积)的理解
Kubernetes characteristic research: sidecar containers
Funny! Pictures and texts give you a comprehensive understanding of the effects of dynamics and mass
嵌入式开发基础之线程间通信
AI structured intelligent security video monitoring technology, supporting the protective umbrella of the reserve / wild animals
B. Ternary Sequence(思维+贪心)Codeforces Round #665 (Div. 2)
[tke] modify the cluster corendns service address
Global and Chinese markets of stainless steel barbecue ovens 2022-2028: Research Report on technology, participants, trends, market size and share
Pytorch 转置卷积
Customized Tile Map cut - based on Tencent map
Video structured intelligent analysis platform easycvr video recording plan function optimization / regularly delete expired videos
AI video structured intelligent security platform easycvr realizes intelligent security monitoring scheme for procuratorate building
[idea] dynamic planning (DP)
Tencent releases the full platform version of reasoning framework TNN, and supports mobile terminal, desktop terminal and server terminal at the same time
How does the effective date of SAP PP ECM affect the work order?
A new weapon to break the memory wall has become a "hot search" in the industry! Persistent memory enables workers to play with massive data + high-dimensional models
MySQL date timestamp conversion
Goby+awvs realize attack surface detection