CSpace

浏览/检索结果: 共3条,第1-3条 帮助

限定条件    
已选(0)清除 条数/页:   排序方式:
Mortar-FP8: Morphing the Existing FP32 Infrastructure for High-Performance Deep Learning Acceleration 期刊论文
IEEE TRANSACTIONS ON COMPUTER-AIDED DESIGN OF INTEGRATED CIRCUITS AND SYSTEMS, 2024, 卷号: 43, 期号: 3, 页码: 878-891
作者:  Li, Hongyan;  Lu, Hang;  Li, Xiaowei
收藏  |  浏览/下载:7/0  |  提交时间:2024/05/20
Deep learning accelerator  deep neural network (DNN)  fp8 format  
Double Precision is not Necessary for LSQR for Solving Discrete Linear Ill-Posed Problems 期刊论文
JOURNAL OF SCIENTIFIC COMPUTING, 2024, 卷号: 98, 期号: 3, 页码: 30
作者:  Li, Haibo
收藏  |  浏览/下载:8/0  |  提交时间:2024/05/20
Mixed precision  Linear ill-posed problem  Regularization  LSQR  Roundoff unit  Semi-convergence  
Fpar: filter pruning via attention and rank enhancement for deep convolutional neural networks acceleration 期刊论文
INTERNATIONAL JOURNAL OF MACHINE LEARNING AND CYBERNETICS, 2024, 页码: 13
作者:  Chen, Yanming;  Wu, Gang;  Shuai, Mingrui;  Lou, Shubin;  Zhang, Yiwen;  An, Zhulin
收藏  |  浏览/下载:12/0  |  提交时间:2024/05/20
Neural network  Model compression  Filter pruning  Attention  Rank enhancement  CNNs