Get our free extension to see links to code for papers anywhere online!

Add to Chrome

Add to Firefox

Picture for Yuchen Zhong

Yuchen Zhong

dPRO: A Generic Profiling and Optimization System for Expediting Distributed DNN Training

May 18, 2022
Hanpeng Hu, Chenyu Jiang, Yuchen Zhong, Yanghua Peng, Chuan Wu, Yibo Zhu, Haibin Lin, Chuanxiong Guo

Figure 1 for dPRO: A Generic Profiling and Optimization System for Expediting Distributed DNN Training

Figure 2 for dPRO: A Generic Profiling and Optimization System for Expediting Distributed DNN Training

Figure 3 for dPRO: A Generic Profiling and Optimization System for Expediting Distributed DNN Training

Figure 4 for dPRO: A Generic Profiling and Optimization System for Expediting Distributed DNN Training

* Accepted by MLSys 2022

Access Paper or Ask Questions

Compressed Communication for Distributed Training: Adaptive Methods and System

May 17, 2021
Yuchen Zhong, Cong Xie, Shuai Zheng, Haibin Lin

Figure 1 for Compressed Communication for Distributed Training: Adaptive Methods and System

Figure 2 for Compressed Communication for Distributed Training: Adaptive Methods and System

Figure 3 for Compressed Communication for Distributed Training: Adaptive Methods and System

Figure 4 for Compressed Communication for Distributed Training: Adaptive Methods and System

Access Paper or Ask Questions

Thin Structure Estimation with Curvature Regularization

Sep 16, 2015
Dmitrii Marin, Yuri Boykov, Yuchen Zhong

Figure 1 for Thin Structure Estimation with Curvature Regularization

Figure 2 for Thin Structure Estimation with Curvature Regularization

Figure 3 for Thin Structure Estimation with Curvature Regularization

Figure 4 for Thin Structure Estimation with Curvature Regularization

* The IEEE International Conference on Computer Vision (ICCV), 2015, pp. 397-405
* D. Marin, Y. Zhong, M. Drangova, Y. Boykov. Thin Structure Estimation with Curvature Regularization. International Conference on Computer Vision (ICCV), Santiago, Chili, December 2015, to appear

Access Paper or Ask Questions