Based on PS-Lite
Welcome to cite our paper if using our parameter servers design.
We will soon provide a soild version of the FluentPS design in Plato. It will support Pytorch and Tensorflow as a federated learning system. If you have any questions, please contact us. We will reply to it as soon as possible.
@INPROCEEDINGS{8891038,
author={Yao, Xin and Wu, Xueyu and Wang, Cho-Li},
booktitle={2019 IEEE International Conference on Cluster Computing (CLUSTER)},
title={FluentPS: A Parameter Server Design with Low-frequency Synchronization for Distributed Deep Learning},
year={2019},
volume={},
number={},
pages={1-12},
doi={10.1109/CLUSTER.2019.8891038}}