Name: ps-lite
Owner: Distributed (Deep) Machine Learning Community
Description: A lightweight parameter server interface
Created: 2015-04-26 19:29:30.0
Updated: 2018-05-24 06:14:33.0
Pushed: 2018-05-02 06:44:09.0
Homepage: http://ps-lite.readthedocs.org
Size: 632
Language: C++
GitHub Committers
| User | Most Recent Commit | # Commits |
|---|
Other Committers
| User | Most Recent Commit | # Commits |
|---|
A light and efficient implementation of the parameter server framework. It provides clean yet powerful APIs. For example, a worker node can communicate with the server nodes by
Push(keys, values): push a list of (key, value) pairs to the server nodesPull(keys): pull the values from servers for a list of keysWait: wait untill a push or pull finished.A simple example:
d::vector<uint64_t> key = {1, 3, 5};
d::vector<float> val = {1, 1, 1};
d::vector<float> recv_val;
::KVWorker<float> w;
Wait(w.Push(key, val));
Wait(w.Pull(key, &recv_val));
More features:
ps-lite requires a C++11 compiler such as g++ >= 4.8. On Ubuntu >= 13.10, we
can install it by
apt-get update && sudo apt-get install -y build-essential git
Instructions for older Ubuntu, Centos, and Mac Os X.
Then clone and build
clone https://github.com/dmlc/ps-lite
s-lite && make -j4
ps-lite provides asynchronous communication for other projects:
We started to work on the parameter server framework since 2010.
The first generation was designed and optimized for specific algorithms, such as logistic regression and LDA, to serve the sheer size industrial machine learning tasks (hundreds billions of examples and features with 10-100TB data size) .
Later we tried to build a open-source general purpose framework for machine learning algorithms. The project is available at dmlc/parameter_server.
Given the growing demands from other projects, we created ps-lite, which provides a clean data communication API and a
lightweight implementation. The implementation is based on dmlc/parameter_server, but we refactored the job launchers, file I/O and machine
learning algorithms codes into different projects such as dmlc-core and
wormhole.
From the experience we learned during developing dmlc/mxnet, we further refactored the API and implementation from v1. The main changes include
less library dependencies
more flexible user-defined callbacks, which facilitate other language bindings
let the users, such as the dependency engine of mxnet, manage the data consistency