/Mxnet_Tutorial

Using Mxnet!!!!

Primary LanguagePython

Mxnet?

  • Flexible and Efficient Library for Deep Learning
  • Symbolic programming or Imperative programming
  • Mixed programming available (Symbolic + imperative)

Introduction

  • It is a tutorial that can be helpful to those who are new to the MXNET Deep-Learning Framework

Official Homepage Tutorial

Let's begin with

  • Required library and very simple code
import mxnet as mx
import numpy as np

out=mx.nd.ones((3,3),mx.gpu(0))
print(mx.asnumpy(out))
  • The below code is the result of executing the above code
<NDArray 3x3 @gpu(0)>
[[ 1.  1.  1.]
 [ 1.  1.  1.]
 [ 1.  1.  1.]]

Topic 1 : Symbolic Programming

Topic 2 : Imperative Programming

Development environment

  • os : window 10.1 64bit and Ubuntu linux 16.04.2 LTS only for tensorboard
  • python version(3.6.1) : anaconda3 4.4.0
  • IDE : pycharm Community Edition 2017.2.2 or visual studio code

Dependencies

  • mxnet-1.0.0
  • numpy-1.12.1, matplotlib-2.0.2 , tensorboard-1.0.0a7(linux) , graphviz -> (Visualization)
  • tqdm -> (Extensible Progress Meter)
  • opencv-3.3.0.10 , struct , gzip , os , glob , threading -> (Data preprocessing)
  • Pickle -> (Data save and restore)
  • logging -> (Observation during learning)
  • argparse -> (Command line input from user)
  • urllib , requests -> (Web crawling)

Author

JONGGON KIM GitHub

rlawhdrhs27@gmail.com