Tensorflow Environment Setup

• Install docker-machine & virtualbox

  • brew install docker-machine
  • install virtualbox by click here

• Create virtual machine

  • docker-machine create -d virtualbox --virtualbox-memory 8196 tensorflow

• Look the ip of the virtual machine

  • docker-machine ip tensorflow

• Add this command into you environment file

  • eval $(docker-machine env tensorflow)

• Running the Docker container from the TW AI Club repository

  • docker run -d -p 8888:8888 -p 6006:6006 -v {your-path}/:/notebooks --name tensorflow gcr.io/tensorflow/tensorflow:1.5.0-rc0-py3
  • if you don't have tensorflow docker image, just docker pull gcr.io/tensorflow/tensorflow:1.5.0-rc0-py3

• Go to: http://localhost:8888, and then you can do everything you want

• See Local Url and Token: git logs tensorflow

• You also can use other ways to setup environment, just reminder please provide more memory as the calculation need.

• And there is a link about this course in udacity. You can get more information from it.

Step

• Run udacity/1_notmnist.ipynb and then it will download the data we need and stored in data folder
• the study process: lr -> dnn -> cnn -> udacity