Docker - Anaconda and Tensorflow

Welcome to the anaconda/tensorflow image. This image was built to be used as a simple way to run a standard ML system with tensorflow.

Structure

The latest branch will (by default) be setup to use the latest version of python and tensorflow that are CPU based. This will allow for the system to use the standard ubuntu base image.

However there are tags that will be created for the gpu specific versions of the system. These will each be tagged as such.

So for every tensorflow example that exists a corresponding version will also exist that will just have the appended tag value of -gpu, for example:

    latest           ->    latest-gpu 
    python35         ->    python35-gpu
    python35-onbuild ->    python35-onbuild-gpu

Running

To use this image you will want to have a volume mounted at /notebooks, and expose port 8888.

    docker run -it --rm -p 8888:8888 -v $PWD:/notebooks mikewright/anaconda-tensorflow:latest-gpu

At this point you can open your browser to localhost:8888.

Building

If you want to have your own tools installed you may do so by creating a Dockerfile based on the onbuild tag. You will need to have your anaconda environment file created and named environment.yml, and you will want to set the environment variable CONDA_ENV to the name of the enviroment you created.

    -- environment.yml
    name: myenv
    dependencies:
      - jupyter
      - pymc 

    -- Dockerfile
    FROM mikewright/anaconda-tensorflow:latest-onbuild
    
    ENV CONDA_ENV myenv