INSTALL.txt

Python 3 is recommended for this lab. Python 2 works as well if you adapt the installation instructions.

Installation instructions for straightforward pip install below.

If you are a power user under a specific Python environment ((virtualenv, anaconda,
docker), please visit tensorflow.org and follow the Python 3 instructions.

MacOS:
	# If you do not have it already, install git from https://git-scm.com/download/mac
	# Install the latest version of python 3 from https://www.python.org/downloads/
	# and run this on the command line so that python3 can access https:// URLs:
	sudo /Applications/Python\ 3.6/Install\ Certificates.command
	# Now you can install tensorflow and matplotlib
	pip3 install --upgrade tensorflow
	pip3 install --upgrade matplotlib

Ubuntu/Linux:
	sudo -H apt-get install git
	sudo -H apt-get install python3
	sudo -H apt-get install python3-matplotlib
	sudo -H apt-get install python3-pip
	sudo -H pip3 install --upgrade tensorflow
	# you might alo need to upgrade matplotlib, the version pulled by
	# apt-get is sometimes stale (but comes with the gfx backend)
	sudo -H pip3 install --upgrade matplotlib

Windows:
	Install Anaconda, Python 3 version: https://www.continuum.io/downloads#windows
	Anaconda comes with matplotlib built in.
	In the Anaconda shell type: pip install --upgrade tensorflow
		If you get the error "Could not find a version that satisfies the requirement (...)" try the following alternative:
		conda config --add channels conda-forge
		conda install tensorflow

TEST YOUR INSTALLATION:
    git clone https://github.com/GoogleCloudPlatform/tensorflow-without-a-phd.git
    cd tensorflow-without-a-phd/tensorflow-mnist-tutorial
    python3 mnist_1.0_softmax.py
    => A window should appear displaying a graphical visualisation and you should also see training data in the terminal.