Install anaconda, Tenserflow GPU, Keras and pycharm on windows 10
Tenserflow gpu is a fast and powerful platform to build, train and maintain Machine Learning flows. Installing them on windows is full of troubles. We will have this page as cheat sheet to help quick installing.
Prerequisites:
Uninstall NVIDIA drives (you can install back later).
Install Visual studio community 2015:
It is a older version. But go ahead with it. You need to singup for download from here. Launch installer and just keep going next with default options.
Install anaconda:

It’s simple. Download anaconda with python 3.x from here. Launch installer and just keep going next with default options.
Install CUDA:
CUDA tool kit is the one that enables tenserflow to use GPU. Unlike CPU, GPUs have thousands of cores. Using those cores in parallel will make your computations run faster (specifically the matrix multiplications of Neural Network and other ML methods). Back to installation. Download installer from here. Launch the installer. Install everything using custom installation instead of express. Note: you may able to install latest CUDA too but I haven’t tried. I tried only CUDA 8.0.

Install Tenserflow gpu:
Goto anaconda prompt. You should be able to locate it @ C:\ProgramData\Microsoft\Windows\Start Menu\Programs\Anaconda2 (64-bit).
type:
> conda create -n tensorflow
> activate tensorflow
> pip install --ignore-installed --upgrade tensorflow-gpuInstall Keras:
In same conda prompt.
> pip install --upgrade kerasInstall Pycharm:
We are done with installing all packages. How about installing IDE (pycharm)? Just download community edition from here. Install with default options. Launch pycharm after installation. Create new project with “Existing interpreter”. Note: In previous steps, We have created new python environment with name tensorflow. Chose that python.exe and click create.

You are all set to program your first transerflow or keras project. All the Best!!
