Installation Instructions for Windows

  1. Download and install Python 3.5.3 from You must use Python 3.5.3 for TensorFlow to install, this is the only version of Python they support on Windows. When installing, make sure you check the box “Add Python 3.5 to PATH”.
  2. You should choose the Windows x86-64 web-based installer or Windows x86-64 executable installer.
  3. If your Windows version is only 32 bit, choose the Windows x86 web or Windows x86 executable installer.
  4. Download the pre-compiled Scipy Binary from
  5. You almost certainly want scipy0.19.0cp35cp35mwin_amd64.whl, which is the binary for Python 3.5.3 running on Windows 64 bit.
  6. If your Windows install is only 32 bit, youwant scipy‑0.19.0‑cp35‑cp35m‑win32.whl
  7. Download the pre-compiled Numpy+mkl binary from This is a large file, it will take some time to download.
  8. You almost certainly want numpy‑1.12.1+mkl‑cp35‑cp35m‑win_amd64.whl, which is the binary for Python 3.5.3 running on Windows 64 bit.
  9. If your Windows install is only 32 bit, you want numpy‑1.12.1+mkl‑cp35‑cp35m‑win32.whl

NOTE: If you have both Python 2 and Python 3 installed on your system, verify pip is installing to the correct version pip with ‘pip –V’. You should see ‘pip 9.0.1 from c:\path\to\pip (Python 3.5)’ If you see (Python 2.7), try pip3 –V. If that reports (Python 3.5), whenever you see pip below, use pip3 (and python3 when running) instead.

  1. Open powershell, and navigate to where you downloaded the Numpy + mkl binary, and run ‘pip install --user my-numpy-binary.whl’
  2. Navigate to where you downloaded the Scipy binary, and run ‘pip install --user my-scipy-binary.whl’
  3. Install Tensorflow with ‘pip install --user --upgrade tensorflow’. This may take some time. If it appears to hang, press enter.
  4. Install Keras and Matplotlib with ‘pip install --user --upgrade kerasmatplotlib’. This may take some time.If it appears to hang, press enter.
  5. Install sklearn with ‘pip install --user --upgrade sklearn’.

Congratulations, you should be good to go. You can verify everything is installed correctly by running the test python file included with this document.

Installation instructions for OS X

(This install was tested on the following system: OS X 10.11.4, Python 3.6.1)

  1. Install Python 3 if it is not already installed. You can find the download on python.org, or the (recommended) way is to install using homebrew, a package manager for OS X. You can install homebrew at then do brew install python3from the command line. Do NOT use sudo when installing using brew.
  2. Once you have installed python3 (we are assuming you used homebrew), verify the versions are correct with pip3 -V and python --version
  3. Install numpy and scipy with pip install --user --upgradenumpyscipy
  4. Install sklearn withpip install --user --upgrade sklearn
  5. Install tensorflow with pip install --user --upgrade tensorflow
  6. Install keras withpip install --user --upgrade keras
  7. Install matplotlib withpip install --user --upgradematplotlib
  8. You should now be good to go. Run test.py to verify everything installed correctly.