Installing Tensorflow on Windows



Installing tensorflow is as simple as running a pip command:

for CPU Version: “pip install tensorflow”

for GPU Version: “pip install tensorflow-gpu”.

If you haven’t installed/configured python yet and cannot run the above commands read further.

Python Configuration on windows

The best and easiest way to install Python on windows is using WinPython

Winpython is a free Python-distribution for Windows platform, including prebuilt packages for Scientific Python (includes most commonly used libraries like numpy,scipy, pandas.. and many more)

Install latest python 3.6 from the below sourceforge links



Once it is installed it contains many utilities like Spyder, Control Panel, Interpreter.

Open WinPython Command Prompt and use the pip command to install any python libraries



Tensorflow CPU installation

Open Winpython command prompt and type

“pip install tensorflow”

This will install tensorflow cpu only version. CPU version takes 10-20 times more training time than GPU. i.e. if you are trying to train a model on a dataset you will spend huge amount of time waiting for your program to complete the execution. Its recommended to use GPU if you want to work on real-time projects. CPU version is good for a very small data set or for testing or for learning the concepts of deep learning programming.

If you have installed successfully you can test in python console by typing

import tensorflow

if you donot see any errors, your tensorflow is ready to use.

Tensorflow GPU Installation

Open Winpython command prompt and type

“pip install tensorflow-gpu”

This will install tensorflow GPU version. The gpu version requires additional installations based on your graphics card and operating system. Read installing tensorflow-gpu on windows for complete instruction


Leave a Reply

Your email address will not be published. Required fields are marked *