Tensorflow Virtualenv PyCharm

• Is PyCharm your python IDE?
• Did you install tensorflow using a virtual environment?

If so, this is a way to combine both, your tensorflow code running in a virtual environment used by PyCharm and from command line.

This article assumes you installed Python3 and pip3 already.

First step would be to create a new virtual environment using PyCharms Project Interpreter dialog for your existing project:

The settings symbol on the left upper corner allows you to initiate a new virtual environment creation:

As in the tensorflow installtion guide you should inherit global site-packages:

After the virtual environment is created, I did test it first from using command line:

# optional
$ cd ~/tensorflow_pycharm
# activate the virtual env
$ source ~/tensorflow_pycharm/bin/activate
# deactivate the virtual env after your prompt should have changed
(tensorflow)$ deactivate

Now its time to define & install the correct tensorflow binary to install as defined in the tensorflow installtion guide. In my case it would be:

$ cd ~/tensorflow_pycharm
$ export TF_BINARY_URL=https://storage.googleapis.com/tensorflow/linux/cpu/tensorflow-0.10.0rc0-cp34-cp34m-linux_x86_64.whl
$ sudo pip3 install --upgrade $TF_BINARY_URL

This virtual environment can then be used in a PyCharm project run configuration:

Select the just created virtual environment in case it is not preselected yet:

Set the script to be run. The intro tensorflow script (from tensorflow introduction) is used for this post:

No you are ready to run your script via virtual environment from both, PyCharm and command line:

• PyCharm:

  

• command line:

  

related links:

• http://stackoverflow.com/questions/37447906/how-to-get-virtualenv-tensorflow-to-work-in-pycharm
• https://www.jetbrains.com/help/pycharm/2016.1/configuring-python-interpreter-for-a-project.html#d29405e148
• https://www.jetbrains.com/help/pycharm/2016.1/adding-existing-virtual-environment.html
• https://www.youtube.com/watch?v=X3yTxWem7iY