Why do I need another conda environment from tensorflow?



I’m currently trying to start working with tensorflow.
I work with anaconda and I tried to install the tensorflow packages in the root environment but it always displays the message: "Several errors encountered".
When I looked it up it says the solution is to create another environment exclusively for tensorflow, I did and it worked. But I’d still like to know what the reason for this is.


I have had the same question when I started out. It seemed like it is the "correct" thing to do, so I just did it, but never understood why. After working with TensorFlow for 2 years now, and on multiple machines, I realised just how specific the set of its requirements is. Only a few versions of python are compatible with it, the same thing with numpy, and if you want to use NVIDIA GPUs, good luck figuring out the specific versions of cuda and cudnn.

You don’t want to have to tailor most of the python-related software on your machine to running tensorflow. In order to avoid breaking it whenever you install something that requires a higher version of numpy, for example, it is best to keep it in a separate environment. This way you have an isolated "container" that keeps everything just the way TensorFlow wants it, while still being able to use other software if needed.
Not to mention that there are several versions of TensorFlow and they all have different requirements.

Answered By – idinu

This Answer collected from stackoverflow, is licensed under cc by-sa 2.5 , cc by-sa 3.0 and cc by-sa 4.0

Leave A Reply

Your email address will not be published.

This website uses cookies to improve your experience. We'll assume you're ok with this, but you can opt-out if you wish. Accept Read More