Standard practice for writing Conda environment file with support for TensorFlow CPU/GPU

Asked Jun 06 '20 at 11:19

Active Jun 06 '20 at 11:19

Viewed 52 times

I would like to add an environments.yml file to my repo. Since I don't know whether the end user will be using the CPU or GPU version of TensorFlow, I am unsure how to proceed. My thoughts:

I could create two environment files, each one for CPU/GPU
I create one environment file for the more commonly used GPU variant and hope the users figure out what's going on

Is there a standard approach to this or a better way I haven't considered?

asked Jun 06 '20 at 11:19

Tim Hargreaves

Do you know what OS user is using? – Frenzy Kiwi Jun 06 '20 at 11:33
@FrenzyKiwi It could be any. I want the code to be accessible to as many users as reasonably possible – Tim Hargreaves Jun 06 '20 at 14:49
1

You'd have to script. Such interaction is not directly available on conda. – Frenzy Kiwi Jun 17 '20 at 06:38

Standard practice for writing Conda environment file with support for TensorFlow CPU/GPU

0 Answers0