Hello,
I’m trying to use tensorflow on the rpbs jupyterhub servers, but I am faced with the following message :
“Could not find cuda drivers on your machine, GPU will not be used.” and the GPUs aren’t identified by tensorflow.
Is there any way to fix this (installing drivers in a conda env…?)
Hello,
Did you select a GPU in the jupyterhub form ?
Yes, I did select a GPU. It is detected when I run the nvidia-smi command, but I don’t know what I need to do in order for it to be recognized by tensorflow.
Have you tried to use tensorflow-gpu/2.6.2 module? In principle it should contain all the necessary libraries. If you could provide a minimal example, I could try to help.
Best wishes
Magali