I recently upgrade TensorFlow to version 2.15.0, after installation the legend “You may see slightly different numerical results due to floating-point round-off errors from different computation orders. To turn them off, set the environment variable TF_ENABLE_ONEDNN_OPTS=0
.”
I have nvidia-smi setup
Why is that and how do I set that variable up?
Hi @Julio_Torres, If you are using Windows Command Prompt, you can set environment variable as
set TF_ENABLE_ONEDNN_OPTS=0
If you are using linux you can set the environment variable as
export TF_ENABLE_ONEDNN_OPTS=0
Thank you.
i set TF_ENABLE_ONEDNN_OPTS=0 ine my enve ,but stil don’t work…
open cmd or powershell as admin and paste this command: setx TF_ENABLE_ONEDNN_OPTS 0
if you did it correctly it shows a success message