1

i'm created a conda environment

conda create -n tfgpu tensorflow-gpu
conda activate tfgpu

but now after running

(tfgpu) > jupyter notebook

I can't import tensoflow as tf, notebook doesn't load this environment! Why it's happen?

Mike
  • 479
  • 5
  • 15

1 Answers1

2

My problem is solved after:

https://stackoverflow.com/a/44786736/12370909

conda create -n tfgpu tensorflow-gpu
conda activate tfgpu
python -m ipykernel install --user --name tfgpu --display-name "Python (tfgpu)"
jupyter notebook

Then you could be able to create a new Notebook using your environment by selecting "Python (tfgpu)" from the drop-down list.

Defne Tunçer
  • 56
  • 3
  • is it's okay that setting stored in "user" data directory, so they LEAK OUT of tfgpu environment? – Mike Nov 30 '19 at 12:31
  • Or, may be, I have two choices: 1) install jupyter into tfgpu environment (that spends additional space) 2) use anaconda's default jupyter notebook and accept the fact that "system-wide package stores it's setting system wide" – Mike Nov 30 '19 at 12:34