Google Cloud config issue : Jupyter loads wrong Python kernel

njgroene · July 4, 2019, 6:08am

Hi everyone,

I’m having trouble getting the 2019 course to work on Google Cloud. It seems that the Jupyter notrebook loads the wrong Python kernel.

I’ve followed these steps to set u GCP : https://course.fast.ai/start_gcp.html
The fastai module doesn’t seem to be recognised from within the jupyter notebook (“no module named fastai”) because it loads the wrong python kernel.
The fastai install seems to be linked to Python 3.7.3, in which fastai works from the jupyter terminal, but jupyter loads Python 3.5.3 and thus can’t find any of the installed libraries

I’ve tried the following:

Installing the Jupyter dependencies manually as describe here https://docs.fast.ai/install.html#jupyter-notebook-dependencies
Installing the nb_conda_kernels package to add all environments (as seen on stackoverflow)

Any suggestions would be more than welcome, I’m really stuck on this one

Cheers
Nick

njgroene · July 4, 2019, 7:20am

Update:
I’ve solved this issue by installing a new jupyter kernel based on the active environment using

python -m ipykernel install --user --name myenv --display-name "Python (myenv)"