JupyterLab Kernels
- Simon Alexander
Here is a list of available kernels that can be selected for a jupyter notebook:
Note: You can not use the module system from the Beartooth cluster within a Jupyer Notebook - any environment you require needs to run within a specific kernel.
Kernels
Name | Description | GPU Enabled? |
|---|---|---|
Python 3 | Python 3.9.5 | N |
Python 3.11.0 | Python 3.11.0 | N |
PyTorch 1.13.1 | Python 3.10.9 | Y |
Dask 2023.8.1 | Python 3.11.4 / pandas 2.0.3 / numpy 1.25.2 | N |
R4.2.2 | A basic R 4.2.2 environment. | N |
The following kernels relate to Intel’s Distribution for Python and AI Analytics Toolkit:
Name | Description | GPU Enabled? |
|---|---|---|
Intel Python3 Base 23.0.0 | Python 3.9.15 | N |
Intel Modin 0.17.0.1 | Python 3.9.15 | N |
Intel PyTorch 1.13.0.0 | Python 3.9.15 | Y |
Intel TensorFlow 2.11.0 | Python 3.9.15 | N |
Intel TensorFlow 2.12 | Python 3.10.10 | N |
Finding available packages
Conda: If the kernel has been created from a conda environment, then you can run conda list from a cell to find all the packages/libraries within it:
Python: Call pip list -v from a cell will list all the python packages installed via the conda environment AND any packages you have installed yourself via pip install listed with a location in your home ~/.local/lib/<python-version> folder.
Look at the Location and Installer columns to see what is directly available within the kernel and what has manually been installed. For example:
Package Version Location Installer
----------------------------- ------------------- --------------------------------------------------------- ---------
...
entrypoints 0.3 /apps/s/jupyterlab/miniconda3/lib/python3.9/site-packages conda
flatbuffers 23.1.4 /pfs/tc1/home/salexan5/.local/lib/python3.9/site-packages pip
...R: Call installed.packages() from a cell will list all the R packages installed via the conda environment AND any packages you have installed yourself via install.packages('<package-name>') listed with a location in your home ~/R/<architecture/<version> folder.
Look at the LibPath column to see what is directly available within the kernel and what has manually been installed. For example:
Package LibPath Version
clipr /pfs/tc1/home/salexan5/R/x86_64-conda-linux-gnu-library/4.2 0.8.0
base /apps/u/opt/jupyter_kernels/r4.2.2/lib/R/library 4.2.2