This repository contains information to create a docker image for jupyterlab. The main python library that is installed is DeepForest. It has been created to utilize GPUs in Cyverse Discover Environment. Also added to this docker image is a script in entry.sh that copies your .ssh and .gitconfig files from you personal data store to the working directory of the docker image.
The docker image is based on the jupyter/minimal-notebook image. It has python 3.11 installed and includes some command line tools like curl, git, and nano.
FROM quay.io/jupyter/minimal-notebook:lab-4.2.1
git clone [email protected]:cyverse-vice/jupyterlab-deepforest.git
cd jupyterlab-deepforest
docker build -t harbor.cyverse.org/vice/jupyter/deepforest:gpu-061424 .
docker run -it --rm --gpus=all -p 8888:8888 harbor.cyverse.org/vice/jupyter/deepforest:gpu-061424
Open your browser and go to https://localhost:8888
docker push harbor.cyverse.org/vice/jupyter/deepforest:gpu-061424
Go to the Cyverse DE and go to 'Tools'
Fill out the tool information as such
Go to the 'Tools' under the Admin. You need admin privaledges for this to be visible.
From the Tool you created, create an interactive Cyverse App
I created an app called 'JupyterLab Minimal GPU'
After launching the Cyverse App, open a terminal in the jupyterlab and type:
nvidia-smi -L
The output should say the name of the GPU (e.g., NVIDIA GeForce RTX 2080)