deeplearning-jupyterhub-docker
Docker file for deep learning and JupyterHub
Prerequisites
docker with nvidia support
Follow the instructions from the TensorFlow project on configuring Docker with GPU support.
Building and running
Build:
$ docker build -t deeplearning-jupyterhub-docker:latest -f Dockerfile .
Run:
$ docker run --gpus all -it -p 8888:8888 -p 6006:6006 -v $(realpath ~/workspace):/home/jovyan/workspace deeplearning-jupyterhub-docker:latest bash
Running remotely
Enable ssh port forwarding
$ ssh -L 8888:localhost:8888 username@domain
Updating conda environments
Conda environments are created from a yaml file. It is easiest to modify the environment from within the docker image.
Update the environment file when a dependency has been added or modified.
#