tattaka / jupyter_playground

Docker environment for ML

jupyter_playground

Overview

Jupyter nootbook server for remote machines. It works on docker.

Dependence

• docker
• docker-compose
• cuda>=9.0

Usage

1. Install jupyter-playground on remote machine.
   git clone https://github.com/tattaka/jupyter_playground.git

2. Put working directory in repository.
   mv /your/workspace jupyter_playground/
   or
   cp /your/workspace jupyter_playground/

3. Build jupyter_playground.
   cd jupyter_playground
./BUILD-DOCKER-IMAGE.sh

4. Run Docker Container.
   ./RUN-DOCKER-CONTAINER.sh

5. Exec jupyter_notebook (and tensorboard).
   docker exec username_jupyter_1 nohup jupyter notebook --allow-root >> jupyter.log 2>&1 &
docker exec username_jupyter_1 nohup tensorboard --logdir=/root/your/log/path

6. Login to remote machine from your client machine.
   ssh -L 8888:localhost:8888 (-L 6006:localhost:6006) -l user_name server.address