/jupyterlab-docker

Docker images for the machine learning environment. This project cover Python and ML packages, GPU support, C++.

Primary LanguageDockerfileMIT LicenseMIT

JupyterLab Dockerfiles

This is JupyterLab's Dockerfiles, create and deploy them to my Docker Hub.

Check the full list of tags for the available images.

Image Description

Base OS is Ubuntu and use the Miniconda3 environment.

  • Ubuntu
    • 16.04
    • 18.04
  • GPU
    • CUDA 10.1
  • Miniconda3
    • Python 3.7.3
      • numpy
      • pandas
      • matplotlib
      • pillow
  • Machine Learning (ml)
    • scipy
    • scikit-learn
    • nltk
    • opencv-python
    • tensorflow==1.14.0
    • pytorch==1.1.0
  • C++
    • xeus-cling

Tag Rules

  • latest: same as xenial-py3
  • latest-gpu: same as xenial-gpu-py3
  • xenial: Ubuntu16.04
  • bionic: Ubuntu18.04
  • -py3: Python 3.7.3
  • -gpu: enable to use CUDA
  • -ml: pre-installed machine learning packages
  • -cpp: enable to use C++ environment from xeus-cling which is the jupyter kernel for the C++

Running Containers

$ docker run -it --rm -p 8888:8888 gzupark/jupyterlab

Run a JupyterLab server, navigate to localhost:8888 in your browser. Then, type default jupyter password which is "jupyterlab", you can see the guide if you want to change it through tutorial_change_passwd.ipynb in the workspace.

Start a machine learning packages and C++ environment:

$ docker run -it --rm -p 8888:8888 gzupark/jupyerlab:py3-ml-cpp

Want to mount with your local machine:

$ docker run -it --rm -v $(realpath ~/project):/workspace -p 8888:8888 gzupark/jupyterlab

Want to use GPU version:

$ docker run -it --rm --runtime=nvidia -p 8888:8888 gzupark/jupyterlab:latest-gpu