Docker container for machine learning, runs jupyter notebook & tensorflow
Simply clone the repo, and run ./build.sh to build a container that is automatically tagged with the branch name. To also build the gpu version, run ./build.sh -g. The GPU version is built from Dockerfile_gpu.
The automatic builds for branch master can be found on Dockerhub as aviatr/tf_jupyter:latest and aviatr/tf_jupyter:gpu. For other branches, the auto-builds are tagged as aviatr/tf_jupyter:<branchname> and aviatr/tf_jupyter:gpu-<branchname>.
By default, the container runs jupyter notebook on port 8888, and can be launched simply by running the ./run.sh script. To use the nvidia-gpu runtime, you can pass the -g option, and to use a different tag other than master, you can use the -t <tagname> option. For more info, run ./run.sh -h.
If you would like to put this in the cloud somewhere, create a private folder, and replace the sha-1 hashed password in pwd.txt file (currently password) with your own password. Furthermore, add your ssl keys underneath private. You can then launch the container with the script ./run.sh which will mount the private folder in the docker container where the jupyter notebook expects to find them, and also passes your password to the jupyter notebook.
Dark theme for Jupyter Notebook/iPython 4 is created by Theodore Pak.