This is a docker image that I used for running projects of the Udacity Deep Learning nanodegree. It is based on Ubuntu 18.04 and contains all of the necessary dependencies, such as conda, PyTorch, Jupyter, numpy and etc.

Prerequisites

• Docker
• Nvidia Container Toolkit

Usage

First, we need to build a Docker image:

• clone this repository and proceed to the root directory of the repository.
• run the build script.

When you need to run code inside the container:

• on the host filesystem, proceed to the directory containing the target source code.
• launch the run script. It will start the container and bring you to the bash session inside the container, with the host's working directory mounted as /workdir inside the container.
• to start Jupyter, enter the notebook command inside the container's bash session. It will start Jupyter server listening on the localhost:8888 address.
• to stop the container, simply exit the container's bash session.

Contributing

If you'd like to improve the usability of the container or install new dependencies inside the docker image, please make a pull request.

License

MIT