Skip to content

Latest commit

 

History

History
33 lines (21 loc) · 1011 Bytes

README.md

File metadata and controls

33 lines (21 loc) · 1011 Bytes

Deep Learning with Docker

A boilerplate used to scaffold projects running on Nvidia-Docker (or even traditional Docker) with both source code and Jupyter notebooks.

Such projecs ensure code reproducibility (which is often an issue when using the Jupyter environment), while also increasing refactor capability from Jupyter code into the main Python directory.

Getting started

To create a new project from this template, install cookiecutter:

pip install cookiecutter

Then add the following to your .bashrc file:

EXPORT PATH=$HOME/.local/bin:$PATH

Then you can run cookiecutter as follows:

cookiecutter gh:rsayn/cookiecutter-dl-docker

Cookiecutter will prompt you for some configuration values, e.g. the project name, package requirements and whether to use CPU or GPU to run the container.

Note: requirements should be provided in a comma-separated form as follows:

sklearn,pipenv,numpy,pandas