This project was originally made by @tobiasnorlund and slightly modified by me (Filip Cornell).

This repo constitutes a skeleton for a typical ML/DS project. Docker is a first class citizen and can be customized by editing the provided Dockerfile.

When starting a new project, please do the following:

1. On GitHub, create your own repository from this template by clicking the "Use this template" button
2. Update DOCKER_IMAGE_NAME in scripts/start.sh
3. Build and start a docker container:

./scripts/start.sh [--gpu] [--notebook] [--tensorboard] [-v|--mount /host/path:/container/path] [--detach]

5. Start a development container in VS Code: There are two ways this can be done.
   • Attach to the already running container (preferred when container is running on remote host)
     • In VS Code, install the Remote-Containers extention
     • Run Remote-Containers: Attach to Running Container... (F1). Select the newly created container
     • In the Explorer pane, click Open Folder and type the workspace directory (by default mounted to /workspace)
   • Let VS Code manage the container (preferred for local development)
     • In VS Code, install the Remote-Containers extention
     • Update name in .devcontainer/devcontainer.json to the value of DOCKER_IMAGE_NAME
     • Run Remote-Containers: Reopen in container (F1). Select the newly created container