/ML-project-template

MIT LicenseMIT

ML-project-template

A simple and flexible template to structure a Machine Learning project.

  pytorch-template/
  │
  ├── data/
  │   ├── processed/            <- Processed data via data scripts
  │   └── raw/                  <- Data as downloaded from the source
  │       └── README.md         <- Data details, e.g., URL, date of download, name, etc.
  │
  ├── saved/
  │   ├── log/                  <- Log information
  │   └── models/               <- Checkpoints and final models
  │
  ├── src/
  │   ├── model/                <- Source code for your own model
  │   └── scripts/
  │   │    ├── data/            <- scripts and programs to process data
  │   │    ├── visualization/   <- Scripts for visualisation of your outputs
  │   │    └── test/            <- The results of your analysis
  │   │
  │   └── utils/                <- Helpers and utilities used
  │
  ├── README.md                 <- description and how to use the project
  ├── Pipfile                   <- requirements for the dedicated environmnet
  ├── .pre-commit-config.yaml   <- configuration of pre-commit
  ├── LICENSE                   <- license details
  └── .gitignore

Project creation

pipenv install
pre-commit install
pipenv shell
pipenv -e . # install the project as editable

TODO

  • Implementation of a Run all script for complete reproducibility
  • Similar to rails to create the project template and associated environment: project new <project-new>
  • Docker containers for servers?

License

This project is licensed under the MIT License. See LICENSE for more details

Acknowledgements

Thanks to Cookiecutter Data Science project , Pytorch project template and Tania Sanchez Monroy's presentation.