Skip to content
Go to file

bootstrap.pytorch is a high-level extension for deep learning projects with PyTorch. It aims at accelerating research projects and prototyping by providing a powerful workflow focused on your dataset and model.

And it is:

  • Scalable
  • Modular
  • Shareable
  • Extendable
  • Uncomplicated
  • Built for reproducibility
  • Easy to log and plot anything

Unlike many others, bootstrap.pytorch is not a wrapper over pytorch, but a powerful extension.

Quick tour

To run an experiment (training + evaluation):

python -m
       -o myproject/options/sgd.yaml

To display parsed options from the yaml file:

python -m
       -o myproject/options/sgd.yaml

Running an experiment will create 4 files, here is an example with mnist:

  • options.yaml contains the options used for the experiment
  • logs.txt contains all the information given to the logger
  • logs.json contains the following data: train_epoch.loss, train_batch.loss, eval_epoch.accuracy_top1, etc
  • view.html contains training and evaluation curves with javascript utilities (plotly)

To save the next experiment in a specific directory:

python -m
       -o myproject/options/sgd.yaml
       --exp.dir logs/custom

To reload an experiment:

python -m
       -o logs/custom/options.yaml
       --exp.resume last


The package reference is available on the documentation website.

It also contains some notes:

Official project modules



Contributions to this repository are welcome and encouraged. We also have a public trello board with prospect features, as well as an indication of those currently being developed. Feel free to contact us with suggestions, or send a pull request.

We use flake8 to perform early semantic checking of submitted code. After installing all the requirements in requirements.txt, please run the following to activate the pre-commit hooks for flake8: pre-commit install

To manually trigger the pre-commit checks for a file without creating a commit, you can run the following command: pre-commit run --files <>

bootstrap.pytorch was conceived and is maintained by Rémi Cadène and Micael Carvalho, with helpful discussions and insights from Thomas Robert and Hedi Ben-Younes. We chose to adopt the [very permissive] BSD-3 license, which allows for commercial and private use, making it compatible with both academy and industry standards.


Highlevel framework for starting Deep Learning projects (lightweight, flexible, easy to extend)




No releases published


No packages published