Essential reproducibility is done using mybinder.org. The service is a way to pack your repository into a free-to-run reproducible container.
Click here to run the notebook server :
To get started with the proposed solution for core reproducibility, we use user Docker+Conda+Binder. You can fork this template, and replace with your own solution.
To help out, we propose few checklists and present an effortless project. Read more in the section Project Goal
Use the following checklists to make sure your work is striving for reproducibility.
- Update the #binder section to use you repository link
- update /binder/enviroment.yml to add you software dependencies
- Add you code in the
/codefolder - Try to fullfill the following checklists:
- Here is a checklist for code reproducibility : Code
- Here is a checklist for method documentation : Method
- If you are publishing data : Data
- Joelle Pineau has an excellent checklist for machine learning reproducibility
This template repo is a simple and barebone solution. A better and didactic deep dive is presented in The turing way Guide for Reproducible Research .
Here is where one should have a description of your project goal. Bonus points if you include Hypothesis/Prediction/Conclusion.
In our case, our goal is to showcase a toy example of reproducibility. We have a code POC.ipynb that was run locally. We believe that we can have the code achieve the same result while running on [binder]
There are three options to run this repository and check results.
They are listed below from prefered to least prefered order.
Use binder to run a jupyter notebook server
Jump to code/POC.ipynb to run the code and verify the result.
What is the protocol to run the code?
- Run each cell of
code/POC.ipynbby pressing the play button
Use a computer with docker and docker-compose.
You will need docker and docker-compose installed at your development machine.
Use the scrip run_localy.sh to start a jupyter-lab with the environment.
Jump to POC.ipynb to run the code and verify the result.
Use conda and the provided environment, to set up things in your local computer manually.