in my PhD and work afterwards I have spent a ton of time reproducing and building new ML models.
To reduce the time it takes me to implement each new approach and have a better comparability I have developed a tool that has stored SOTA and my custom implementations in a standardized way.
To launch a new training I can then simply fill out the forms with dropdowns etc in my locally run web application.
I have started for computer vision, but would love to understand how that can be applied to your area.
It is still not perfect, you can run the netlify preview (in the ReadMe) so you do not have to run it locally.
Appreciate your feedback.