Hi, Jakub from neptune.ai here.
First of all, I want to say that I have massive respect for this project and all the people who contribute. It is essential in day-to-day for so many teams. Great job, and thank you!
A few months ago, some of our users asked if we could connect the experiment and metadata management capabilities of neptune.ai with all the great stuff that Kedro does.
The main things people wanted to improve were:
- Browsing, sorting, filtering pipeline executions on various metadata
- Comparing pipeline executions and nodes inside a pipeline on metrics, outputs, visualizations
- Display additional pipeline outputs like prediction images, ROC curves, or learning curves for deep learning models
And since neptune.ai is really good at those things, we figured it could be a nice integration.
It took some time, but we wanted to make it easy to use yet powerful.
Here is a video walkthrough:
More info and links are in the thread (couldn't fit it all in here).
I would absolutely love to hear what you think about it, how we could make it work for your workflow, where we need to improve it. If you have any questions, I am here!