Skore lets you "Own Your Data Science." It provides a user-friendly interface to track and visualize your modeling results, and perform evaluation of your machine learning models with scikit-learn.
We can offer sane defaults regarding what's trusted with skops, and if we want to, we can offer a migration from an older version of sklearn to a new one, which is what I worked on already.
We can start running a bunch of tests with usual pipelines using sklearn and other third party library estimators, to see what comes up that needs to be trusted. That list then needs to be maintained across library versions, and you need to manually check those libraries to make sure trusting them doesn't open paths to exploits.
tuscland
commented
1 week ago
As recently discussed here:
We can offer sane defaults regarding what's trusted with skops, and if we want to, we can offer a migration from an older version of sklearn to a new one, which is what I worked on already.
We can start running a bunch of tests with usual pipelines using sklearn and other third party library estimators, to see what comes up that needs to be trusted. That list then needs to be maintained across library versions, and you need to manually check those libraries to make sure trusting them doesn't open paths to exploits.