Model progress with event.

[MayukhSobo]

Hello, I am stuck with a weird problem. I have a setup where based on the options selected in the web, I am choosing multiple algorithms like Neaural networks and sometimes Logistic Regression using SGDClassifier. I can always write a custom callback function where I can transmitt the values like epoch, validation loss and training progress to the web backend in case of neural network. But I want to achieve the same thing with logistic regression using SGDClassifer. I know that it works on multiple iterations. So if I can somehow get the current iteration and after every iter, if I can call a custom function, it would be really great.

[rth]

There is WIP Pull Request to add callbacks which would allow doing this in #16925

[glemaitre]

I am closing this issue since we already have a PR and associated issue and an entry in the scikit-learn road map.