IterationControl.jl is a lightweight package for controlling iterative algorithms, with a view to training and optimizing machine learning models. It builds on our earlier package EarlyStopping.jl.
This package is similar to LearningStrategies.jl. The API is a little simpler (at least to us) and it has been thoroughly tested.
IterationControl.jl is the engine behind a newly released iterative model wrapper for the MLJ machine learning platform, to be described in a separate announcement.
Hey, nice, I think we are new neighbors GitHub - mschauer/DynamicIterators.jl: Iterators with message passing and feedback loops
I’m very happy to see that people are coming up with various designs here.
Looks like I’m not the only one who felt constrained by iterators only having output and no input 
Thanks for sharing @ablaom @mschauer I’ll look into your packages with great interest!
@mschauer Hey, thanks for the feedback. Your package looks very nice and I’m sorry I had not known about it.
I will add a link to it from the readme.