Reactive Optimizers: rewrite the optimizer's source code during training

This is really interesting and fun!

Reactive optimizers could actually be helpful if you used them to dynamically tweak training data mix and use them to make choices about RL environments (if you recall from the Mimo training dashboards, it’s quite tricky to figure out how to value and chose RL traces during post training)

I do that constantly in a lot of apps. Mostly I write a include file of a param table and retest under new conditions. It's called calibration, just that I compile the values in, and recompile.