[Feature] Add a monitoring callback to MIPROv2

[dsblank]

What feature would you like to see?

I'd like to have the ability for online monitoring of an optimizer. I guess I could add a logger, but that seems a roundabout way and requires parsing the log messages.

If there is not a better way, I'd be glad to make a PR to add a callback parameter to the class and call it when things like scores are computed and logged.

Also, I don't know if it is possible to get learned internal details like:

But if we can use the callback to log such details, that would also be great in monitoring and understanding the optimizer.

Would you like to contribute?

• Yes, I'd like to help implement this.
• No, I just want to request it.

Additional Context

No response

[dsblank]

Am I missing something, or would this be a good way to monitor the MIPRO processing?

[TomeHirata]

Hi, @dsblank. Thank you for the feature request. We are working on improving the traceability of DSPy optimizers, which will be released soon.