Better verbose

[Qazalbash]

Researchers require gradient values throughout training to diagnose issues, particularly when encountering NaN values. Currently, flowMC does not provide access to these values in debug/verbose mode. We have implemented this functionality in our project, which can be reviewed in the following pull request: gwkokab/gwkokab#216. We propose adding this feature to flowMC. Do you consider this a beneficial enhancement to the project?

[kazewong]

This is something in the back of our head. Currently I have been busy on other stuff, but I expect there should be some push on several enhancements on flowMC in October when things are more calm for me

[Qazalbash]

Would you like me to push this enhancements?

[kazewong]

If there is a clean PR we can look at it, but just to be clear I don't have much time for PR review now, and I want to make sure any modification to flowMC is thoroughly reviewed. With the disclaimer, I would be delighted to see a PR if you can do that.