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: kokabsc/gwkokab#216. We propose adding this feature to flowMC. Do you consider this a beneficial enhancement to the project?
Researchers require gradient values throughout training to diagnose issues, particularly when encountering
NaNvalues. Currently,flowMCdoes 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: kokabsc/gwkokab#216. We propose adding this feature toflowMC. Do you consider this a beneficial enhancement to the project?