Hi, pm.Potential seems really handy to do the weighting. It works fine with using minibatches, as long as I use only one likelihood in my model. To be more specific the likelihoods all need to have observed data with the same row indexes.
Out of curiosity, what are the options to make it work with a model with more than one likelihood where each likelihood has a different set of observed data? Is it possible to just add the potential to one specific likelihood (although by looking into the model code I don’t think so)?
Is it possible in another way?