When running sample_posterior predictive() the model takes a lot of time. Much more than the sample method to get the posterior. Is this normal? I’m using pymc4 due to environment constraints.
Posterior predictive is way way faster in PyMC 5 thanks to speed up posterior predictive sampling by OriolAbril · Pull Request #6208 · pymc-devs/pymc · GitHub
I was trying to avoid moving to that version but maybe that’s the only way Thanks!