How to properly do out-of-sample prediction for hierarchical model

Why did you say that can’t be right? You shouldn’t have to change RV dims that were already present in the original model, even if not all are used for posterior predictive