Getting the current sample variable value so it can be used to compute the observed inputs in a secondary model likelihood

What do you mean by scaling?