Undertanding difference between numpyro and pyro for time series forecasting

Relatively new to pyro and numpyro
I’m trying to understand the difference between how time series forecasting models in pyro and numpyro are defined and what their main differences are and how they deal with different input lengths for training and prediction?

Taking the example from the univariate time series model Forecasting I: univariate, heavy tailed — Pyro Tutorials 1.8.6 documentation
It appears that the class uses this method to incorporate the predicted values in the model under the hood: Forecasting — Pyro documentation

When doing forecasting in numpyro, is that what is being done with the transition method and scan, as in the numpyro AR2 example?

Sorry for the question