How can i sample from an mcmc posterior?

so ive been following this tutorial:

https://pyro.ai/examples/bayesian_regression_ii.html

but I got stuck toward the end. We draw 1000 samples and come up with a posterior distribution using MCMC, however, how do I go abut making predictions from this point?

i found this discussion:

but it seems to be discontinued.

please help me! ty

@patc Have you tried to use Predictive? Also, I think baseball example will be helpful.