Behrens' Bayesian Learner Model : How to share parameters between steps?

[ricardoV94]

@dddd1007

But when I add more steps, I still meet -inf logp.

That’s surprising. Do you see any r equal 0 or 1? You can remove the observations and just sample from the prior to see what they look like.

Can you print bayesian_lerner_model.check_test_point, to see where the -inf are coming from?