Binomial GLM (Stat Rethinking)

Hello everyone,

I’ve been going through exercises of Stat Rethinking (Winter 2022) and got problem with one of the models:

Expected:
parameters roughly the same as in exercise 1 solution

Actual:
practically only one parameter converges

(kinda the same happends if i do .expand([gender_card, discipline_card])))

Am I misapplying API somewhere?

Indeed, I misused the API in two different places and it failed silently at me.

The corrected version, also rewritten in plate notation, is in the second .ipynb of the updated gist: