Hierarchical model from Statistical Rethinking takes a long time to build

Questions
3

I will try the first 2 suggestions. Regarding point 3, I’m running what’s in the notebook and exponential priors on standard deviations worked just fine before all throughout different chapters of this book (including in the same notebook), so will not change that.

Here’s an example from the beginning of the same notebook that worked fine before.

d = pd.read_csv("Data/reedfrogs.csv", sep=",")

with pm.Model() as m_13_2:
    a_bar = pm.Normal("a_bar", 0.0, 1.5)
    sigma = pm.Exponential("sigma", 1.0)

    a = pm.Normal("a", a_bar, sigma, shape=n_tanks)
    p = pm.math.invlogit(a[tank])

    S = pm.Binomial("S", n=d.density, p=p, observed=d.surv)
    trace_13_2 = pm.sample(random_seed=RANDOM_SEED)