Hi!

I have a custom pymc model that I'd like to port to Bambi in the future. The model is of the form:

y ~ N(1 + treatment + row + col, s_y)

In this model we have a pattern like this to set priors:

spatial_variance_ratios = pm.Dirichlet(
    "spatial_variance_ratios", a=alphas, dims="sources"
)
sigmas = {source: np.sqrt(spatial_variance_ratios[i]) for i, source in enumerate(sources)}

where sources = ["row", "col", "s_y"] and alphas is a list of the same size. Then later we would build the model with terms like this:

predictor_row = pm.Normal("predictor_row", sigma=sigmas["row"], dims="row")

I was wondering if it would be possible to support defining priors similarly in Bambi. Thanks for your thoughts on this!