Paul Xu

Using custom distributions in `bmb.Prior` for group-specific terms

10

Hi, I know that `bambi` supports arbitrary distributions as priors through the `dist` argument. However, when I wanted to use a custom distribution as the prior for a group-specific term,...

TypeError: cannot pickle '_contextvars.Context' object or 'PyCapsule' object during design matrix creation

7

The `hssm` package relies on `bambi` for model creation. A Type error is thrown during the creation of certain models with certain term combinations: ```python import hssm cav_data = hssm.load_data("cavanagh_theta")...

### Describe the issue: In `pymc.sampling.jax.sample_numpyro_nuts`, a jitter is always applied to the initial values through function `_get_batched_jittered_initial_points()`: https://github.com/pymc-devs/pymc/blob/904a0eaac216732bc358dd91680cd428d95704f0/pymc/sampling/jax.py#L662-L668 This function actually accepts a `jitter` argument that allows jittering to...

Raise an error when there are invalid prior specifications

1

Is there a way for `bambi` to throw an error if there are entries in the `priors` dict passed to `bmb.Model` that are not assigned to any term?

Make `var_names` accessible in `model.fit()`

3

In PyMC, `pm.Sample()` has a `var_names` parameter which allows the users to specify the variable names to be included in `InferenceData`. In Bambi, it seems that `_run_mcmc()` will override this...

Setting alias to random intercept of the parent parameter unexpectedly changes the name of the factor dimension

4

Thank you so much, the Bambi team, for the latest version! The graphs look much cleaner now. There is one small inconsistency, though. For example: I have a model built...