Mixed-Effects VI
VI (variational inference) is not supported for models with random effects. Calling fit_model with NoLimits.VI on a mixed-effects model will throw an error.
For full Bayesian inference on mixed-effects models, use MCMC. For likelihood-based mixed-effects estimation, use Laplace, LaplaceMAP, MCEM, or SAEM.
For a variational inference tutorial on a fixed-effects model, see Fixed-Effects Tutorial 2: Variational Inference (VI).