We often need samples from the posterior for downstream analyses (for example). We currently export a JSON of nested records without labels. Ideally, we would use the save_posterior function to save a pickle of the posterior dictionary to a specific path.