I’m fitting models using brms version 2.19 and cmdstan 2.32.1 as the backend on a Windows 10 machine. I need to get marginal likelihoods for each model. When I use add_criterion() to the models get recompiled to Rstan. This is a pretty severe bottleneck in my code, as I’m doing this for hundreds of models. Is there a way to get the marginal likelihood without recompiling to rstan?