Complexity penalty k for a Bayesian version of Adjusted McFadden's R-Squared

General
2

How about LOO-R2 as described in the online appendix of R-squared for Bayesian regression models? No need for separate adjustment.