That sounds correct as it uses a multivariate approximation to the posterior. The deterministics are then the different dimensions of the multivariate + transforms if specified.
The number of parameters should be the same as the number of dimensions in the multivariate