Dropping parameters I'm not interested in

I’m trying to fit an IRT model using brms, and the formula is:
y ~ 1 + (1 | user) + (1 | item),

however, I am not really interested in the estimations for the user’s and the dataset it rather large (about 4k users, and 50 items), so in order to reduce the memory used by the fitted model, is it possible to not save the samples for the “user” parameters?

In the documentation for rstan::sampling there’s the pars and include parameter, but I’ve tried to specify include = FALSE and pars = “user”, but that doesn’t work…

Right now, you can either store all “random effects” (the default) or non of them via the argument save_ranef = FALSE. There is currently nothing in between these two things, but this may very well be a reasonable feature request. Would you mind opening an issue on https://github.com/paul-buerkner/brms about this?