Different sds for random slopes of sub-populations

Hi everybody, I’m trying to estimate a model in which I allow not just for different random intercepts of sub-populations (fit1 below) but also random slopes (fit2 below):

df <- data.frame(y = rnorm(1000), 
                 x = rnorm(1000), 
                 g = as.factor(rep(1:20,each=50)), 
                 f = as.factor(rep(c('a','b'),each=500)))

fit1 <- brm(y ~ 1 + (1|gr(g,by=f)),
            data = df, chains = 1) 

fit2 <- brm(y ~ 1 + (x|gr(g,by=f)),
            data = df, chains = 1) 

fit1 runs fine but fit2 crashes. What am I doing wrong? Thanks for your help!

I think this should be fixed in the brms github version already.

ah great thanks, just tried the github version and it indeed works!