Sampling from a posterior with a weighted log-likelihood function

[martinjankowiak]

you just need to use scale; see here:

Using scale in numpyro

That worked. Thanks!