Pointwise log-likelihood for loo/waic with custom "black-box" probability distribution

Thanks for your quick response and the link to the examples!

I tried following the first example in that link, but still get a NotImplemented error. The error is occuring when I perform my sampling:

idata_mh = pm.smc.sample_smc(draws=1000, cores=10, chains=4, return_inferencedata=True, idata_kwargs=dict(log_likelihood=True))

When you say I need to pass the parameters function, is this refering to mu in logp, or the parameters that I am trying to infer? Further, should I be specifying the value variable in the logp function?