Is it possible at the moment to also obtain the logp value of the samples? If not is that something you would consider adding?