FAQ# What should I do when my ‘posterior samples are outside the prior support’ in SNPE? Can the algorithms deal with invalid data, e.g., NaN or inf? When using multiple workers, I get a pickling error. Can I still use multiprocessing?
