I thank the author for MNP package.
I would like to report an issue I am facing while using the package.
I found some of the predicted probabilities to be zero. Hence the computed log-likelihood becomes -ve infinity and the code returns NaN.
Is there a way to overcome this??

Thanks