ANN with two outputs and only one target - custom loss function returns NaN

Question

I have a dataset with 212 input columns and one target. The thing is that I dont want to predict the value of target, but values of parameters of Normal distribution. As a l