Closed RJT1990 closed 7 years ago
Issue is resolved and will be in next release. The solution for GAS models was to clip the gradients. This still does not mean it's fool-proof, but now if it does fail, it will be because a insensibly large count dataset would have been chosen (where a Normal distribution would be a better choice...)
NaNs/infs out for large count data. Needs to be written to avoid overflow in the exponentiated terms.