Xgb trained in R and loaded into Java giving different predictions

Hi all,

I have a xgb.Booster model that was trained in R and saved to a file using ‘xgb.save’. I then loaded it into a Booster model in Java using XGBoost.loadModel().

The predictions from the R model and Java model are different and we are unsure why.

Has anyone had this problem before or knows how to solve it?

Thank you!