Xgboost4j-spark and treelite


I am using xgboost4j-spark for the first time. I managed to train a model using the example in xgboost repo. I would like to perform realtime prediction using the trained model. I saw issue 4119 recommendation to use treelite to convert the model to another format and then load it back into a Scala Web service for prediction.

Does anyone know where I can find a sample code to convert a Xgboost4j-spark model to another format using treelite and then load the model back in for realtime prediction?

Thank you in advance for your help!


The treelite tutorials should help here, provided that XGB4j saves the usual binary model format.