Load XGBoost model made in Python to spark

I got a xgboost model made in python (xgb_model.json) and I need to load it to Spark and on Scala too. When I try to do it it searches for /metadata in the path. What do I do?

Begin by reproducing the load attempt with xgb_model.json in Spark and Scala, recording where the /metadata lookup occurs. Check whether the Spark/Scala loader supports a Python-produced XGBoost model in this format; done means the supported loading or conversion path is identified and the model loads without the missing-metadata error.

Written by the indexing model from the issue text.