Does xgboost4j-spark support continuation training

#1

Hi,
It seems the python API of xgboost support the continuation training by providing a trained model: https://xgboost.readthedocs.io/en/latest/python/python_api.html#module-xgboost.training. Is the same functionality also supported by xgboost4j/xgboost4j-spark? I did not find one. Are you planning to add it?
Thank you.

#2

Hi @gexu,
You could try training on initial prediction of a model (baseMarginCol param in model), but seems it is working properly only on small datasets - Margin issue in distributed xgboost

#3

Thank you @Clamoris, I am working at a similar problem as you described in Margin issue in distributed xgboost.