XGBoost4J (with Spark) 0.9 unable to train with udt column?

Does anyone know how to Explicitly convert the Vector returned from VectorAssembler to a DenseVector to return the zeros to the dataset. in Scala? I’m using a Pipeline containing a VectorAssembler and XGBoostClassifier and I’m not sure how to force the VectorAssembler to transform the dataset into a DenseVector in the pipeline.