Commits


Markus Cozowicz authored and Guolin Ke committed beeb6e0fd58
Added additional APIs to better support JNI on Spark (#2032) * added API changes required for JNI performance optimizations (e.g. predict is 3-4x faster) * removed commented variables * removed commented header * renamed method to make it obvious it is created for Spark * fixed comment alignment * replaced GetPrimitiveArrayCritical with GetIntArrayElements for training. fixed dead-lock on databricks