Calling the Library

For a project that involves native Spark algorithms, you can directly call the Kunpeng BoostKit machine learning algorithm library without modifying the code. You only need to reference the JAR files of the library when submitting an algorithm. For details, see Deployment Process. If the project does not call a native Spark algorithm, develop applications based on the Machine Learning Algorithm Library Developer Guide.

spark.executor.userClassPathFirst is an experimental parameter. If this parameter is set to true, the class loading sequence will be changed, which may cause algorithm package incompatibility. Therefore, you are not advised to use this parameter.