Interface | Description |
---|---|
TrainingHook |
A hook for the workers when training.
|
TrainingMaster<R extends TrainingResult,W extends TrainingWorker<R>> |
A TrainingMaster controls how distributed training is executed in practice
In principle, a large number of different approches can be used in distributed training (synchronous vs. |
TrainingResult |
TrainingResult: a class used by
TrainingMaster implementations
Each TrainingMaster will have its own type of training result. |
TrainingWorker<R extends TrainingResult> |
TrainingWorker is a small serializable class that can be passed (in serialized form) to each Spark executor
for actually conducting training.
|
Class | Description |
---|---|
WorkerConfiguration |
A simple configuration object (common settings for workers)
|
Enum | Description |
---|---|
RDDTrainingApproach |
Approach to use when training from a
JavaRDD<DataSet> or JavaRDD<MultiDataSet> . |
Repartition |
Enumeration that is used for specifying the behaviour of repartitioning in
ParameterAveragingTrainingMaster
(and possibly elsewhere. |
RepartitionStrategy |
RepartitionStrategy: different strategies for conducting repartitioning on training data, when repartitioning is required.
SparkDefault: repartition using Spark's standard RDD.repartition(int) method. |