org.deeplearning4j.spark.impl.multilayer.gradientaccum

Class GradientAccum

java.lang.Object

org.deeplearning4j.spark.impl.multilayer.gradientaccum.GradientAccum

All Implemented Interfaces:

Serializable, org.apache.spark.api.java.function.Function<org.nd4j.linalg.dataset.DataSet,Gradient>

public class GradientAccum
extends Object
implements org.apache.spark.api.java.function.Function<org.nd4j.linalg.dataset.DataSet,Gradient>

Runs iterative reduce on the dataset

Author:

Adam Gibson

See Also:

Serialized Form

Constructor Summary

Constructors

Constructor and Description
GradientAccum(org.apache.spark.broadcast.Broadcast<org.nd4j.linalg.api.ndarray.INDArray> params, String json) Train and average over mini batches from a dataset

Method Summary

Modifier and Type	Method and Description
Gradient	call(org.nd4j.linalg.dataset.DataSet dataSet)

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

GradientAccum

public GradientAccum(org.apache.spark.broadcast.Broadcast<org.nd4j.linalg.api.ndarray.INDArray> params,
                     String json)

Train and average over mini batches from a dataset

Parameters:

params - the parameters that were broadcast

json - the configuration for the network

Method Detail

call

public Gradient call(org.nd4j.linalg.dataset.DataSet dataSet)
              throws Exception

Specified by:

call in interface org.apache.spark.api.java.function.Function<org.nd4j.linalg.dataset.DataSet,Gradient>

Throws:

Exception