org.deeplearning4j.spark.parameterserver.modelimport.elephas

Class ElephasModelImport

java.lang.Object

org.deeplearning4j.spark.parameterserver.modelimport.elephas.ElephasModelImport

public class ElephasModelImport
extends Object

Reads HDF5-persisted Elephas models stored with `model.save()` for both underlying `Sequential` and `Model` Keras models

Author:

Max Pumperla

Constructor Summary

Constructors

Constructor and Description
ElephasModelImport()

Method Summary

Modifier and Type	Method and Description
static SparkComputationGraph	importElephasModelAndWeights(org.apache.spark.api.java.JavaSparkContext sparkContext, String modelHdf5Filename) Load Elephas model stored using model.save(...) in case that the underlying Keras model is a functional `Model` instance, which corresponds to a DL4J SparkComputationGraph.
static SparkDl4jMultiLayer	importElephasSequentialModelAndWeights(org.apache.spark.api.java.JavaSparkContext sparkContext, String modelHdf5Filename) Load Elephas model stored using model.save(...) in case that the underlying Keras model is a functional `Sequential` instance, which corresponds to a DL4J SparkDl4jMultiLayer.

Methods inherited from class java.lang.Object

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

Constructor Detail

ElephasModelImport

public ElephasModelImport()

Method Detail

importElephasModelAndWeights

public static SparkComputationGraph importElephasModelAndWeights(org.apache.spark.api.java.JavaSparkContext sparkContext,
                                                                 String modelHdf5Filename)
                                                          throws IOException,
                                                                 UnsupportedKerasConfigurationException,
                                                                 InvalidKerasConfigurationException

Load Elephas model stored using model.save(...) in case that the underlying Keras model is a functional `Model` instance, which corresponds to a DL4J SparkComputationGraph.

Parameters:

sparkContext - Java SparkContext

modelHdf5Filename - Path to HDF5 archive storing Elephas Model

Returns:

SparkComputationGraph Spark computation graph

Throws:

IOException - IO exception

InvalidKerasConfigurationException - Invalid Keras config

UnsupportedKerasConfigurationException - Unsupported Keras config

See Also:

SparkComputationGraph

importElephasSequentialModelAndWeights

public static SparkDl4jMultiLayer importElephasSequentialModelAndWeights(org.apache.spark.api.java.JavaSparkContext sparkContext,
                                                                         String modelHdf5Filename)
                                                                  throws IOException,
                                                                         UnsupportedKerasConfigurationException,
                                                                         InvalidKerasConfigurationException

Load Elephas model stored using model.save(...) in case that the underlying Keras model is a functional `Sequential` instance, which corresponds to a DL4J SparkDl4jMultiLayer.

Parameters:

sparkContext - Java SparkContext

modelHdf5Filename - Path to HDF5 archive storing Elephas model

Returns:

SparkDl4jMultiLayer Spark computation graph

Throws:

IOException - IO exception

InvalidKerasConfigurationException - Invalid Keras config

UnsupportedKerasConfigurationException - Unsupported Keras config

See Also:

SparkDl4jMultiLayer