Package org.nd4j.autodiff.samediff.internal

Class TrainingSession

java.lang.Object

org.nd4j.autodiff.samediff.internal.AbstractSession<INDArray,Pair<SameDiffOp,OpContext>>

org.nd4j.autodiff.samediff.internal.InferenceSession

org.nd4j.autodiff.samediff.internal.TrainingSession

public class TrainingSession
extends InferenceSession

Nested Class Summary

Nested classes/interfaces inherited from class org.nd4j.autodiff.samediff.internal.InferenceSession

InferenceSession.ConstantDep, InferenceSession.Dep, InferenceSession.ExecDoneDep, InferenceSession.OpDep, InferenceSession.PlaceholderDep, InferenceSession.ReqOutputDep, InferenceSession.VariableDep

Nested classes/interfaces inherited from class org.nd4j.autodiff.samediff.internal.AbstractSession

AbstractSession.ExecStep, AbstractSession.ExecStepPredicate, AbstractSession.ExecType, AbstractSession.VarId

Field Summary

Fields

Modifier and Type	Field	Description
protected TrainingConfig	config
protected double[]	currIterLoss
protected Map<Class<?>,AtomicDouble>	currIterRegLoss
protected Map<String,String>	gradVarToVarMap
protected List<Listener>	listeners
protected Map<String,Integer>	lossVarsToLossIdx
protected Map<String,GradientUpdater>	updaters

Fields inherited from class org.nd4j.autodiff.samediff.internal.InferenceSession

freedArrays, KERAS_TRAIN_TEST

Fields inherited from class org.nd4j.autodiff.samediff.internal.AbstractSession

dt, nodeValueOutputs, OUTER_FRAME, sameDiff, subgraph, subgraphOps, zeroInputOpsInSubgraph

Constructor Summary

Constructors

Constructor	Description
TrainingSession(SameDiff sameDiff)

Method Summary

Modifier and Type	Method	Description
ExecutionResult	getOutputs(Pair<SameDiffOp,OpContext> opPair, FrameIter outputFrameIter, Set<AbstractSession.VarId> opInputs, Set<AbstractSession.VarId> allIterInputs, Set<String> constAndPhInputs, List<Listener> listeners, At at, MultiDataSet batch, Set<String> allReqVariables, Map<String,SDValue> otherPlaceHolders)	Execute the op - calculate INDArrays, or shape info, etc
Loss	trainingIteration(TrainingConfig config, Map<String,INDArray> placeholders, Set<String> paramsToTrain, Map<String,GradientUpdater> updaters, MultiDataSet batch, List<String> lossVariables, List<Listener> listeners, At at)	Perform one iteration of training - i.e., do forward and backward passes, and update the parameters

Methods inherited from class org.nd4j.autodiff.samediff.internal.InferenceSession

doExec, getAndParameterizeOp, getArray, getConstantOrVariable, getOutputsHelperTensorArrayOps, postProcessOutput, postProcessOutputValues, preprocessPlaceholders

Methods inherited from class org.nd4j.autodiff.samediff.internal.AbstractSession

addDependenciesForOp, addVarControlDeps, contains, execFailed, get, get, getExecStepForVar, getSdValue, getTensorArraysInSession, getTensorArraysInSession, getTensorFromOutputs, initSubgraph, lookup, lookup, output, output, preprocessValuePlaceholders, putNodeValue, setArrayAtIndex, updateDescendantDeps

Methods inherited from class java.lang.Object

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

Field Detail

config

protected TrainingConfig config

gradVarToVarMap

protected Map<String,String> gradVarToVarMap

updaters

protected Map<String,GradientUpdater> updaters

lossVarsToLossIdx

protected Map<String,Integer> lossVarsToLossIdx

currIterLoss

protected double[] currIterLoss

currIterRegLoss

protected Map<Class<?>,AtomicDouble> currIterRegLoss

listeners

protected List<Listener> listeners

Constructor Detail

TrainingSession

public TrainingSession(SameDiff sameDiff)

Method Detail

trainingIteration

public Loss trainingIteration(TrainingConfig config,
                              Map<String,INDArray> placeholders,
                              Set<String> paramsToTrain,
                              Map<String,GradientUpdater> updaters,
                              MultiDataSet batch,
                              List<String> lossVariables,
                              List<Listener> listeners,
                              At at)

Perform one iteration of training - i.e., do forward and backward passes, and update the parameters

Parameters:

config - Training configuration

placeholders - Current placeholders

paramsToTrain - Set of parameters that will be trained

updaters - Current updater state

batch - Current data/batch (mainly for listeners, should have already been converted to placeholders map)

lossVariables - Loss variables (names)

listeners - Listeners (if any)

at - Current epoch, iteration, etc

Returns:

The Loss at the current iteration

getOutputs

public ExecutionResult getOutputs(Pair<SameDiffOp,OpContext> opPair,
                                  FrameIter outputFrameIter,
                                  Set<AbstractSession.VarId> opInputs,
                                  Set<AbstractSession.VarId> allIterInputs,
                                  Set<String> constAndPhInputs,
                                  List<Listener> listeners,
                                  At at,
                                  MultiDataSet batch,
                                  Set<String> allReqVariables,
                                  Map<String,SDValue> otherPlaceHolders)

Description copied from class: AbstractSession

Execute the op - calculate INDArrays, or shape info, etc

Overrides:

getOutputs in class InferenceSession

Parameters:

opPair - Operation to exit. This should be parameterized (i.e., all inputs set)

outputFrameIter - The frame and iteration of the outputs

opInputs - The specific input arrays for the op

allReqVariables - All required variables requested for the current session execution (not just the current op outputs)

Returns:

The outputs of the op