org.tensorflow.op.nn

Class FusedBatchNormGrad<T extends Number,U extends Number>

java.lang.Object

org.tensorflow.op.PrimitiveOp

org.tensorflow.op.nn.FusedBatchNormGrad<T,U>

Type Parameters:

T - data type for xBackprop() output

U - data type for scaleBackprop() output

All Implemented Interfaces:

Op

@Operator(group="nn")
public final class FusedBatchNormGrad<T extends Number,U extends Number>
extends PrimitiveOp

Gradient for batch normalization.

Note that the size of 4D Tensors are defined by either "NHWC" or "NCHW". The size of 1D Tensors matches the dimension C of the 4D Tensors.

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	FusedBatchNormGrad.Options Optional attributes for FusedBatchNormGrad

Field Summary

Fields inherited from class org.tensorflow.op.PrimitiveOp

operation

Method Summary

Modifier and Type	Method and Description
static <T extends Number,U extends Number> FusedBatchNormGrad<T,U>	create(Scope scope, Operand<T> yBackprop, Operand<T> x, Operand<Float> scale, Operand<U> reserveSpace1, Operand<U> reserveSpace2, FusedBatchNormGrad.Options... options) Factory method to create a class wrapping a new FusedBatchNormGrad operation.
static FusedBatchNormGrad.Options	dataFormat(String dataFormat)
static FusedBatchNormGrad.Options	epsilon(Float epsilon)
static FusedBatchNormGrad.Options	isTraining(Boolean isTraining)
Output<U>	offsetBackprop() A 1D Tensor for the gradient with respect to offset.
Output<U>	reserveSpace3() Unused placeholder to match the mean input in FusedBatchNorm.
Output<U>	reserveSpace4() Unused placeholder to match the variance input in FusedBatchNorm.
Output<U>	scaleBackprop() A 1D Tensor for the gradient with respect to scale.
Output<T>	xBackprop() A 4D Tensor for the gradient with respect to x.

Methods inherited from class org.tensorflow.op.PrimitiveOp

equals, hashCode, op, toString

Methods inherited from class java.lang.Object

clone, finalize, getClass, notify, notifyAll, wait, wait, wait

Method Detail

create

public static <T extends Number,U extends Number> FusedBatchNormGrad<T,U> create(Scope scope,
                                                                                 Operand<T> yBackprop,
                                                                                 Operand<T> x,
                                                                                 Operand<Float> scale,
                                                                                 Operand<U> reserveSpace1,
                                                                                 Operand<U> reserveSpace2,
                                                                                 FusedBatchNormGrad.Options... options)

Factory method to create a class wrapping a new FusedBatchNormGrad operation.

Parameters:

scope - current scope

yBackprop - A 4D Tensor for the gradient with respect to y.

x - A 4D Tensor for input data.

scale - A 1D Tensor for scaling factor, to scale the normalized x.

reserveSpace1 - When is_training is True, a 1D Tensor for the computed batch mean to be reused in gradient computation. When is_training is False, a 1D Tensor for the population mean to be reused in both 1st and 2nd order gradient computation.

reserveSpace2 - When is_training is True, a 1D Tensor for the computed batch variance (inverted variance in the cuDNN case) to be reused in gradient computation. When is_training is False, a 1D Tensor for the population variance to be reused in both 1st and 2nd order gradient computation.

options - carries optional attributes values

Returns:

a new instance of FusedBatchNormGrad

epsilon

public static FusedBatchNormGrad.Options epsilon(Float epsilon)

Parameters:

epsilon - A small float number added to the variance of x.

dataFormat

public static FusedBatchNormGrad.Options dataFormat(String dataFormat)

Parameters:

dataFormat - The data format for y_backprop, x, x_backprop. Either "NHWC" (default) or "NCHW".

isTraining

public static FusedBatchNormGrad.Options isTraining(Boolean isTraining)

Parameters:

isTraining - A bool value to indicate the operation is for training (default) or inference.

xBackprop

public Output<T> xBackprop()

A 4D Tensor for the gradient with respect to x.

scaleBackprop

public Output<U> scaleBackprop()

A 1D Tensor for the gradient with respect to scale.

offsetBackprop

public Output<U> offsetBackprop()

A 1D Tensor for the gradient with respect to offset.

reserveSpace3

public Output<U> reserveSpace3()

Unused placeholder to match the mean input in FusedBatchNorm.

reserveSpace4

public Output<U> reserveSpace4()

Unused placeholder to match the variance input in FusedBatchNorm.