Package org.tensorflow.op.quantization

Class FakeQuantWithMinMaxVarsGradient

java.lang.Object

org.tensorflow.op.PrimitiveOp

org.tensorflow.op.quantization.FakeQuantWithMinMaxVarsGradient

All Implemented Interfaces:

Op

@Operator(group="quantization")
public final class FakeQuantWithMinMaxVarsGradient
extends PrimitiveOp

Compute gradients for a FakeQuantWithMinMaxVars operation.

Nested Class Summary

Nested Classes

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

Field Summary

Fields inherited from class org.tensorflow.op.PrimitiveOp

operation

Method Summary

Modifier and Type	Method	Description
Output<java.lang.Float>	backpropsWrtInput()	Backpropagated gradients w.r.t.
Output<java.lang.Float>	backpropWrtMax()	Backpropagated gradients w.r.t.
Output<java.lang.Float>	backpropWrtMin()	Backpropagated gradients w.r.t.
static FakeQuantWithMinMaxVarsGradient	create(Scope scope, Operand<java.lang.Float> gradients, Operand<java.lang.Float> inputs, Operand<java.lang.Float> min, Operand<java.lang.Float> max, FakeQuantWithMinMaxVarsGradient.Options... options)	Factory method to create a class to wrap a new FakeQuantWithMinMaxVarsGradient operation to the graph.
static FakeQuantWithMinMaxVarsGradient.Options	narrowRange(java.lang.Boolean narrowRange)
static FakeQuantWithMinMaxVarsGradient.Options	numBits(java.lang.Long numBits)

Methods inherited from class java.lang.Object

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

Methods inherited from class org.tensorflow.op.PrimitiveOp

equals, hashCode, toString

Method Detail

create

public static FakeQuantWithMinMaxVarsGradient create(Scope scope,
                                                     Operand<java.lang.Float> gradients,
                                                     Operand<java.lang.Float> inputs,
                                                     Operand<java.lang.Float> min,
                                                     Operand<java.lang.Float> max,
                                                     FakeQuantWithMinMaxVarsGradient.Options... options)

Factory method to create a class to wrap a new FakeQuantWithMinMaxVarsGradient operation to the graph.

Parameters:

scope - current graph scope

gradients - Backpropagated gradients above the FakeQuantWithMinMaxVars operation.

inputs - Values passed as inputs to the FakeQuantWithMinMaxVars operation. min, max: Quantization interval, scalar floats.

min -

max -

options - carries optional attributes values

Returns:

a new instance of FakeQuantWithMinMaxVarsGradient

numBits

public static FakeQuantWithMinMaxVarsGradient.Options numBits(java.lang.Long numBits)

Parameters:

numBits - The bitwidth of the quantization; between 2 and 8, inclusive.

narrowRange

public static FakeQuantWithMinMaxVarsGradient.Options narrowRange(java.lang.Boolean narrowRange)

Parameters:

narrowRange - Whether to quantize into 2^num_bits - 1 distinct values.

backpropsWrtInput

public Output<java.lang.Float> backpropsWrtInput()

Backpropagated gradients w.r.t. inputs: `gradients * (inputs >= min && inputs <= max)`.

backpropWrtMin

public Output<java.lang.Float> backpropWrtMin()

Backpropagated gradients w.r.t. min parameter: `sum(gradients * (inputs < min))`.

backpropWrtMax

public Output<java.lang.Float> backpropWrtMax()

Backpropagated gradients w.r.t. max parameter: `sum(gradients * (inputs > max))`.