Package org.tensorflow.op

Class QuantizationOps

java.lang.Object

org.tensorflow.op.QuantizationOps

public final class QuantizationOps
extends java.lang.Object

An API for adding quantization operations to a Graph

Method Summary

Modifier and Type	Method	Description
<T> Dequantize	dequantize(Operand<T> input, Operand<java.lang.Float> minRange, Operand<java.lang.Float> maxRange, Dequantize.Options... options)	Adds an Dequantize operation to the graph
FakeQuantWithMinMaxArgs	fakeQuantWithMinMaxArgs(Operand<java.lang.Float> inputs, FakeQuantWithMinMaxArgs.Options... options)	Adds an FakeQuantWithMinMaxArgs operation to the graph
FakeQuantWithMinMaxArgsGradient	fakeQuantWithMinMaxArgsGradient(Operand<java.lang.Float> gradients, Operand<java.lang.Float> inputs, FakeQuantWithMinMaxArgsGradient.Options... options)	Adds an FakeQuantWithMinMaxArgsGradient operation to the graph
FakeQuantWithMinMaxVars	fakeQuantWithMinMaxVars(Operand<java.lang.Float> inputs, Operand<java.lang.Float> min, Operand<java.lang.Float> max, FakeQuantWithMinMaxVars.Options... options)	Adds an FakeQuantWithMinMaxVars operation to the graph
FakeQuantWithMinMaxVarsGradient	fakeQuantWithMinMaxVarsGradient(Operand<java.lang.Float> gradients, Operand<java.lang.Float> inputs, Operand<java.lang.Float> min, Operand<java.lang.Float> max, FakeQuantWithMinMaxVarsGradient.Options... options)	Adds an FakeQuantWithMinMaxVarsGradient operation to the graph
FakeQuantWithMinMaxVarsPerChannel	fakeQuantWithMinMaxVarsPerChannel(Operand<java.lang.Float> inputs, Operand<java.lang.Float> min, Operand<java.lang.Float> max, FakeQuantWithMinMaxVarsPerChannel.Options... options)	Adds an FakeQuantWithMinMaxVarsPerChannel operation to the graph
FakeQuantWithMinMaxVarsPerChannelGradient	fakeQuantWithMinMaxVarsPerChannelGradient(Operand<java.lang.Float> gradients, Operand<java.lang.Float> inputs, Operand<java.lang.Float> min, Operand<java.lang.Float> max, FakeQuantWithMinMaxVarsPerChannelGradient.Options... options)	Adds an FakeQuantWithMinMaxVarsPerChannelGradient operation to the graph
<T> Quantize<T>	quantize(Operand<java.lang.Float> input, Operand<java.lang.Float> minRange, Operand<java.lang.Float> maxRange, java.lang.Class<T> T, Quantize.Options... options)	Adds an Quantize operation to the graph
<T extends java.lang.Number> QuantizeAndDequantize<T>	quantizeAndDequantize(Operand<T> input, Operand<T> inputMin, Operand<T> inputMax, Operand<java.lang.Integer> numBits, QuantizeAndDequantize.Options... options)	Adds an QuantizeAndDequantize operation to the graph
<U,T> QuantizeDownAndShrinkRange<U>	quantizeDownAndShrinkRange(Operand<T> input, Operand<java.lang.Float> inputMin, Operand<java.lang.Float> inputMax, java.lang.Class<U> outType)	Adds an QuantizeDownAndShrinkRange operation to the graph
<T> RequantizationRange	requantizationRange(Operand<T> input, Operand<java.lang.Float> inputMin, Operand<java.lang.Float> inputMax)	Adds an RequantizationRange operation to the graph
<U,T> Requantize<U>	requantize(Operand<T> input, Operand<java.lang.Float> inputMin, Operand<java.lang.Float> inputMax, Operand<java.lang.Float> requestedOutputMin, Operand<java.lang.Float> requestedOutputMax, java.lang.Class<U> outType)	Adds an Requantize operation to the graph

Methods inherited from class java.lang.Object

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

Method Detail

fakeQuantWithMinMaxVarsGradient

public FakeQuantWithMinMaxVarsGradient fakeQuantWithMinMaxVarsGradient(Operand<java.lang.Float> gradients,
                                                                       Operand<java.lang.Float> inputs,
                                                                       Operand<java.lang.Float> min,
                                                                       Operand<java.lang.Float> max,
                                                                       FakeQuantWithMinMaxVarsGradient.Options... options)

Adds an FakeQuantWithMinMaxVarsGradient operation to the graph

Parameters:

gradients - Backpropagated gradients above the FakeQuantWithMinMaxVars operation.

inputs - Values passed as inputs to the FakeQuantWithMinMaxVars operation.

min -

max -

options - carries optional attributes values

Returns:

a new instance of FakeQuantWithMinMaxVarsGradient

See Also:

FakeQuantWithMinMaxVarsGradient

requantize

public <U,T> Requantize<U> requantize(Operand<T> input,
                                      Operand<java.lang.Float> inputMin,
                                      Operand<java.lang.Float> inputMax,
                                      Operand<java.lang.Float> requestedOutputMin,
                                      Operand<java.lang.Float> requestedOutputMax,
                                      java.lang.Class<U> outType)

Adds an Requantize operation to the graph

Parameters:

input -

inputMin - The float value that the minimum quantized input value represents.

inputMax - The float value that the maximum quantized input value represents.

requestedOutputMin - The float value that the minimum quantized output value represents.

requestedOutputMax - The float value that the maximum quantized output value represents.

outType - The type of the output. Should be a lower bit depth than Tinput.

Returns:

a new instance of Requantize

See Also:

Requantize

quantize

public <T> Quantize<T> quantize(Operand<java.lang.Float> input,
                                Operand<java.lang.Float> minRange,
                                Operand<java.lang.Float> maxRange,
                                java.lang.Class<T> T,
                                Quantize.Options... options)

Adds an Quantize operation to the graph

Parameters:

input -

minRange - The minimum scalar value possibly produced for the input.

maxRange - The maximum scalar value possibly produced for the input.

T -

options - carries optional attributes values

Returns:

a new instance of Quantize

See Also:

Quantize

fakeQuantWithMinMaxVarsPerChannel

public FakeQuantWithMinMaxVarsPerChannel fakeQuantWithMinMaxVarsPerChannel(Operand<java.lang.Float> inputs,
                                                                           Operand<java.lang.Float> min,
                                                                           Operand<java.lang.Float> max,
                                                                           FakeQuantWithMinMaxVarsPerChannel.Options... options)

Adds an FakeQuantWithMinMaxVarsPerChannel operation to the graph

Parameters:

inputs -

min -

max -

options - carries optional attributes values

Returns:

a new instance of FakeQuantWithMinMaxVarsPerChannel

See Also:

FakeQuantWithMinMaxVarsPerChannel

quantizeAndDequantize

public <T extends java.lang.Number> QuantizeAndDequantize<T> quantizeAndDequantize(Operand<T> input,
                                                                                   Operand<T> inputMin,
                                                                                   Operand<T> inputMax,
                                                                                   Operand<java.lang.Integer> numBits,
                                                                                   QuantizeAndDequantize.Options... options)

Adds an QuantizeAndDequantize operation to the graph

Parameters:

input -

inputMin -

inputMax -

numBits -

options - carries optional attributes values

Returns:

a new instance of QuantizeAndDequantize

See Also:

QuantizeAndDequantize

fakeQuantWithMinMaxArgsGradient

public FakeQuantWithMinMaxArgsGradient fakeQuantWithMinMaxArgsGradient(Operand<java.lang.Float> gradients,
                                                                       Operand<java.lang.Float> inputs,
                                                                       FakeQuantWithMinMaxArgsGradient.Options... options)

Adds an FakeQuantWithMinMaxArgsGradient operation to the graph

Parameters:

gradients - Backpropagated gradients above the FakeQuantWithMinMaxArgs operation.

inputs - Values passed as inputs to the FakeQuantWithMinMaxArgs operation.

options - carries optional attributes values

Returns:

a new instance of FakeQuantWithMinMaxArgsGradient

See Also:

FakeQuantWithMinMaxArgsGradient

fakeQuantWithMinMaxVarsPerChannelGradient

public FakeQuantWithMinMaxVarsPerChannelGradient fakeQuantWithMinMaxVarsPerChannelGradient(Operand<java.lang.Float> gradients,
                                                                                           Operand<java.lang.Float> inputs,
                                                                                           Operand<java.lang.Float> min,
                                                                                           Operand<java.lang.Float> max,
                                                                                           FakeQuantWithMinMaxVarsPerChannelGradient.Options... options)

Adds an FakeQuantWithMinMaxVarsPerChannelGradient operation to the graph

Parameters:

gradients - Backpropagated gradients above the FakeQuantWithMinMaxVars operation,

inputs - Values passed as inputs to the FakeQuantWithMinMaxVars operation, shape

min -

max -

options - carries optional attributes values

Returns:

a new instance of FakeQuantWithMinMaxVarsPerChannelGradient

See Also:

FakeQuantWithMinMaxVarsPerChannelGradient

quantizeDownAndShrinkRange

public <U,T> QuantizeDownAndShrinkRange<U> quantizeDownAndShrinkRange(Operand<T> input,
                                                                      Operand<java.lang.Float> inputMin,
                                                                      Operand<java.lang.Float> inputMax,
                                                                      java.lang.Class<U> outType)

Adds an QuantizeDownAndShrinkRange operation to the graph

Parameters:

input -

inputMin - The float value that the minimum quantized input value represents.

inputMax - The float value that the maximum quantized input value represents.

outType - The type of the output. Should be a lower bit depth than Tinput.

Returns:

a new instance of QuantizeDownAndShrinkRange

See Also:

QuantizeDownAndShrinkRange

fakeQuantWithMinMaxArgs

public FakeQuantWithMinMaxArgs fakeQuantWithMinMaxArgs(Operand<java.lang.Float> inputs,
                                                       FakeQuantWithMinMaxArgs.Options... options)

Adds an FakeQuantWithMinMaxArgs operation to the graph

Parameters:

inputs -

options - carries optional attributes values

Returns:

a new instance of FakeQuantWithMinMaxArgs

See Also:

FakeQuantWithMinMaxArgs

fakeQuantWithMinMaxVars

public FakeQuantWithMinMaxVars fakeQuantWithMinMaxVars(Operand<java.lang.Float> inputs,
                                                       Operand<java.lang.Float> min,
                                                       Operand<java.lang.Float> max,
                                                       FakeQuantWithMinMaxVars.Options... options)

Adds an FakeQuantWithMinMaxVars operation to the graph

Parameters:

inputs -

min -

max -

options - carries optional attributes values

Returns:

a new instance of FakeQuantWithMinMaxVars

See Also:

FakeQuantWithMinMaxVars

dequantize

public <T> Dequantize dequantize(Operand<T> input,
                                 Operand<java.lang.Float> minRange,
                                 Operand<java.lang.Float> maxRange,
                                 Dequantize.Options... options)

Adds an Dequantize operation to the graph

Parameters:

input -

minRange - The minimum scalar value possibly produced for the input.

maxRange - The maximum scalar value possibly produced for the input.

options - carries optional attributes values

Returns:

a new instance of Dequantize

See Also:

Dequantize

requantizationRange

public <T> RequantizationRange requantizationRange(Operand<T> input,
                                                   Operand<java.lang.Float> inputMin,
                                                   Operand<java.lang.Float> inputMax)

Adds an RequantizationRange operation to the graph

Parameters:

input -

inputMin - The float value that the minimum quantized input value represents.

inputMax - The float value that the maximum quantized input value represents.

Returns:

a new instance of RequantizationRange

See Also:

RequantizationRange