org.tensorflow.op.quantization

Class FakeQuantWithMinMaxArgs

java.lang.Object

org.tensorflow.op.PrimitiveOp

org.tensorflow.op.quantization.FakeQuantWithMinMaxArgs

All Implemented Interfaces:

Op, Operand<Float>

@Operator(group="quantization")
public final class FakeQuantWithMinMaxArgs
extends PrimitiveOp
implements Operand<Float>

Fake-quantize the 'inputs' tensor, type float to 'outputs' tensor of same type.

Attributes `[min; max]` define the clamping range for the `inputs` data. `inputs` values are quantized into the quantization range (`[0; 2^num_bits - 1]` when `narrow_range` is false and `[1; 2^num_bits - 1]` when it is true) and then de-quantized and output as floats in `[min; max]` interval. `num_bits` is the bitwidth of the quantization; between 2 and 16, inclusive.

Before quantization, `min` and `max` values are adjusted with the following logic. It is suggested to have `min <= 0 <= max`. If `0` is not in the range of values, the behavior can be unexpected: If `0 < min < max`: `min_adj = 0` and `max_adj = max - min`. If `min < max < 0`: `min_adj = min - max` and `max_adj = 0`. If `min <= 0 <= max`: `scale = (max - min) / (2^num_bits - 1) `, `min_adj = scale * round(min / scale)` and `max_adj = max + min_adj - min`.

Quantization is called fake since the output is still in floating point.

Nested Class Summary

Nested Classes

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

Field Summary

Fields inherited from class org.tensorflow.op.PrimitiveOp

operation

Method Summary

Modifier and Type	Method and Description
Output<Float>	asOutput() Returns the symbolic handle of a tensor.
static FakeQuantWithMinMaxArgs	create(Scope scope, Operand<Float> inputs, FakeQuantWithMinMaxArgs.Options... options) Factory method to create a class wrapping a new FakeQuantWithMinMaxArgs operation.
static FakeQuantWithMinMaxArgs.Options	max(Float max)
static FakeQuantWithMinMaxArgs.Options	min(Float min)
static FakeQuantWithMinMaxArgs.Options	narrowRange(Boolean narrowRange)
static FakeQuantWithMinMaxArgs.Options	numBits(Long numBits)
Output<Float>	outputs()

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 FakeQuantWithMinMaxArgs create(Scope scope,
                                             Operand<Float> inputs,
                                             FakeQuantWithMinMaxArgs.Options... options)

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

Parameters:

scope - current scope

inputs -

options - carries optional attributes values

Returns:

a new instance of FakeQuantWithMinMaxArgs

min

public static FakeQuantWithMinMaxArgs.Options min(Float min)

Parameters:

min -

max

public static FakeQuantWithMinMaxArgs.Options max(Float max)

Parameters:

max -

numBits

public static FakeQuantWithMinMaxArgs.Options numBits(Long numBits)

Parameters:

numBits -

narrowRange

public static FakeQuantWithMinMaxArgs.Options narrowRange(Boolean narrowRange)

Parameters:

narrowRange -

outputs

public Output<Float> outputs()

asOutput

public Output<Float> asOutput()

Description copied from interface: Operand

Returns the symbolic handle of a tensor.

Inputs to TensorFlow operations are outputs of another TensorFlow operation. This method is used to obtain a symbolic handle that represents the computation of the input.

Specified by:

asOutput in interface Operand<Float>

See Also:

OperationBuilder.addInput(Output)