Tune in to the first Women in ML Symposium this Tuesday, October 19 at 9am PST Register now

RequantizePerChannel

public final class RequantizePerChannel

Requantizes input with min and max values known per channel.

Public Methods

static <U, T> RequantizePerChannel<U>
create(Scope scope, Operand<T> input, Operand<Float> inputMin, Operand<Float> inputMax, Operand<Float> requestedOutputMin, Operand<Float> requestedOutputMax, Class<U> outType)
Factory method to create a class wrapping a new RequantizePerChannel operation.
Output<U>
output()
Output tensor.
Output<Float>
outputMax()
The maximum value of the final output tensor.
Output<Float>
outputMin()
The minimum value of the final output tensor

Inherited Methods

Public Methods

public static RequantizePerChannel<U> create (Scope scope, Operand<T> input, Operand<Float> inputMin, Operand<Float> inputMax, Operand<Float> requestedOutputMin, Operand<Float> requestedOutputMax, Class<U> outType)

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

Parameters
scope current scope
input The original input tensor.
inputMin The minimum value of the input tensor
inputMax The maximum value of the input tensor.
requestedOutputMin The minimum value of the output tensor requested.
requestedOutputMax The maximum value of the output tensor requested.
outType The quantized type of output tensor that needs to be converted.
Returns
  • a new instance of RequantizePerChannel

public Output<U> output ()

Output tensor.

public Output<Float> outputMax ()

The maximum value of the final output tensor.

public Output<Float> outputMin ()

The minimum value of the final output tensor