FakeQuantMAOutputScaleLayer

class paddle.nn.quant.quant_layers. FakeQuantMAOutputScaleLayer ( layer, weight_bits=8, activation_bits=8, moving_rate=0.9, name=None, reduce_type=None, *args, **kwargs ) [source]

Add FakeQuantMovingAverageAbsMax layer to the behind of the input layer.

forward ( *inputs, **kwargs )

forward

Defines the computation performed at every call. Should be overridden by all subclasses.

Parameters
  • *inputs (tuple) – unpacked tuple arguments

  • **kwargs (dict) – unpacked dict arguments