FakeQuantChannelWiseAbsMax

class paddle.nn.quant.quant_layers. FakeQuantChannelWiseAbsMax ( name=None, channel_num=None, quant_bits=8, quant_axis=0, dtype='float32', quant_on_weight=False, reduce_type=None ) [source]
forward ( input )

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