tensorplay.ao.quantization.default_observer

tensorplay.ao.quantization.default_observer = MinMaxObserver(quant_min=0, quant_max=127)

observer classes be specialized with constructor arguments while staying callable as observer_cls(**kwargs).