tensorplay.ao.quantization.default_per_channel_qconfig

tensorplay.ao.quantization.default_per_channel_qconfig = (functools.partial(<class 'tensorplay.ao.quantization.observer.MinMaxObserver'>, quant_min=0, quant_max=127){}, functools.partial(<class 'tensorplay.ao.quantization.observer.PerChannelMinMaxObserver'>, dtype=tensorplay.qint8, qscheme=tensorplay.per_channel_symmetric){})

observer factories for activations and weights.

Both fields must be observer classes or callables returning observer instances. Use MyObserver.with_args(x=1) to override constructor arguments.

Type:

Describes how to quantize a layer