ZeroPointDomain# class torch.ao.quantization.observer.ZeroPointDomain(value)[source]# Enum that indicate whether zero_point is in integer domain or floating point domain integer domain: quantized_val = (float_val / scale) (integer) + zero_point (integer) float domain: quantized_val = (float_val - (zero_point (float) - scale * mid_point)) / scale none domain: quantized_val = (float_val / scale)
