Why 'running_var' is not quantized though 'running_mean' is quantized in WgtScaleBatchNorm in slayer code

[gwgknudayanga]

Hi,

It can be seen that in the slayer code for WgtScaleBatchNorm, though the ' running_mean' quantized (and hence quantization aware training) the 'running_var' is not quantized. So this will affect when mapping the model from slayer to lava/loihi2 since in loihi2 these parameter and buffer values are going to be quantized, isn't it?

Thank you.