ConvBn2d

class ConvBn2d(in_channels, out_channels, kernel_size, stride=1, padding=0, dilation=1, groups=1, bias=True, conv_mode='cross_correlation', compute_mode='default', eps=1e-5, momentum=0.9, affine=True, track_running_stats=True, padding_mode='zeros', **kwargs)

A fused QATModule including Conv2d and BatchNorm2d with QAT support. Could be applied with Observer and FakeQuantize.