PyTorch implementations of EfficientNetV2B2 and B3 are not working properly

[y-otn]

Thank you for creating such an amazing library.

I've noticed that the EfficientNetV2B2 and B3 models are not functioning correctly.
It seems that the in_channels of the top_conv layer are not being scaled.

deepvision/deepvision/models/classification/efficientnet/efficientnetv2_pt.py

Lines 177 to 184 in f45e928

	self.top_conv = nn.Conv2d(
	in_channels=blockwise_output_filters[-1],
	out_channels=top_channels,
	kernel_size=1,
	stride=1,
	padding="same",
	bias=False,
	)

I would appreciate it if you could check this issue. Thank you in advance.