UNet uses zero-padding even when it shouldn't #647
Description
There's an option to make the UNet use circular convolutions and even though it's set here
deepinv/deepinv/models/unet.py
Line 110 in 7116f70
deepinv/deepinv/models/unet.py
Line 137 in 7116f70
it's not set there (meaning it defaults to zero-padding)
deepinv/deepinv/models/unet.py
Line 119 in 7116f70
deepinv/deepinv/models/unet.py
Line 141 in 7116f70
deepinv/deepinv/models/unet.py
Line 151 in 7116f70
deepinv/deepinv/models/unet.py
Line 164 in 7116f70