Question around building cudnn* variants

Hello!

This is more of a question than an issue.

I read through the documentation and was planning to build some outdated images for a pretty old system. (needing cuda 10.0 cudnn7)
One thing I have not seen is the procedure to build the cudnn variants of the docker images.

I've managed to build the old ones with:

./build.sh -d --image-name XXXXXXXXX.dkr.ecr.eu-central-1.amazonaws.com/cuda --cuda-version 10.0 --os ubuntu --os-version 18.04 --arch x86_64 --push

but only the base/runtime/devel variants were present.

Looking thought build.sh i could not figure out how to build the cudnn variants too

Edit 1:
What I ended up doing was cloning 10.0 to 10.0-cudnn7 and copying over the Dockefile's from the cudnn7 folders into their respective parents inside the ubuntu1804 folder, then i ran:

./build.sh -d --image-name XXXXXXXXX.dkr.ecr.eu-central-1.amazonaws.com/cuda --cuda-version 10.0-cudnn7 --os ubuntu --os-version 18.04 --arch x86_64 --push