Trained models used in the publication for brain and liver registration.
The liver models correspond to those finetuned in one step, since these yielded better results than the models finetuned in two steps (see publication for more details).

And the SpatialTransformer model needed for the HF demo.