The dimensions of the training set and the test set do not match

[Joy2Boy]

I want to know if the dimension of the training set is 256x256 and the test set is 512x512, can such a model perform inference? Also, can the author provide the inference code for reference, because the inference code on deep mind seems to be read remotely through tfhub.

[jacobbieker]

Hi, yes, the test set is a different shape than the training set, but the model should be able to perform inference. We are not the authors of the paper though, and this is more just the model architecture in PyTorch. It is the same as any other PyTorch model though, so you should be able to adapt the deepmind inference code, or write one yourself that works okay. There are some examples of model inference in the unit tests.