When cbs is used in a unet learner as,
unet_learner(dls, resnet34, loss_func=CrossEntropyLossFlat(axis=1), cbs=cutmix, metrics=Dice(axis=1)) where cutmix=CutMix(alpha=1.)
The size of output tensors I am getting from learn.predict is as follows:
Could you kindly suggest what went wrong, how to solve the problem?