NirroAs advised by my tutor, I've implemented Precision-Recall and ROC curves to evaluate my trained model.
...
Haven't yet tried them, though. It might work, it might not. Who knows? I've had enough for today.
I was surprised to learn that you cannot do ROC curves on #tensorboard directly... but after banging my head against the keyboard, I've found that you can send #matplotlib grahps to it.
(And also torchmetrics gives you nice matplotlib graphs for your metrics.)