run_iterationRAW parameter

[mhmd-mst]

Hey, I am trying to apply the techniques you proposed, I want to ask you regarding run_iterationRAW has parameter data_generatorRAW, what is the difference between it and data_generator?

[ZerojumpLine]

Hey,

Sorry for the delayed reply and thank you very much for your interests! data_generatorRAW: it would give data with the original format, without any data augmentation. data_generator: it is from default nnunet code, data with built-in data augmentation. I would use data_generatorRAW to generate our version of transformed samples (w/ mixup and w/ adversarial training). Because I found the tranformed data w/ data_generator would be unstable as it becomes quite different from the original data.

Hope it can resolve your problem!

Best, Zeju