mlr3torch
mlr3torch copied to clipboard
CallbackCheckpoint
It would be great if continuing of training was possible from checkpoint, currently the optimizer and loss state are not stored.