What happens when redefining learn = ConvLearner obect


(Fabio Graetz) #1

Hey guys,

I need some clarification.

Assume we set learn = ConvLearner.pretrained(arch, data, precompute=True) and train a few epochs. Then we set learn = ConvLearner.pretrained(arch, data, precompute=True, ps=0.5). Is the progress of the first learn deleted and we start with new weights? Or do we keep the old weights and dropout is added from now on?

Thanks!

F


(William Horton) #2

If you redefine the learn variable like that, the training will not be saved