Lesson 2 Larger Images Clarification

In Lesson 2 about the Dog Breed Classification. Jeremy mentions about the retraining with a larger images. The line that I am referring to in the code is below

learn.set_data(get_data(299, bs))

Will start small then move to larger general images.

How does calling this method learn.set_data(get_data(299, bs)) re-train with larger size images, because the method get_data returns
return data if sz > 300 else data.resize(340, 'tmp')

The sz parameter that is passed is β€œ299” so what is the explanation, The only difference is calling the transformation model to transform the images will now use sz as 299


1 Like

Did you figure this out? I’m also wondering about this…

set_data does not retrain the model. If you look carefully, he runs learn.fit in the next cell, which actually does the training.