After I load a learner from from a specific path where I had saved it previously how can I resume its learning for further epochs? Does it automatically resumes or it starts from the beginning if I do fit_one_cycle after loading ? Can someone please make this clear to me?
After you load a learner it resumes the state of the save you did earlier. To fit it further, just load it and call fit_one_cycle; it does not start from the beginning.
I think this is mentioned in one of the first two lectures. After fitting for a few epochs, unfreezing and then fitting again, accuracy dropped. Then Jeremy said as we’ve broken the model, it is better to load it from a better instance and proceeded to load it.
Basically loading means replacing the current model weights with saved ones.
Hope this helped.
Ok I will try thank you very much