Assuming that all other parameters remain the same, is running learn.fit() with epoch=1 two times the same as running learn.fit() once with epoch=2?
My confusion stems from the fact that whenever I run learn.fit() on any of the notebooks, given that learn.fit() has already been run once, the model seems to update on what it has already learnt, as observed by the validation loss going down.
Assuming that all other parameters remain the same, is running learn.fit() with epoch=1 two times the same as running learn.fit() once with epoch=2?
Yep. The weights that were learned don’t get reset in between fit calls. If you want to do that you should rerun your code that instantiates your model/learner