Hi
I set out to use Kaggle’s free GPU for learning fastai right from the beginning as I do not want to subscribe for paid services as much as possible. The problem was, the current part 1 video tutorials are out of date (v2) w.r.t fastai library which is now 1.0, while video uses 0.7. I initially tried using downgraded fastai in kaggle, but that mean switching off GPU in that which was very very very very time consuming. So I tried porting current lesson 1 to use fastai v1.0.
But being newbie here, it was highly challenging, with lots of earlier posts but not yet useful help. Finally have been able to finish, but I am sure there would be bugs. This is understandable, as I am porting w.o understanding the underlying concepts. But since this is top down approach, this should be ok, thus I request the current experts here to review my note book once. For this purpose, I have also not yet ported the documentation part.
This effort might totally go irrelevant once Part 1 v3 tutorials are out in mid Jan 2019, but no one is sure about that, and there could be some like me, who just cannot wait and also not ready to use paid GPU services. I hope this could be heavily useful for them. Also it does not make sense to learn fastai 0.7 provided it is already expired with newer major version on the table.
Here is the kaggle notebook for your review. Last version I attempted to commit multiple times but was unsuccessful as code was taking too long time to execute. So modifications to make it faster also would be highly appreciated. For now, I am thinking of splitting it further in to parts.