Hello and thank you Jeremy and all for the amazing work!
Last week I just finished the lesson 2 of the 2018 version of the course, which is now located at http://course18.fast.ai/lessons/lesson2.html
Today I had the surprise to encountered a brand new website, a brand new fast.ai library release and a new set of videos for the course. That’s amazing.
But then, where should I migrate from the lesson I was about to follow (i.e. 2018 lesson 3)?
Shall I start again from scratch? Can I go directly to 2019 lesson 2? Will I miss something (I wouldn’t want to…)?
I definitely want to migrate, as suggested here, but the question is where
I am on a similar boat. My goal is to also complete Part 2 of the course. And given that v3 for Part 2 won’t release until June, I was wondering whether Part 1 v3 to Part 2 v2 would be a smooth sail? Or should I stick to v2?
If you are in part1 V2018, I would suggest go directly to part1 V2019. The content are quite similar but with completely new library. Some concepts you understand already (lr finder, …) so skim the video by playing with x2 or even just read the notes from the others student (Hiromi or PoonamV). It doesn’t take you much time and you will get the support from the very active community rather than V2018.
For the part2 v2019, I think it will be different for v2018 as Jeremy will use the dev notebook to explain how the library is built. But Part2 v2018 is still very interesting, you can learn by watching it and reimplementing in the new library
I was on Lesson 4 v2 and started all over from Lesson 1 v3. And for my surprise, there was a lot of new cool stuff which I didn’t see in v2, specifically regarding image segmentation & regression. You can always skip parts you already know but for me, it was really useful to go through all the material.
I was on lesson 3 v2 when v3 was released. Despite my trying not to, I found the best course of action was to restart from scratch, as there was a lot of new information and the library is quite different.
