Fast.ai courses

Hello everybody!
I have been taking fast.ai Part 1 (2019). I am a bit confused with the different courses of fast.ai out there. The lectures are amazing, so is the community here.

I found there
Deep Learning
Part 1 Deep learning for coders (2019)
Part 2 Deep learning from foundations (2019)
Part 1 Deep learning for coders (2018)
Part 2 Cutting Edge Deep learning (2018)
Code First NLP course
2017 versions are also there.

Machine Learning
Machine learning course

DL math
Computational Linear Algebra

I am currently taking Part 1 2019 course. From a bit of research, the 2018 part 1 uses fastai v0.7.
The part 2 of 2019 and 2018 look to be completely different courses. Part 2 (2019) looks to be implementing and learning more about fastai and Pytorch (DL in general)

While in part 2 of 2018, Jeremy seems to be teaching more sophisticated applications of DL.

  • Please point out the differences between Part 1 and Part 2 of 2019 and 2018.

  • Why Jeremy decided to make a different Part 2 in 2019? To make a complete package of implementation + application? I mean different in the concepts taught.

  • Which course should one take after Part 1 (2019) i.e Part 2 of 2018 or part 2 of 2019 first?
    I would like to take both the Part 2’s though in the future. I want to learn more sophisticated applications but I want to learn to read research papers as well.(You learn that by doing but Jeremy gives so much great advice)

If this a duplicate thread, please inform. I could not find other threads addressing this.

Hi Sankalp

I suggest 2019 P1 and P2 then Machine Learning because it covers random forests and then finally Computation Linear Algebra but be aware the audio quality is poor. It is worth remembering 2020 P1 and 2020 P2 should be coming soon.

Regards Conwyn

I made one already here actually: What course is what? Where should I start? A Guide

:slight_smile:

2 Likes

Thanks @Conwyn for replying. I am mostly confused between Part 2 Cutting Edge Deep learning (2018) and Deep Learning from foundations(2019). Anyways, by the time I will reach to any of the part 2, 2020 version will be released by then.

I think after completing Part 1, i will do the latest part 2. I need to learn to read research papers so I will be doing Deep Learning from Foundations. So, I will follow your suggestion.
OR
Only, I am considering to watch Intro to ML right now since new fastai is going to come out in July.
Till then, I can watch a bit of Part 1 v3 and intro to ML. Basically, wait and do other stuff.

For now, I can watch Intro to ML

Thanks for replying @muellerzr. I had been watching your interview with Sanyam a few days back. Glad to see you here.

I have gone through that forum. I could not find any mention of Part 2 Cutting Edge Deep learning (2018) other than this. I think you have included it under Practical Deep Learning for Coders. Maybe, you can add their separately like you have added for v3 Part 2.

Also the link for lecture in Practical Deep Learning for Coders (2018) is pointing towards 2019 lectures.

Hi Sankalp

I like 2019 P2 because I like to understand the machinery rather than typing parameters into a function. Where as you might have a practical use for deep learning. I think everybody takes different things from the courses.

Regards Conwyn

1 Like

Yes. I thought the same. I saw the initial few minutes of 2019 P2. Jeremy says that he can’t teach all the different types of architectures but if you know the foundations, you can get going on your own.
For now, I have decided to keep watching v3P1 and Intro to ML slowly. Then, when v4 comes, I will watch it from scratch. Thanks for helping @Conwyn.