I am a newbie in Data Science, and wanted to gain deeper application + hands-on knowledge on the same. I was wondering should I start with fastai’s Machine Learning course from 2018 - Introduction to Machine Learning for Coders: Launch · fast.ai , or the latest Deep Learning course :- https://course.fast.ai/ would suffice. I believe the ML course is on an older version of fastAi.
Thanks in advance!
This is just my own, personal opinion, and that is that if I were starting out right now, I’d just go with the 2020 version of the fast.ai course. If you can wait a couple months you may get access to the 2022 version of the course, but having been part of it, it is not too divergent from the 2020 version and so if you get started with the 2022 version in let’s say August of this year, your time spent on the 2020 version would have been well spent. It is still using the fastbook but has some new concepts like transformers and Jeremy has used Kaggle notebooks to make it more accessible for participants. Also, there are a lot of walkthru videos which will be made available which cover topics like Vim/Bash/setting up Paperspace instances/ Submitting to Kaggle competitions etc.
HTH.
3 Likes
Hi Kashish
It might be worth waiting for the 2022 couse. You could try
[Computational Linear Algebra] and [Code-First Introduction to Natural Language Processing]
whilst you are waiting. You could also try [www.probabilitycourse.com].
Regards Conwyn
1 Like
Thanks a lot for your inputs. Is there any planned date for the launch of 2022 course online ? I’m planning to wait for the course, in the 2 month window, I can cover some other aspects of data science (like stats, SQL, etc)
Thanks for your inputs Conwyn. Will try out the courses you recommended!
The problem with doing courses from 3 years ago is that you’ll be dealing with versions of the library from 3 years ago, and you may get caught up with trying to figure those things out and get distracted from the main goal which is learning about Deep Learning. And when you ask questions, most people won’t be in a position to answer those questions. There isn’t a major diff between the 2020 course (vs the 2022 course) and if you have time to get started now, start with chapter 1 of the fastbook because by the time you’re done with ch1 and ch2 the new course might already be released. The focus (still) is the book aka “fastbook” and you can add the 2022 lectures to your learning when they are released; because 2022 version isn’t too different from the 2020 course.
As Jeremy has pointed out on various occasions, you don’t need to take tons of courses (math/probability/NLP etc) before getting started with the fast.ai course (also mentioned in the first few pages of the book.) So I’ll advise anyone starting out with fast.ai /fastbook course to beware of falling into the trap of going down many rabbit holes of stats/maths/data science “knowledge gain”. I would just read the book and watch the 2020 lectures. They are sufficient and still quite relevant to get you on the path. 2022 course is along very similar lines, so if you have time this summer, I wouldn’t spend too much of it on taking other obliquely related courses which you really wouldn’t need for the 2022 version (or the 2020 version for that matter). Also it all depends on how much time you have and what your goals are. Personally I’ve found that going down too many rabbit holes has ultimately been counter-productive, and I didn’t accomplish my original goal either. But then again, YMMV.
2 Likes
I found this confusing when I encountered the website for the first time. Fortunately I started and stood with the book but… I’m wondering, would it make sense to add the “Where to start” advice to the front page? Did I miss it there?
Either a where to start or a learning map(s).
This is something I find in some books and like it (even though I mostly follow the original order).
Hi Diogo, have you seen Jeremy’s Lecture 0 ? It’s posted on his youtube channel where the 2020 lectures are also posted.
HTH.
1 Like
I have not yet no. Mostly just looking at the book, but I’ll add it to the watch list, thanks!