Just so people have a clear understanding of what I have planned (as I have no affiliation with Jeremy’s wonderful courses ), yesterday I conducted a Twitter poll to see what the community would be wanting in terms of next in regards to WWF2 and v2 walkthroughs. The results were pretty darn close, but continuing WWF2 text won, so, let’s discuss how this will work and what I have in the planning phases:
-
We’re still going to finish up the
fastinference
sub-course, that’s not going away -
I won’t have an ETA when it’s done, I want it to be as good as vision and tabular and I’m almost there in the knowledge level, I just need a little bit more work, so soon ish but I’m not putting a date here.
-
There was want for an interpretability course, and for those of you who joined/have taken the tabular module, you know I try to involve interpretability when and how I can. So to that regard, despite it coming in second, I won’t be making the interpretability course with fastai until after the next course…which is:
-
The course after WWF2 will be a debugging guide to fastai2. I’ll be attempting to make a series of videos in-which we walk through from the most basic of problems in the high-level API to the most complex and mysterious problems hidden away (such as those referencing
fastcore
instead offastai2
). IMO this provides more value at this moment compared to the interpretability module, so we’ll be doing this second. -
Until the newest version of Part 1 is released, the course megathreads will be pinned to the forum, for easier access to them
So that is overall my plans for the next few months, I will of course update the v2 text thread with important dates and make megathreads for the other two when it is a bit closer.
I’m very glad that so many of you are really enjoying these series, and I want to get back to them, and I think this would be the right approach. Thanks guys
Edit: Also I am open to ideas from the community towards what you would specifically want out of those three