Oh, yes you're right...hmmm. Unsure what to do about that, I'd love a neat graphical solution. Wasn't quite sure how to represent repeating units without just duplicating subdiagrams!
Many thanks! I have residuals going from before to after the MHA block, and from before to after MLP+FFNN. Should I have others in other places?
edit: Those links are super!
I think there's this problem that occurs often when explaining something technical. Let's say you wanted to explain how a computer works. You could pick a specific computer, e.g. a Macbook Pro Retina. This would hang together nicely as an item, you could explain, in principle, why each specific bit of it was there and was designed the way it was. Of course, this would be insanely complicated for a beginner. They'd come away without understanding (in part because they can't triangulate from other computer architectures to what you've taught them about this one), nor intuition (because they can't leverage what you've taught them about this one to others).
You could instead come... (read 279 more words →)
Many thanks!