Yes, You Need To Work Through Concrete Examples

They may feel tedious, they may get under your skin, but that's the only way they get into your bones.

Want to get notified about new posts? Join the mailing list and follow on X/Twitter.

Yes, you need to work through concrete examples. They may feel tedious, they may get under your skin, but that’s the only way they get into your bones.

If they’re not in your bones, you’re just cargo-culting abstractions that you don’t really understand at a visceral level, and unable to push boundaries because you don’t actually know where they are.

“You really need to do those computations, as painful as they are, you really need to do them so they get under your skin, into your bones. So when you actually come to code something like that, you know exactly what’s going on.

Yeah, the computations are gonna get under your skin in more ways than one. I know you didn’t love it, but now you actually know it. You’ve mastered it. Intuition through repetition.

Lots of people start the top down method of becoming machine learning engineers and never really do the bottom up part. So they kind of vaguely know what gradient descent is, but not really. And that’s not a good position to be in, if you really want to make cutting edge technology or push the boundaries on things.

It is a necessary evil to kind of go through calculations by hand. That’s where you learn, you develop that intuition, and then the abstractions all fall into place.

But if you try and skip the concrete examples, skip the repetitions and go straight to the abstract, you just cargo cult math, you’re just parroting stuff. You sound like you know what you’re talking about, but you can’t really solve anything.”

(weaving together snippets from our discussion ~1:16:48)

Want to get notified about new posts? Join the mailing list and follow on X/Twitter.