Differentiating between the 3 fastai layers

I am in the intermediatish zone with the course. Got to know a lot about the fast.ai library (how it has these multiple layers and so on).

It gave me clarity to focus on code layer wise and not feel insecure too much with what’s going on behind the scenes. I mean, when we are using the high-level API with which we begin the course, what more can we do? You grasp all that layer can provide, relax yourself with the questions, and enter the next layer and then satiate your curiosity by going deeper and tweaking stuff.

Anyways, I wanted to know what one can leverage from at each stage of these API layers. When to go deeper and when it is not required to? Curiosity might always take one to jump into other layers but from the POV of practicality, what are the distinguishable stuff these layers provide? What if one choses not at all to bother with other layers and just wants to do great things with the higher-level API? Can he/she do it?

Also, am I right on these:

  • Part 1 covers the high-level API and (sort of a little of the medium-level API and part 2 covers the low-level API (Deep Learning from Foundations).
  • The book covers all the layers.

My new course is based solely on this explanation, and I highly recommend it as the next stage :slight_smile: