Fork of fastai with longer variable names

Jeremy has made clear his preference for a “tight” format, with as much as possible visible in one eye-glance. His Style Guide asks for terse variable names, tight vertical spacing, and avoiding comments.

I appreciate that that works for Jeremy, and that it might make it faster to develop code.

However, it doesn’t work for me. In particular, I spend a lot of time trying to figure out what the variable names mean when I am reading the code.

I have thus forked fastai and am making a “looser” version: long, explanatory variable names, PEP8-compliant, and more comments. It’s at
in the branch called loose in case you want to use it to help you understand the code OR if you want to contribute. (Note: I need to keep master available for keeping in sync with the fastai code base.)

I guarantee that today there are things are currently broken, and it’s not nearly done, but I am working on it little by little. (I did make sure that Lesson1 – Dogs&Cats – does work.)


Are you taking about making this change for the lesson notebooks, or the library itself? I say leave the library as is. It’s his library. I don’t think it makes sense to maintain a separate version of the library just so it’s easier for non-core devs to work on. The lesson notebooks though! That could stand to benefit hugely from more explanatory var names. I also spent a lot of my time just trying to figure out what the var name referred to.

I’m working on both. I found that when I was trying to recreate a class notebook without looking at the original notebook, I had to dive into the library a lot – and I couldn’t figure out what was going on in the libraries.

1 Like

omg so useful