Finding documentation and possible arguments

fastai
1

Hey everyone,

I’ve just starting with the course, python and fast.ai. Now I’m trying to build an image classifier not looking at existing code from lesson 2/3. I have trouble figuring out what options I have, for example for creating a DataBlock.

First, I’m trying ?DataBlock and ??Datablock, giving me some insights into which arguments are there and the source code.
Then I look at the documentation using doc(Datablock). But from here, many things remain unclear, e.g. which blocks are there? Which get_items functions are there? What object type do the arguments have to be?

Is the only way looking through existing code from other people? How do you managed to find your way around in the beginning?

Any help is very much appreciated.

2

Not sure how others did it, but unfortunately, this is pretty much what I did in addition to asking questions on the forum.

(Edit) Zach’s code below were the main source outside of the fastbook for my getting comfortable fastai.

2 Likes
3

I wrote numerous examples to try and help folks out with how different DataBlocks can look like in https://walkwithfastai.com, but yes much of it is looking at examples and mimicking them. (Personally, you should look at the code when learning it. There isn’t a need to memorize it all, I definitely haven’t)

3 Likes
4

Thanks for this, looks like a great resource!