Hacker News Reader

His whole blog is fantastic. If you want more background (e.g. how transformers work) he's got all the posts you need

This looks nice, thank you, but I'm looking for a more hands-on tutorial, with e.g. Python code, like Andrej Karpathy makes them.

SD3 is a new architecture using DiT (diffusion transformers), so those would be out of date.

The older ones have drawbacks like not being able to spell.

Not too out of date. Just replace the magic UNet with a DiT and squint. It’s doing the same thing - removing noise.

fast.ai has a whole free course

I liked this 18 minute video ( https://www.youtube.com/watch?v=1CIpzeNxIhU ). Computerphile has other good videos with people like Brian Kernighan.