Every attention weight matrix in GPT-2, visualized

Haha, yeah I was kind of thinking that as well! Like with different error-diffusion patterns, one method may be more visually appealing than the other.

Although, with either approach, I definitely feel that the fact that sRGB is non-linear should be acknowledged, and that’s something I was completely unaware of. So, I’m happy I learned something new today :)

Hi, OP here!

Thank you so much for pointing this out! Just read the post you linked and did some of my own research on the non-linearity of sRGB - really fascinating stuff :)

For now, I've acknowledged this limitation of my implementation so that any new readers are aware of it: https://amanvir.com/blog/writing-my-own-dithering-algorithm-...

But I'll definitely revisit the article to add proper linearization to my implementation when I have the time. Thanks again for mentioning this!

I love them too :)

Visual Studio Code with this extension: https://marketplace.visualstudio.com/items/?itemName=kisstko...

Writing my own dithering algorithm in Racket