
Boris Dayma — The Story Behind DALL·E mini, the Viral Phenomenon
Gradient Dissent: Conversations on AI
00:00
Is It a Convolution?
Each patch is about like a 16 by 16. You don't notice that the others are big patches, because it's a unete. And ar the patchesan like a gred, like a two by two gred. T, thats right. Thon, agreed. So often i thing, for me, the picture is divided. The idea was that, the idea was that it would maybe more efficient. My model is slightly different in the sense that ave the anqader, it's like a translation model where the ancuader and de is separate.
Transcript
Play full episode