i've been trying to understand the transformer architecture for ages but it never clicked. so recently i started trying to learn its history and that has helped a LOT. to try to cement what i learned imma write up a essay.
Undefined Behavior presents: The Evolution of Attention
one of my history superstitions is that we VASTLY overestimate how much progress is bottlenecked by knowledge and VASTLY underestimate how much it's bottlenecked by infrastructure and capital. knowledge is worthless if it's too early to exploit it
I wonder what's the smallest amount of today's knowledge you'd need to give to the Romans back in 27 B.C. in order to spark industrialization/modernity roughly equivalent to our own but two millennia earlier
every paper that mentions category theory defines what a category is and then two pages later is throwing the traced monoidal cocartesian yoneda triple backflip at you. mfer if i didn't know what a category was before reading this paper there is no way id pick it up now.
idk what your background is, so apologize if this is gibberish, but ML with just matrix multiplications is linear regression. a lot of interesting probability distributions are not very well approximated by a line of best fit, which is what linear regression does, and need
u could wrap up ur securitized debts with other ingredients and then hold the debt package together with a flour tortilla to make a debt burrito for your debt burrito
โayahausca copypastaโ is insane because it appears to be one of the most legitimately dangerous memes with the potential to gigafry your brain but is exclusively used by literal turbonormies who unironically want to like "say something funny" and basically get oneshotted by it
if you are in vibe coding, pivot to vibe proving. im making $0k/year using chatgpt to write proofs in lean4 without ever having gone to grad school. you can too. here's how: