The most cited paper of the century is a brilliant hack
12community score
2mentions
2discussions
2sharers
2days active
—story pointsall mentions are comments
Hacker News discussions
2 independent threadsWe're releasing the weights of Mistral Large 4
“AFAICT there are several things that affect the size of the model: 1. the number of transformer layers -- e.g. each layer adds the ability to provide more information to the residual stream [1], [2]; …”— rhdunn, when sharing this video · read the comment ↗+ 1 comment mention in this thread ↗
OpenAI's agents talked to each other through the CPU Cache
“I recently learned that NN layers talk to each other via unused neurons. https://www.youtube.com/watch?v=QgH9sr7G13Q”— est, when sharing this video · read the comment ↗+ 1 comment mention in this thread ↗