Keep on to blur preview images; turn off to show them clearly
Thread
行道途中。非求速成，惟求通达。 2023 年扎进AI ，打通Know-How，不少赚钱项目，踩过坑，也见过光。 围城里待得够久了，出来聊聊世界，聊聊技术、聊聊赚钱。
Landing page design tip: communicate the value with a before and after visual. It converts 🔥
UI/UX Designer ϟ Prev: lead designer at @super_ ✱ UI inspiration @DamnGoodUI
could we just collectively rise as a people and just refuse to change our clocks? 4:30pm sunset is outrageous.
gpus and tractors Neural networks from Scratch book: https://t.co/hyMkWyUP7R https://t.co/8WGZRkUGsn
*overfit to recent data finally, how would you create these boundaries where you decide to minimise KLD? if temporally decided, it's suboptimal if decided based on abstract ideas, it runs into the same challenges as BLT or LCM.
RL and efficient distributed pretraining • eXperiments lab • memes and training lores
the cost angle is definitely one of the factors. another is that it all gets handled at scale and we shouldn't overfit the data in pretraining that can be noisy and possibly lower quality.
*overfit to recent data finally, how would you create these boundaries where you decide to minimise KLD? if temporally decided, it's suboptimal if decided based on abstract ideas, it runs into the same challenges as BLT or LCM.
https://t.co/k8dEZBW68s
Wonderer. Amor fati. Scaling trust.