can mcts be made to work with quasi-infinite state space?

>figure made a manlet

tbh feels beyond stupid that every token is weighted the same in llm training loss, particularly for instruct or task-specific use cases. most tokens are filler and thus noise relative to what you want the model to really learn.

mogged with 17% of the effective compute

can anyone point me to research on using chain-of-thought (or tree-of-thought) to generate a synthetic dataset that basically converts system 2 thought (the chain or tree) into system 1 thought (just standard next token prediction)?

i assume someone has done this

a dynamic bundle of compute is all you need

“8b is not small enough”

yes, thank you mark, you’re damn right

cora since the aughts we’ve all received nonstop and pervasive media/social media pipe gen activity from VC’s and founders parading around as self actualization performance narratives. this has shifted egoistic preferences materializing in everyone thinking they need to be founders.

friendly reminder that these things are COMMODITIES and will end up priced to reflect that

zuck has the right strategy here. openai microsoft and anthropic do not.

Many people are hating on this video, but I actually think it's a fascinating display of the two very distinct modes that exist to relate with reality: mimesis vs. first principles thinking.

95% of people operate by mimesis. Truth doesn't matter to them as much as getting

lina Well, no, but why would I if it's not solving a problem anyone actually has?