“Next-token predictor” is the wrong mental model for LLMs
gmcgoldr.github.io
“Next-token predictor” is the wrong mental model for LLMs
1–10 of 312 posts
Re: “Next-token predictor” is the wrong mental model for LLMs
#2A better statement might be:
Current agentic systems may be *built* from next-token predictors which are conceptually simple, but because of agentic frameworks, recursive invocation, tool use, and *heavy* investment in reinforcement learning in these contexts and for specific applications, they can no longer be thought of as "Merely" next token predictors.
Re: “Next-token predictor” is the wrong mental model for LLMs
#3It's the fitness function: Make a model which is capable of predicting the next token. The next token of what? EVERYTHING.
So what does this lead to? To a generic intelligence which is capable of responding/answering everything.
If overfitted, the model just remembers every possibility in the world but this is not possible anyway so it will start to identify patterns and rules and will use them instead.
Basically 'compressing' every possibility to every question someone could ask -> compression leads to intelligence.
Re: “Next-token predictor” is the wrong mental model for LLMs
#4yep "next-embedding" predictor is more correct, and not just at the end but through the layers, and folding back dimensions into that one next token is one small final step, and next-embedding could be named "next-meaning" as well, and we're getting there...
this sentence above would made a longer article if I bothered to so blog as is being blogged here
Re: “Next-token predictor” is the wrong mental model for LLMs
#5> Calling the second system a “next-move predictor” would be strange. It is not trying to predict what move appeared next in a dataset. It is trying to choose a move that wins.
i dont understand the distinction here. does working backwards from a set of win states instead of working forwards from the current state somehow change whether it’s a prediction or not?
Re: “Next-token predictor” is the wrong mental model for LLMs
#6I understand how a computer can know that a chess move is more likely to lead to a win, and therefore “correct”, but I don’t understand how it can know that a token is correct. Can someone explain?
Re: “Next-token predictor” is the wrong mental model for LLMs
#7slop slop and more slop
Re: “Next-token predictor” is the wrong mental model for LLMs
#8> Calling the second system a “next-move predictor” would be strange. It is not trying to predict what move appeared next in a dataset. It is trying to choose a move that wins. i dont understand the distinction here. does working backwards from a set of win states instead of working forwards from the current state somehow change whether it’s a prediction or not?
The distinction is that it's not 'predicting the next token'. Instead it's _determining_ the next token based on a prediction of its reward signal.
Re: “Next-token predictor” is the wrong mental model for LLMs
#9Sure, I get the gist of the article. I have never liked the reductionist argument that LLMs are nothing more than next-token predictors. By that rational, the human brain is really not that much different. When I am having a conversation with another person, I do not usually have every word I will respond with stored in my limited working memory. My output is often predicted based on the previous word I spoke.
Re: “Next-token predictor” is the wrong mental model for LLMs
#10Better title: Continue thinking of LLMs as Next-Token Predictors
Because no, post training doesn't change that.