141–150 of 396 posts
Thank God there is still neverending wars, otherwise authoritarian governments would have no fun left.
Earlier quoted context omitted.
High dimensional vectors are thought (insofar as you can define what that even means). Tokens are one dimensional input that navigates the thought, and output that renders the thought. The "thinking" takes place in the high dimension space, not the one dimensional stream of tokens.
But isn't the one dimensional tokens a reflex of high dimensional space? What you see is "sure let's take a look at that" but behind the curtains it's actually an indication that it's searching a very specific latent space which might be radically different if those tokens didn't exist. Or not. In any case, you can't just make that claim and isolate those two processes. They might be totally unrelated but they also m…
Okay, I like how it reduces token usage, but it kind of feels that, it will reduce the overall model intelligence. LLMs are probabilistic models, and you are basically playing with their priors.
Earlier quoted context omitted.
[flagged]
Can't you know that tokens are units of thinking just by... like... thinking about how models work?
Seems reasonable, but this doesn't settle probably-empirical questions like: (a) to what degree is 'more' better?; (b) how important are filler words? (c) how important are words that signal connection, causality, influence, reasoning?
Earlier quoted context omitted.
But isn't the one dimensional tokens a reflex of high dimensional space? What you see is "sure let's take a look at that" but behind the curtains it's actually an indication that it's searching a very specific latent space which might be radically different if those tokens didn't exist. Or not. In any case, you can't just make that claim and isolate those two processes. They might be totally unrelated but they also m…
I assume in practice, filler words do nothing of value. When words add or mean nothing (their weights are basically 0 in relation to the subject), I don't see why they'd affect what the model outputs (except cause more filler words)?
Oh boy. Someone didn't get the memo that for LLMs, tokens are units of thinking . I.e. whatever feat of computation needs to happen to produce results you seek, it needs to fit in the tokens the LLM produces. Being a finite system, there's only so much computation the LLM internal structure can do per token, so the more you force the model to be concise, the more difficult the task becomes for it - worst case, you ca…
However, another potential issue is that LLMs are continuation engines, and I'd have thought that talking like a caveman may be "interpreted" as meaning you want a dumbed down response, not just a smart response in caveman-speak.
It's a bit like asking an LLM to predict next move in a chess game - it's not going to predict the best move that it can, but rather predict the next move that would be played given what it can infer about the ELO rating of the player whose moves it is continuing. If you ask it to continue the move sequence of a poor player, it'll generate a poor move since that's the best prediction.
Of course there's not going to be a lot of caveman speak on stack overflow, so who knows what the impact is. Program go boom. Me stomp on bugs.
Earlier quoted context omitted.
Can't you know that tokens are units of thinking just by... like... thinking about how models work?
Can't you just know that the earth is the center of the world by... like... just looking at how the world works?
There's linguistic term for this kind of speech: isolating grammars, which don't decline words and use high context and the bare minimum of words to get the meaning across. Chinese is such a language btw. Don't know what Chinese think about their language being regarded as cavemen language...
All languages must have means for marking the syntactic roles of the words in a sentence.
The roles may be marked with prepositions or postpositions in isolating languages, or with declensions in fusional languages, or there may be no explicit markers when the word order is fixed (i.e. the same distinction as between positional arguments and arguments marked by keywords, in programming languages). The most laconic method for both programming languages and natural languages is to have a default word order where role markers are omitted, but to also allow any other word order if role markers are present.
Besides the mandatory means for marking syntactic roles, many languages have features that add redundancy without being necessary for understanding, i.e. which repeat already known information, for instance by repeating the information about gender and number that is attached to a noun also besides all its attributes. Whether a language requires redundancy or not is independent on whether it is an isolating language or a fusional language.
English has somewhat less syntactic role markers than other languages because it has a rigid word order, but for the other roles than the most frequent roles (agent, patient, beneficiary) it has a lot of prepositions.
Despite being more economic in role markers, English also has many redundant words that could be omitted, e.g. subjects or copulative verbs that are omitted in many languages. Thus for English it is possible to speak "like a caveman" without losing much information, but this is independent of the fact that modern English is a mostly isolating language with few remnants of its old declensions.
Oh boy. Someone didn't get the memo that for LLMs, tokens are units of thinking . I.e. whatever feat of computation needs to happen to produce results you seek, it needs to fit in the tokens the LLM produces. Being a finite system, there's only so much computation the LLM internal structure can do per token, so the more you force the model to be concise, the more difficult the task becomes for it - worst case, you ca…
I wonder if a language like Latin would be useful. It's a significantly much succinct semantic encoding than English while being able to express all the same concepts, since it encodes a lot of glue words into the grammar of the language, and conventionally lets you drop many pronouns. e.g. "I would have walked home, but it seemed like it was going to rain" (14 words) -> "Domum ambulavissem, sed pluiturum esse videba…