Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Simply explained: How does GPT work?

confusedbit.dev

Post reply on HN
Print view View original

71–80 of 392 posts

Re: Simply explained: How does GPT work?

#71

Post by seydor »

nitnelave wrote: post #50
seydor wrote: post #17

This is confusing, using the semantic vectors arithmetic of embeddings is not very relevant to transformers and its completely missing the word 'attention'. I don't think transformers are that difficult to explain to people , but it is hard to explain "why" they work. But i think it's important for everyone to look under the hood and know that there are no demons underneath.

I was trying to keep the article at a level that everyone understands, from middle school up. I thought about going a bit deeper in the structure and mentioning attention, but my problem is that the intuitive concept of "attention" is quite different from the mathematical reality of an attention layer, and I'm sure I would have lost quite a few people there. It's always a trade-off :)

Indeed the word 'attention' implies some active process, which is also unfortunate, because it is really similarity maps over long ranges. This process apparenty creates a cross-reference maps of items that are commonly co-activated. But then what do the different heads do? Some kind of different "modalities of cross-reference"?

I struggle to understand why this thing works the way it does. It's possible that Vaswani et al. have made one of the greatest discoveries of this century that solved the language problem in an unintuitive, and yet very unappreciated way. It's also possible that there are other architectures that can simulate the same level of intelligence with such large numbers of parameters.

I think you re right that it's not intuitive, it's like basic arithmetic is laughing at us

Re: Simply explained: How does GPT work?

#72

Post by myrmidon »

TuringTest wrote: post #59
myrmidon wrote: post #48

Earlier quoted context omitted.

What gives you any confidence that the way GPT4 comes up with answers is qualitatively different from humans? Why should the emulation of human though, a result of unguided evolution, require anything more than properly wired silicon?

> What gives you any confidence that the way GPT4 comes up with answers is qualitatively different from humans? For a start, GPT-4 doesn't include in its generation the current state of its internal knowledge used so far; any text built can only use at most the few words already generated in the current session as a kind of short-term memory. Biological brains OTOH have a rhythm with feedback mechanisms which adapt t…

> For a start, GPT-4 doesn't include in its generation the current state of its internal knowledge used so far

Sure. But are you certain that you NEED write access to long term memory to think? Would your thinking capabilities degrade meaningfully if that was taken away?

Re: Simply explained: How does GPT work?

#73

Post by tel »

alkonaut wrote: post #41

What I wonder most is how it encodes knowledge/state other than in the sequence of queries/responses. Does it not have a "mind"? If I play a number guessing game, can I tell it to "think of a number between 0 and 100" and then tell me if the secret number is higher/lower than my guess (For a sequence of N guesses where it can concistently remember it's original number)? If not, why? Because it doesn't have context? I…

Yes, you absolutely can. I just did it and despite an initial warning that it couldn't play the game (always fun to see these) it absolutely did.

The way it works is that each time it's tasked to produce a new response, it can view the entire history of the game. It knows that if it's said "higher" to 65 then it would be inconsistent to say "lower" to 64. Eventually this process terminates and the AI admits I "got" the number. The chat transcript up to that point is consistent with a "win".

What's wild though is that I can ask it to "regenerate" it's response. Over and over. Using this, I can convert a situation where a transcript which leads to a "too high" response into one that reads "too low". I'm, in essence, simulating fresh games each time and sampling over the choices of random numbers that GPT offers.

But it should also break the illusion of GPT specifically "having a mind". As I was chatting with it interactively, it was not really selecting a number but instead evaluating the probability of my particular guess sequence having the set of responses it actually saw. It then samples possible continuations. The more questions I've asked (and the more informative they were) the less variation remains in that selection of possible consistent continuations.

Or perhaps more consistent is the idea that within any single "call" to GPT to generate one further token (not even one further response) it may "have a mind", a particular choice of number, or it may not. It's actual behavior is indistinguishable either way. A whole chat dialogue, indeed even the rolling out of tokens from a single response it gives, are certainly (autoregressive) probabilistic samples over this process in either case.

(Edit, also worth noting that some evidence suggests GPT, including 4, is pretty bad at randomly drawing numbers.)

Re: Simply explained: How does GPT work?

#74

Post by ryanwaggoner »

mithr wrote: post #66
alkonaut wrote: post #41

What I wonder most is how it encodes knowledge/state other than in the sequence of queries/responses. Does it not have a "mind"? If I play a number guessing game, can I tell it to "think of a number between 0 and 100" and then tell me if the secret number is higher/lower than my guess (For a sequence of N guesses where it can concistently remember it's original number)? If not, why? Because it doesn't have context? I…

It doesn't have a mind. When you ask it "think of a number between 0 and 100", it delivers the most statistically likely (with some random variance) response to your input text, based on what it's been trained on. In other words, it's seen (via its extremely large training set) that when asked that specific question, the response is most often a character from a particular set of characters, which happens to represen…

It's more complicated than that, because it can competently answer many questions that it has a 0% chance of having seen before.

Re: Simply explained: How does GPT work?

#76

Post by nicpottier »

EForEndeavour wrote: post #32
onetrickwolf wrote: post #7

I've been using GPT4 to code and these explanations are somewhat unsatisfactory. I have seen it seemingly come up with novel solutions in a way that I can't describe in any other way than it is thinking. It's really difficult for me to imagine how such a seemingly simple predictive algorithm could lead to such complex solutions. I'm not sure even the people building these models really grasp it either.

The advanced capabilities of scaled up transformer models fed oodles of training data has burdened me with pseudo-philosophical questions about the nature of cognition that I am not well equipped to articulate, and make me wish I'd studied more neuroscience, philosophy, and comp sci earlier in life. A possibly off-topic thought dump: - What is thinking, exactly? - Does human (or superhuman) thinking require conscious…

The hypothesis that I find most compelling and intuitive is that language is thought and vice versa. We made a thing really good at language and it turns out that's also pretty good at thought.

One possible conclusion might be that the only thing keeping GPT algos from going full AGI is a loop and small context windows.

Re: Simply explained: How does GPT work?

#77

Post by danaris »

rootusrootus wrote: post #49
vadansky wrote: post #31

Earlier quoted context omitted.

I hate being the bearish guy during the hype cycle, but I think a lot of that is just anthropomorphizing it. They fed it TBs of human text, it spits out human text, we think it's humanesque. Of course maybe I'm wrong and it's AGI and it will find this comment and torture me for for insulting it's intelligence.

> I hate being the bearish guy No, please keep it up. Someone needs to keep pushing back against all the "I don't understand it, but it says smart-sounding things, and I don't understand the human brain either, so they're probably the same, it must be sentient!" It's a pretty handy technology, to be sure. But it's still just a tool.

Yeah; there's way too much "humanity of the gaps" here recently.

We don't have to fully understand the brain, or fully understand what LLMs are doing, to be able to say that what LLMs are doing is neither that close to what the brain does, nor anything that we would recognize as consciousness or sentience. There is enough that we do understand about those things—and the ways in which they differ—to be able to say with great confidence that we are not particularly close to AGI with this.

Re: Simply explained: How does GPT work?

#78

Post by ryanwaggoner »

Workaccount2 wrote: post #44
EForEndeavour wrote: post #32

Earlier quoted context omitted.

The advanced capabilities of scaled up transformer models fed oodles of training data has burdened me with pseudo-philosophical questions about the nature of cognition that I am not well equipped to articulate, and make me wish I'd studied more neuroscience, philosophy, and comp sci earlier in life. A possibly off-topic thought dump: - What is thinking, exactly? - Does human (or superhuman) thinking require conscious…

> has burdened me with pseudo-philosophical questions about the nature of cognition that I am not well equipped to articulate, and make me wish I'd studied more neuroscience, philosophy, and comp sci earlier in life Welcome to the club. There pretty much are no answers, just theories primarily played out as thought experiments. Its on of those areas where you can pick out who knows less (or is being disingenuous) by…

It's interesting to me how many commenters on HN are absolutely convinced that GPT4 is incapable of thought or understanding or reasoning, it's "just" predicting the next word. And then they'll insist that it'll never be able to do things that it's already capable of doing...

Interestingly, more than one of these folks have turned out to be religious. I wonder if increasingly intelligent AI systems will be challenging for religious folks to accept, because it calls into question our place at the pinnacle of God's creation, or it casts doubt upon the existence of a soul, etc.

Re: Simply explained: How does GPT work?

#80

Post by prerok »

agentultra wrote: post #29

A good article and well articulated! I would change the introduction to be more impartial and not anthropomorphize GPT. It is not smart and it is not skilled in any tasks other than that for which it is designed. I have the same reservations about the conclusion. The whole middle of the article is good. But to then compare the richness of our human experience to an algorithm that was plainly explained? And then to sp…

I agree that it is a good effort. I don't agree with the criticism.

It really depends on who the target audience is. There's been a lot of scare mongering in the news about it lately and I think the last part tries to address that. It first offers an explanation that my parents can understand and then addresses what they have been hearing about in the news.

So, I would say it is great to share it with them and I think they are the intended audience.

Post reply on HN