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

TimeCapsuleLLM: LLM trained only on data from 1800-1875

github.com

Post reply on HN
Print view View original

111–120 of 334 posts

Re: TimeCapsuleLLM: LLM trained only on data from 1800-1875

#111

Post by pegasus »

echoangle wrote: post #90
scottlamb wrote: post #77

Earlier quoted context omitted.

What's the bar here? Does anyone say "we don't know if Einstein could do this because we were really close or because he was really smart?" I by no means believe LLMs are general intelligence, and I've seen them produce a lot of garbage, but if they could produce these revolutionary theories from only <= year 1900 information and a prompt that is not ridiculously leading, that would be a really compelling demonstrati…

> Does anyone say "we don't know if Einstein could do this because we were really close or because he was really smart?" Kind of, how long would it have realistically taken for someone else (also really smart) to come up with the same thing if Einstein wouldn't have been there?

But you're not actually questioning whether he was "really smart". Which was what GP was questioning. Sure, you can try to quantify the level of smarts, but you can't still call it a "stochastic parrot" anymore, just like you won't respond to Einstein's achievements, "Ah well, in the end I'm still not sure he's actually smart, like I am for example. Could just be that he's just dumbly but systematically going through all options, working it out step by step, nothing I couldn't achieve (or even better, program a computer to do) if I'd put my mind to it."

I personally doubt that this would work. I don't think these systems can achieve truly ground-breaking, paradigm-shifting work. The homeworld of these systems is the corpus of text on which it was trained, in the same way as ours is physical reality. Their access to this reality is always secondary, already distorted by the imperfections of human knowledge.

Re: TimeCapsuleLLM: LLM trained only on data from 1800-1875

#112

Post by mlinksva »

metalliqaz wrote: post #42
dogma1138 wrote: post #2

Would be interesting to train a cutting edge model with a cut off date of say 1900 and then prompt it about QM and relativity with some added context. If the model comes up with anything even remotely correct it would be quite a strong evidence that LLMs are a path to something bigger if not then I think it is time to go back to the drawing board.

Yann LeCun spoke explicitly on this idea recently and he asserts definitively that the LLM would not be able to add anything useful in that scenario. My understanding is that other AI researchers generally agree with him, and that it's mostly the hype beasts like Altman that think there is some "magic" in the weights that is actually intelligent. Their payday depends on it, so it is understandable. My opinion is that…

Do you have a pointer to where LeCun spoke about it? I noticed last October that Dwarkesh mentioned the idea off handedly on his podcast (prompting me to write up https://manifold.markets/MikeLinksvayer/llm-trained-on-data-...) but I wonder if this idea has been around for much longer, or is just so obvious that lots of people are independently coming up with it (parent to this comment being yet another)?

Re: TimeCapsuleLLM: LLM trained only on data from 1800-1875

#113

Post by nickdothutton »

dogma1138 wrote: post #2

Would be interesting to train a cutting edge model with a cut off date of say 1900 and then prompt it about QM and relativity with some added context. If the model comes up with anything even remotely correct it would be quite a strong evidence that LLMs are a path to something bigger if not then I think it is time to go back to the drawing board.

I would love to ask such a model to summarise the handful of theories or theoretical “roads” being eyed at the time and to make a prediction with reasons as to which looks most promising. We might learn something about blind spots in human reasoning, institutions, and organisations that are applicable today in the “future”.

Re: TimeCapsuleLLM: LLM trained only on data from 1800-1875

#114

Post by nickpsecurity »

dogma1138 wrote: post #2

Would be interesting to train a cutting edge model with a cut off date of say 1900 and then prompt it about QM and relativity with some added context. If the model comes up with anything even remotely correct it would be quite a strong evidence that LLMs are a path to something bigger if not then I think it is time to go back to the drawing board.

That would be an interesting experiment. It might be more useful to make a model with a cut off close to when copyrights expire to be as modern as possible.

Then, we have a model that knows quite a bit in modern English. We also legally have a data set for everything it knows. Then, there's all kinds of experimentation or copyright-safe training strategies we can do.

Project Gutenberg up to the 1920's seems to be the safest bet on that.

Re: TimeCapsuleLLM: LLM trained only on data from 1800-1875

#115

Post by samuelson »

dogma1138 wrote: post #2

Would be interesting to train a cutting edge model with a cut off date of say 1900 and then prompt it about QM and relativity with some added context. If the model comes up with anything even remotely correct it would be quite a strong evidence that LLMs are a path to something bigger if not then I think it is time to go back to the drawing board.

I think it would be fun to see if an LLM would reframe some scientific terms from the time in a way that would actually fit in our current theories.

I imagine if you explained quantum field theory to a 19th century scientists they might think of it as a more refined understanding of luminiferous aether.

Or if an 18th century scholar learned about positive and negative ions, it could be seen as an expansion/correction of phlogiston theory.

Re: TimeCapsuleLLM: LLM trained only on data from 1800-1875

#118

Post by feanaro »

pegasus wrote: post #101
palmotea wrote: post #94

Earlier quoted context omitted.

But that's not the OP's challenge, he said "if the model comes up with anything even remotely correct ." The point is there were things already "remotely correct" out there in 1900. If the LLM finds them, it wouldn't "be quite a strong evidence that LLMs are a path to something bigger."

It's not the comment which is illogical, it's your (mis)interpretation of it. What I (and seemingly others) took it to mean is basically could an LLM do Einstein's job ? Could it weave together all those loose threads into a coherent new way of understanding the physical world? If so, AGI can't be far behind.

This alone still wouldn't be a clear demonstration that AGI is around the corner. It's quite possible a LLM could've done Einstein's job, if Einstein's job was truly just synthesising already available information into a coherent new whole. (I couldn't say, I don't know enough of the physics landscape of the day to claim either way.)

It's still unclear whether this process could be merely continued, seeded only with new physical data, in order to keep progressing beyond that point, "forever", or at least for as long as we imagine humans will continue to go on making scientific progress.

Re: TimeCapsuleLLM: LLM trained only on data from 1800-1875

#119

Post by anthonypasq »

opponent4 wrote: post #102
strbean wrote: post #86

Earlier quoted context omitted.

> And no, the "brain is a computer" is not a scientific description, it's a metaphor. Disagree. A brain is turing complete, no? Isn't that the definition of a computer? Sure, it may be reductive to say "the brain is just a computer".

Not even close. Turing complete does not apply to the brain plain and simple. That's something to do with algorithms and your brain is not a computer as I have mentioned. It does not store information. It doesn't process information. It just doesn't work that way. https://aeon.co/essays/your-brain-does-not-process-informati...

ive gotta say this article was not convincing at all.
Post reply on HN