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

Billion-Parameter Theories

worldgov.org

Post reply on HN
Print view View original

71–80 of 92 posts

Re: Billion-Parameter Theories

#71

Post by jayd16 »

phyzix5761 wrote: post #62
wavemode wrote: post #58

Not to sound condescending, but this reads like someone fimiliar with LLMs but very unfamiliar with statistics in general. If we could understand economics, or poverty, or any number of other social structures, simply by cramming data into a statistical model with billions of parameters, we would've done that decades ago and these problems would already be understood. In the real world, though, there is a phenomenon…

I could be wrong but I think we crossed the 1 billion parameter threshold in 2019. I'm not sure we had this ability for decades.

They mean with traditional hard computation, not LLM magic.

Re: Billion-Parameter Theories

#72

Post by gozucito »

xikrib wrote: post #26

Let's gather authors of 15 different world languages together in a room and see if they can collaboratively write a short story. Surely their inability to do so will prove their inadequacy in their native language. /s Simplicity brings us closer to truth — Occam's razor has underpinned the development of our species for centuries. It's enterprise, empire, and capital that feed off of complexity. We're entering a peri…

>Simplicity brings us closer to truth — Occam's razor has underpinned the development of our species for centuries.

I keep thinking of emergent complexity. Even starting with very simple rules and components, the amount of complexity that arises as a consequence of ever rising interactions can boggle the mind and seems to validate our current predilection for elegant and succinct laws of physics to be enough to model the universe.

Coincidentally, LLMs being so good at coding that it became the #1 source of income for Anthropic is one such example of emerging complexity from deceptively simple ingredients:

A giant pile of matrix multiplies, next-token prediction, and enough data somehow climbs the ladder from autocomplete to writing code well enough that people will pay $20-200/month per seat for it. It is completely bonkers.

Re: Billion-Parameter Theories

#73

Post by DoctorOetker »

JackFr wrote: post #43
pdonis wrote: post #25

Earlier quoted context omitted.

> I also don't think poverty is a complex problem, but that's a minor point. I'm not sure it's a minor point. I don't think poverty is a "complex" problem either, as that term is used in the article, but that doesn't mean I think it fits into one of the other two categories in the article. I think it is in a fourth category that the article doesn't even consider. For lack of a better term, I'll call that category "po…

When all of humanity was hunting and gathering and living at subsistence levels, the was no poverty. It only shows up with wealth. Pretty simple.

This.

Every sedentary society has historically scared its members of the dangers of the nomadic lifestyle, heathens, ...

The implied conclusion being that since our ancestors switched from nomadic to sedentary it must have been preferable, a kind of informal democratic collectively and individually approved choice.

Surely sedentary must have been better, how else could such a transition have been sustained?

Rather easy how else: its perfectly possible for average or mean life quality under sedentary lifestyle to be a net setback compared to nomadic lifestyle, since slavery can't be effectively implemented in a nomadic lifestyle, whereas the sedentary lifestyle creates both the demand for labor (routine monotonous work in the fields) and the means to enable slavery (escaping nomadic tribes under Brownian motion is much easier than escaping from a randomly assigned position deep in a larger sedentary empire, even if you escape the sedentary village, the stable neighbouring village will happily return you to "your owner" so that he would hopefully return the favor if ever he catches one of "their slaves").

It's easy to claim a net improvement in life quality ... by discounting the loss of life quality of the slaves!

Nomadic lifestyle was simply outcompeted by sedentary-enabled slavery!

Re: Billion-Parameter Theories

#74

Post by _hark »

harperlee wrote: post #5

Two handwavey ideas upon reading this: - Even for billion-parameter theories, a small amount of vectors might dominate the behaviour. A coordinate shift approach (PCA) might surface new concepts that enable us to model that phenomenon. "A change in perspective is worth 80 IQ points", said Alan Kay. - There is analogue of how we come up with cognitive metaphors of the mind ("our models of the mind resemble our latest…

You literally can do a kind of model PCA, using the Hessian (matrix of second derivatives of the loss function w/r/t the parameters, aka the local curvature of the loss landscape), and diagonalizing. These eigenvectors and eigenvalues (the spectrum of the Hessian) tend to be power-law distributed in just about every deep NN you can think of [1].

That is, there are a few "really important" (highly curved) dimensions in parameter space (the top eigenvectors) which control the model's performance (the loss function). Conversely, there are very many "unimportant"/low curvature dimensions in the model. There was a recent interesting paper that showed that "deleting" these low-curvature dimensions appeared to correspond to removing "memorized" information in LLMs, such that their reasoning performance was left unchanged while their ability to answer questions which require some memorized knowledge was reduced [2].

It appears that sometimes models undergo dramatic transitions from memorization to perfect generalization, which corresponds to the models becoming much more compressible [3].

I'm hopeful that we'll find a way to distill the models down to the most useful core cognitive/reasoning capabilities, and that that core will be far simpler than the current scale of LLMs. But they might need to look stuff up like we do without all that memorized world knowledge!

[1]: https://openreview.net/pdf?id=o62ZzfCEwZ

[2]: https://www.goodfire.ai/research/understanding-memorization-...

[3]: https://arxiv.org/abs/2412.09810

Re: Billion-Parameter Theories

#75

Post by galaxyLogic »

empath75 wrote: post #44
galaxyLogic wrote: post #22

Earlier quoted context omitted.

Mark Cubain had a good line, I don't know if he came up with it or who, but he reportedly said: " There are 2 types of people using AI: Those who use it so they can know everything, and those who use it so they don't have to know anything. " :-

I think probably the sweet spot is using them so you can focus on knowing only the things you care about or need to know about about.

True, but I think the observation is spot on. There are people who want to know things, whether with some help from AI or by other means. Then there are people who prefer ignorance.

Personally I take great comfort from the fact that I no longer, to a large degree, face the dilemma of "Who should I ask about this?"

Re: Billion-Parameter Theories

#76

Post by pdonis »

DoctorOetker wrote: post #73
JackFr wrote: post #43

Earlier quoted context omitted.

When all of humanity was hunting and gathering and living at subsistence levels, the was no poverty. It only shows up with wealth. Pretty simple.

This. Every sedentary society has historically scared its members of the dangers of the nomadic lifestyle, heathens, ... The implied conclusion being that since our ancestors switched from nomadic to sedentary it must have been preferable, a kind of informal democratic collectively and individually approved choice. Surely sedentary must have been better, how else could such a transition have been sustained? Rather ea…

If you think a subsistence nomadic lifestyle is preferable to a modern "sedentary" one, then how are you able to post here? Subsistence nomads don't have Internet access (to name just one of umpteen things we "sedentary" moderns have access to that they don't). There are ways to live off grid if you really think it's preferable.

Re: Billion-Parameter Theories

#77

Post by pdonis »

DoctorOetker wrote: post #73
JackFr wrote: post #43

Earlier quoted context omitted.

When all of humanity was hunting and gathering and living at subsistence levels, the was no poverty. It only shows up with wealth. Pretty simple.

This. Every sedentary society has historically scared its members of the dangers of the nomadic lifestyle, heathens, ... The implied conclusion being that since our ancestors switched from nomadic to sedentary it must have been preferable, a kind of informal democratic collectively and individually approved choice. Surely sedentary must have been better, how else could such a transition have been sustained? Rather ea…

> even if you escape the sedentary village, the stable neighbouring village will happily return you to "your owner" so that he would hopefully return the favor if ever he catches one of "their slaves")

Tell that to all the people who ran the Underground Railroad in the pre-Civil War US, not to mention all the other ways that Fugitive Slave laws were persistently violated.

I think you are vastly underestimating the benefits of a modern "sedentary" society. But as I pointed out in my other post, if you really don't think they're benefits, then you can simply forgo them. Go and live an off grid subsistence lifestyle. There are people who do that. But of course they don't post on the Internet.

Re: Billion-Parameter Theories

#78

Post by mistivia »

> The deepest truths fit on a napkin.

If you have really done physics or engineering, you would never believe this. Simple and elegant formulas usually can only solve the "spherical chicken in a perfect vacuum" kind of problems. The real world is incredibly messy. Beneath those clean and beautiful-looking partial differential equations lies a mathematical nightmare. And these equations often only hold at certain scales or rely on extremely strict boundary conditions.

Re: Billion-Parameter Theories

#79

Post by munificent »

js8 wrote: post #40
munificent wrote: post #29

Earlier quoted context omitted.

> I think it is always possible to come up with reasonably small theories that capture most of the given phenomena. I can write a program (call it a simulation of some artificial phenomenon) whose internal logic is arbitrarily complex. The result is irreducible: the entire byzantine program with all of its convoluted logic is the smallest possible theory to describe the phenomenon, and yet the theory is not reasonabl…

That's true but I can still approximate what the system does with a simpler model. For example, I can split states of the system into n distinct groups, and measure transition probabilities between them. Thermodynamics is a classic example of a phenomenological model like that.

> That's true but I can still approximate what the system does with a simpler model.

For any strategy you might apply to do that, I can craft a program that similates a phenomenon that defies that strategy.

Re: Billion-Parameter Theories

#80

Post by stevenhuang »

curao_d_espanto wrote: post #65
wavemode wrote: post #58

Not to sound condescending, but this reads like someone fimiliar with LLMs but very unfamiliar with statistics in general. If we could understand economics, or poverty, or any number of other social structures, simply by cramming data into a statistical model with billions of parameters, we would've done that decades ago and these problems would already be understood. In the real world, though, there is a phenomenon…

> The emerging field of mechanistic interpretability suggests otherwise. Researchers are developing tools to understand how neural networks do what they do, from network ablation and selective activation to feature visualization and circuit tracing. These techniques let you study a trained model the way a biologist studies an organism, through careful experimentation and observation. honestly, when I read that part o…

The author is not wrong. You seem unaware of how nascent the field of LLM interpretability research is.

See this thread and article from earlier today showing what we're still able to learn from these interpretability experiments.

https://news.ycombinator.com/item?id=47322887

Post reply on HN