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

Things we learned about LLMs in 2024

simonwillison.net

Post reply on HN
Print view View original

101–110 of 615 posts

Re: Things we learned about LLMs in 2024

#101

Post by airstrike »

antirez wrote: post #75
hdjjhhvvhga wrote: post #69

Earlier quoted context omitted.

While Claude Sonnet is superior than 4o for most my use cases, there are still occasionally some specific tasks where it performs slightly better.

Probably. But statistically to work with 4o is a lose of time for me. LLMs is like an investment: you write the prompts, you "work" with them. If the LLM is too weak, this is a lose of time. You need to have a return on the investment that is positive. With ChatGPT 4o / o1 most of the times for me the investment of time has almost zero return. Before Claude Sonnet 3.5 I already had a ChatGPT PRO account but never use…

This mirrors my experience 100%. I'm not even sure why I still pay for OpenAI at this point. Claude 3.5 is just incredibly superior. And I totally agree on the point about dropping in context and asking very specific questions. I've had Claude pinpoint a bug in a 2k LOC module that I was struggling to find the cause for. After wasting a lot of time on it on my own, I thought "what the heck, maybe Claude can figure it out" and it did. It's objectively useful, even if flawed sometimes.

Re: Things we learned about LLMs in 2024

#102

Post by minimaxir »

nkingsy wrote: post #62
pkoird wrote: post #9

I'd love to read a semi-technical book on everything that we've learned about what works and what does not on LLMs.

It would be out of date in months. Things that didn’t work 6 months ago do now. Things that don’t work now, who knows…

There are still some tropes from the GPT-3 days that are fundamental to the construction of LLMs that affect how they can be used and will not change unless they no longer are trained to optimize for next-token-prediction (e.g. hallucinations and the need for prompt engineering)

Re: Things we learned about LLMs in 2024

#103

Post by datadrivenangel »

dxbydt wrote: post #92
antirez wrote: post #54

About "people still thinking LLMs are quite useless", I still believe that the problem is that most people are exposed to ChatGPT 4o that at this point for my use case (programming / design partner) is basically a useless toy. And I guess that in tech many folks try LLMs for the same use cases. Try Claude Sonnet 3.5 (not Haiku!) and tell me if, while still flawed, is not helpful. But there is more: a key thing with L…

> best LLMs are able to accelerate you https://www2.math.upenn.edu/~ghrist/preprints/LAEF.pdf - this math textbook was written in just 55 days! Paraphrasing the acknowledgements - ...Begun November 4, 2024, published December 28, 2024. ...assisted by Claude 3.5 sonnet, trained on my previous books... ...puzzles co-created by the author and Claude ...GPT-4o and -o1 were useful in latex configurations...doing proof-rea…

Going faster isn't good if the quality drops enough that overall productivity decreases... Infinite slop is only a good thing for pigs.

Re: Things we learned about LLMs in 2024

#104

Post by com2kid »

benreesman wrote: post #32
shihab wrote: post #20

Earlier quoted context omitted.

The last OpenAI valuation I read about was 157 billion. I am struggling to understand what justifies this. To me, it feels like OpenAI is at best few months ahead of competitors in some areas. But even if I am underestimating the advantage, it's few years instead of few months, why does it matter? It's not like AI companies are going to enjoy the first-mover advantage internet giants had over the competition.

Us skeptics believe that valuation prices in some form of regulatory capture or other non-market factor. The non-skeptical interpretation is that it's a threshold function, a flat-out race with an unambiguous finish line. If someone actually hit self-improving AGI first there's an argument that no one would ever catch up.

There are some really good books about wars between cultures that have AGI and it always comes down to math - whoever can get their hands on more compute faster wins.

Re: Things we learned about LLMs in 2024

#105

Post by Animats »

datadrivenangel wrote: post #33
shihab wrote: post #20

Earlier quoted context omitted.

The last OpenAI valuation I read about was 157 billion. I am struggling to understand what justifies this. To me, it feels like OpenAI is at best few months ahead of competitors in some areas. But even if I am underestimating the advantage, it's few years instead of few months, why does it matter? It's not like AI companies are going to enjoy the first-mover advantage internet giants had over the competition.

It's justified if AGI is possible. If AGI is possible, then the entire human economy stops making sense as far as money goes, and 'owning' part of OpenAI gives you power. That is of course, assuming AGI is possible and exponential, and that marketshare goes to a single entity instead of a set of entities. Lots of big assumptions. Seems like we're heading towards a slow-lackluster singularity though.

We may not need smarter AI. Just less stupid AI.

The big problem with LLMs is that most of the time they act smart, and some of the time they do really, really dumb things and don't notice. It's not the ceiling that's the problem. It's the floor. Which is why, as the article points out, "agents" aren't very useful yet. You can't trust them to not screw up big-time.

Re: Things we learned about LLMs in 2024

#106

Post by orbital-decay »

Der_Einzige wrote: post #26

RE: Slop: Having Slop generations from an LLM is a choice. There are so many tricks to make models genuinely creative just at the sampler level alone. https://github.com/sam-paech/antislop-sampler https://openreview.net/forum?id=FBkpCyujtS

This won't solve anything. There's a myriad of sampling strategies, and they all have the same issue: samplers are dumb. They have no access to the semantics of what they're sampling. As a result, things like min-p or XTC will either overshoot or undershoot as they can't differentiate between the situations. For the same reason, samplers like DRY can't solve repetition issues.

Slop is over-representation of model's stereotypes and lack of prediction variety in cases that need it. Modern models are insufficiently random when it's required. It's not just specific words or idioms, it's concepts on very different abstraction levels, from words to sentence patterns to entire literary devices. You can't fix issues that appear on the latent level by working with tokens. The antislop link you give seems particularly misguided, trying to solve an NLP task programmatically.

Research like [1] suggests algorithms like PPO as one of the possible culprits in the lack of variety, as they can filter out entire token trajectories. Another possible reason is training on outputs from the previous models and insufficient filtering of web scraping results.

And of course, prediction variety != creativity, although it's certainly a factor. Creativity is an ill-defined term like many in these discussions.

[1] https://arxiv.org/abs/2406.05587

Re: Things we learned about LLMs in 2024

#107

Post by kianN »

dxbydt wrote: post #92
antirez wrote: post #54

About "people still thinking LLMs are quite useless", I still believe that the problem is that most people are exposed to ChatGPT 4o that at this point for my use case (programming / design partner) is basically a useless toy. And I guess that in tech many folks try LLMs for the same use cases. Try Claude Sonnet 3.5 (not Haiku!) and tell me if, while still flawed, is not helpful. But there is more: a key thing with L…

> best LLMs are able to accelerate you https://www2.math.upenn.edu/~ghrist/preprints/LAEF.pdf - this math textbook was written in just 55 days! Paraphrasing the acknowledgements - ...Begun November 4, 2024, published December 28, 2024. ...assisted by Claude 3.5 sonnet, trained on my previous books... ...puzzles co-created by the author and Claude ...GPT-4o and -o1 were useful in latex configurations...doing proof-rea…

^ This perfectly encapsulates the story I see every time someone digs into the details of any llm generated or assisted content that has any level of complexity.

Great on the surface but lacks any depth, cohesive, or substance

Re: Things we learned about LLMs in 2024

#108

Post by philipkglass »

AnimalMuppet wrote: post #94
philipkglass wrote: post #44

Earlier quoted context omitted.

If AGI is possible, then the entire human economy stops making sense as far as money goes, and 'owning' part of OpenAI gives you power. That's if AGI is possible and not easily replicated . If AGI can be copied and/or re-developed like other software then the value of owning OpenAI stock is more like owning stock in copper producers or other commodity sector companies. (It might even be a poorer investment. Even AGI…

The GP said, "and exponential". If AGI is exponential, then the first one will have a head start advantage that compounds over time. That is going to be hard to overcome.

I believe that AGI cannot be exponential for long because any intelligent agent can only approach nature's limits asymptotically. The first company with AGI will be about as much ahead as, say, the first company with electrical generators [1]. A lot of science fiction about a technological singularity assumes that AGI will discover and apply new physics to develop currently-believed-impossible inventions, but I don't consider that plausible myself. I believe that the discovery of new physics will be intellectually satisfying but generally inapplicable to industry, much like how solving the cosmological lithium problem will be career-defining for whoever does it but won't have any application to lithium batteries.

https://en.wikipedia.org/wiki/Cosmological_lithium_problem

[1] https://en.wikipedia.org/wiki/Siemens#1847_to_1901

Re: Things we learned about LLMs in 2024

#109

Post by mhh__ »

fosterfriends wrote: post #5

My fav part of the writeup at the end: """ LLMs need better criticism # A lot of people absolutely hate this stuff. In some of the spaces I hang out (Mastodon, Bluesky, Lobste.rs, even Hacker News on occasion) even suggesting that “LLMs are useful” can be enough to kick off a huge fight. I like people who are skeptical of this stuff. The hype has been deafening for more than two years now, and there are enormous quan…

I can think of some runaway scenario's where LLMs are definitely bad but, indeed, this particular line of criticism is really just luddites longing for a world that probably doesn't exist anymore.

These are the people who regulate and legislate for us, they are the risk-adverse fools who would rather things be nice and harmless lest they be bad but work.

Personally, I think my only serious ideology in this area is that I am fundamentally biased towards the power of human agency. I'd rather not need to, but in a (perhaps) Nietzschean sense I view so-called AI as a force multiplier to totally avoid the above people.

AI will enable the creative to be more concrete, and drag those on the other end of the scale towards the normie mean. This is of great relevance to the developing world too - AI may end up a tool for enforcing western culture upon the rest of the world but perhaps a force decorrelating it from the McKinsey's of tall buildings in big cities.

Re: Things we learned about LLMs in 2024

#110

Post by mhh__ »

vunderba wrote: post #11
fosterfriends wrote: post #5

My fav part of the writeup at the end: """ LLMs need better criticism # A lot of people absolutely hate this stuff. In some of the spaces I hang out (Mastodon, Bluesky, Lobste.rs, even Hacker News on occasion) even suggesting that “LLMs are useful” can be enough to kick off a huge fight. I like people who are skeptical of this stuff. The hype has been deafening for more than two years now, and there are enormous quan…

I agree, but I think my biggest issue with LLMs (and a lot of GenAI) is that they act as a massive accelerator for the WORST (and unfortunately most common) type of human - the lazy one. The signal-to-noise ratio just goes completely out of control. https://journal.everypixel.com/ai-image-statistics

The people who are lazy but have taste will do well, then.
Post reply on HN