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

I got the highest score on ARC-AGI again swapping Python for English

jeremyberman.substack.com

Post reply on HN
Print view View original

101–110 of 136 posts

Re: I got the highest score on ARC-AGI again swapping Python for English

#101

Post by modeless »

Workaccount2 wrote: post #83
modeless wrote: post #5

I've been testing LLMs on Sokoban-like puzzles (in the style of ARC-AGI-3) and they are completely awful at them. It really highlights how poor their memory is. They can't remember abstract concepts or rules between steps, even if they discover them themselves. They can only be presented with lossy text descriptions of such things which they have to re-read and re-interpret at every step. LLMs are completely helpless…

I really think that the problem is with tokenizing vision. Any kind of visually based reasoning and they become dumb as rocks. It feels similar to having a person play sokoban but blindfolded and only with text prompts. The same issue cropped up with playing pokemon. Like the image gets translated to text, and then the model works on that. I'm no expert on transformers, but it just feels like there is some kind of li…

Yes, vision is a problem, but I don't think it's the biggest problem for the specific task I'm testing. The memory problem is bigger. The models frequently do come up with the right answer, but they promptly forget it between turns.

Sometimes they forget because the full reasoning trace is not preserved in context (either due to API limitations or simply because the context isn't big enough to hold dozens or hundreds of steps of full reasoning traces). Sometimes it's because retrieval from context is bad for abstract concepts and rules vs. keyword matching, and to me the reason for that is that text is lossy and inefficient. The models need to be able to internally store and retrieve a more compact, abstract, non-verbal representation of facts and procedures.

Re: I got the highest score on ARC-AGI again swapping Python for English

#102

Post by adastra22 »

Workaccount2 wrote: post #99
adastra22 wrote: post #85

Earlier quoted context omitted.

I’m sympathetic to your point, but this isn’t quite fair. The field of psychology does exist.

Neuroscience is the field that would be closest to this. But even they are empty handed with evidence and heavy with hypotheses.

No, psychology is right. Psychology studies what the properties of thought are. Neuroscience studies the specific biochemical mechanisms of the brain. Psychology is the study of what mental reasoning IS, while neuroscience is the study of HOW neurons in our brain implement it.

If you are asking “ok, but what is reasoning, really? What definition of reasoning would enable us to recognize whether it is going on in this AI or not?” it is a question of psychology. Unless we are restricting ourselves to whole brain emulation only.

Re: I got the highest score on ARC-AGI again swapping Python for English

#103

Post by freejazz »

kenjackson wrote: post #97
freejazz wrote: post #80

Earlier quoted context omitted.

This is unhelpfully obtuse

What's obtuse about it? It's honestly a very straightforward statement. Every thing we think or say is a function of past events. We don't incorporate future events into what we think or say. Even speculation or imagination of future events occurred in the past (that is the act of imagining it occurred in the past). It's really a super simple concept -- maybe it's so simple that it seems obtuse.

Because the other poster's point wasn't that it was a 'past event.' The point was that it's just predicting based upon the previous token. It's disingenuous to mix the two concepts up.

Re: I got the highest score on ARC-AGI again swapping Python for English

#104

Post by freejazz »

ACCount37 wrote: post #88
freejazz wrote: post #76

Earlier quoted context omitted.

I haven't seen LLMs perform common sense reasoning. Feel free to share some links. Your post reads like anthropomorphized nonsense.

What? Do you even know what "commonsense reasoning" means?

Do you?

Re: I got the highest score on ARC-AGI again swapping Python for English

#105

Post by schrectacular »

HarHarVeryFunny wrote: post #54
gwd wrote: post #23

Earlier quoted context omitted.

> Pattern matching questions on a contrived test is not the same thing as understanding or reasoning. Do submarines swim? I don't really care if it gets me where I want to go. The fact is that just two days ago, I asked Claude to look at some reasonably complicated concurrent code to which I had added a new feature, and asked it to list what tests needed to be added; and then when I asked GPT-5 to add them, it one-sh…

> Do submarines swim? It doesn't matter HOW LLMs "swim" as long as they can, but the point being raised is whether they actually can. It's as if LLMs can swim in the ocean, in rough surf, but fail to swim in rivers or swimming pools, because they don't have a generalized ability to swim - they've just been RL-trained on the solution steps to swimming in surf, but since those exact conditions don't exist in a river (w…

>> Do submarines swim?

>It doesn't matter HOW LLMs "swim" as long as they can, but the point being raised is whether they actually can.

>It's as if LLMs can swim in the ocean, in rough surf, but fail to swim in rivers or swimming pools

Just like submarines!

Re: I got the highest score on ARC-AGI again swapping Python for English

#107

Post by Workaccount2 »

modeless wrote: post #101
Workaccount2 wrote: post #83

Earlier quoted context omitted.

I really think that the problem is with tokenizing vision. Any kind of visually based reasoning and they become dumb as rocks. It feels similar to having a person play sokoban but blindfolded and only with text prompts. The same issue cropped up with playing pokemon. Like the image gets translated to text, and then the model works on that. I'm no expert on transformers, but it just feels like there is some kind of li…

Yes, vision is a problem, but I don't think it's the biggest problem for the specific task I'm testing. The memory problem is bigger. The models frequently do come up with the right answer, but they promptly forget it between turns. Sometimes they forget because the full reasoning trace is not preserved in context (either due to API limitations or simply because the context isn't big enough to hold dozens or hundreds…

I think the problem is though that they need to store it in text context.

When I am solving a sokoban style game, it's entirely visual. I don't need to remember a lot because the visual holds so much information.

It's like the average person trying to play a game of chess with just text. It's nightmarishly hard compared to having a board in front of you. The LLMs seem stuck having to play everything through just text.

Re: I got the highest score on ARC-AGI again swapping Python for English

#108

Post by MobiusHorizons »

schrectacular wrote: post #105
HarHarVeryFunny wrote: post #54

Earlier quoted context omitted.

> Do submarines swim? It doesn't matter HOW LLMs "swim" as long as they can, but the point being raised is whether they actually can. It's as if LLMs can swim in the ocean, in rough surf, but fail to swim in rivers or swimming pools, because they don't have a generalized ability to swim - they've just been RL-trained on the solution steps to swimming in surf, but since those exact conditions don't exist in a river (w…

>> Do submarines swim? >It doesn't matter HOW LLMs "swim" as long as they can, but the point being raised is whether they actually can. >It's as if LLMs can swim in the ocean, in rough surf, but fail to swim in rivers or swimming pools Just like submarines!

What? Submarines can definitely “swim” in rivers, although shallow water is certainly more challenging for a submerged vessel. Most submarines are a bit big for most swimming pools, but small ones like ROVs are frequently tested in pools.

Re: I got the highest score on ARC-AGI again swapping Python for English

#109

Post by DenisM »

mjr00 wrote: post #68
naasking wrote: post #60

Earlier quoted context omitted.

> are they really learning to reason, or are they just learning to pattern match to steer generation in the direction of problem-specific reasoning steps that they had been trained on? Are you sure there's a real difference? Do you have a definition of "reasoning" that excludes this?

It's trivial to demonstrate that LLMs are pattern matching rather than reasoning. A good way is to provide modified riddles-that-aren't. As an example: > Prompt: A man working at some white collar job gets an interview scheduled with an MBA candidate. The man says "I can't interview this candidate, he's my son." How is this possible? > ChatGPT: Because the interviewer is the candidate’s mother. (The riddle plays on t…

We kinda move from the situation “LLM can only do what it seen before” to “LLM can do something by composing several things it has seen before”. We didn’t get to the situation “LLM can do things it has not seen before”.

The practicality of the situation is that a lot of problems fall into the second bucket. We all like to think we deal with novel problems, but most of what we can think of was already considered by another human and captured by llm. You had to invent something deliberately unique, and that’s telling. Most startup ideas are invented more than once, for example.

The key shortcoming of the llm is that it is not aware of its own limits. If it ever becomes aware it can outsource such rare things to mechanical Turk.

Re: I got the highest score on ARC-AGI again swapping Python for English

#110

Post by modeless »

Workaccount2 wrote: post #107
modeless wrote: post #101

Earlier quoted context omitted.

Yes, vision is a problem, but I don't think it's the biggest problem for the specific task I'm testing. The memory problem is bigger. The models frequently do come up with the right answer, but they promptly forget it between turns. Sometimes they forget because the full reasoning trace is not preserved in context (either due to API limitations or simply because the context isn't big enough to hold dozens or hundreds…

I think the problem is though that they need to store it in text context. When I am solving a sokoban style game, it's entirely visual. I don't need to remember a lot because the visual holds so much information. It's like the average person trying to play a game of chess with just text. It's nightmarishly hard compared to having a board in front of you. The LLMs seem stuck having to play everything through just text…

It's not just visual. You also need a representation of the rules of the game and the strategies that make sense. The puzzles I'm solving are not straight Sokoban, they have per-game varying rules that need to be discovered (again, ARC-AGI-3 style) that affect the strategies that you need to use. For example, in classic Sokoban you can't push two crates at once, but in some of the puzzles I'm using you can, and this is taught by forcing you to do it in the first level, and you need to remember it through the rest of the levels. This is not a purely visual concept and models still struggle with it.
Post reply on HN