Rendered at 17:55:29 GMT+0000 (Coordinated Universal Time) with Cloudflare Workers.
evilmathkid 7 hours ago [-]
Hi! Author here. Surprised to see this on HN now. Happy to answer any questions!
Some context about this:
- This is NOT an LLM. its a small ar transformer trained from scratch. One of the points was that extremely complex problems can be tackled without LLMs
- Till the v1 of this result, this benchmark was only scaled by LLMs or their finetunes (ofc w enormous training costs). Other attempts performed okayish but used v complex architectures or extremely high amounts of training compute. No one expected a simple AR transformer to perform this well, at this low cost and w these few training samples.
- Sample Efficiency is one of the most important unsolved problems today in AI. That's what I was targetting with this work. We know it is easy to increase SE by increasing compute/params, so it was important to constrain cost as much as possible (also why OpenAI's Parameter Golf had fixed compute and why Modded NanoGPT is considered very sample efficient)
- Can the perf be improved? Yes but the competition is ongoing so can't talk about it
- Personally I think today's frontier models can be beat by training from scratch. Haven't proved this yet tho
- Fun: I was new to ML when I posted this first (dec '25). I basically used ARC as a way to learn ML
nullbio 53 minutes ago [-]
"- Personally I think today's frontier models can be beat by training from scratch. Haven't proved this yet tho"
Also, in curating the training data in a deliberate manner, with attention to detail. Most people just use existing datasets and call it a day. It's a lot of work, which is why there are gains on the table.
pj_mukh 4 hours ago [-]
>> NOT an LLM. its a small ar transforme
Super cool project! Though, aren't most modern LLM's ar transformers internally?
dpoloncsak 1 hours ago [-]
Pretty sure it's one of those "All squares are rectangles but not all rectangles are squares" situations.
Transformers are what really started the LLM Boom, and seem to be crucial to the technology. They also have other applications, such as what OP created
dfdydx 4 hours ago [-]
Not all transformers are _language_ models - the sequences of tokens don't have to be sequences of words.
shsshs 38 minutes ago [-]
Words are just symbols. We interpret it as language. The distinction is meaningless IMO.
He uses the mechanism that makes LLMs clever but he doesn’t use “words” but some other symbols. Transformers can do images and videos and whatever, they are not tied to to language.
Forgeties79 18 minutes ago [-]
This is incredibly pedantic if you ask me.
perfmode 3 hours ago [-]
In this case, what are the tokens?
evilmathkid 2 hours ago [-]
9 color tokens + 4 special tokens (start, end, newline, inp_out_sep)
dnautics 4 hours ago [-]
It's an slm
bkaae 3 hours ago [-]
There is no language in the training of this, so there is no l.
HNDevsSuck2 3 hours ago [-]
[dead]
_superposition_ 54 minutes ago [-]
I've been interested in training a transformer from scratch for the same learning reasons. The GPU cost/availability seemed prohibitive to do anything useful but you seem to have flipped that on its head. I love your outside the box approach.
kuczmama 6 hours ago [-]
Hey thanks for sharing this. Was curious did you find the more you trained the model the more perf improved, or did it start plateauing. For example, let's say you didn't spend 67 cents, but you spent 67 dollars do you think you would get major benefits from that?
evilmathkid 6 hours ago [-]
Yeah I've reached much higher perf but
- it feels logarithmic (like most perf-compute graphs), and eventually plateaus. 44% @ 67 cents was a good stopping point for me
- more compute would require a lot of effort and dealing with new problems like training stability, cost of iterations/sweeps (didnt have the money to convincingly run larger iterations)
bbor 3 hours ago [-]
First: this is really great technical writing, especially when you get into the rebuttals. Firm & clear without polemics -- props, and thanks for open-sourcing!
That said; I don't have the time, energy, or anywhere near the expertise to challenge you on the DL specifics, but I feel compelled to add another voice to the chorus of doubters nonetheless.
Using other ARC examples at runtime (effectively, yes?) for "transduction" may not violate what some officer behind ARC said on Twitter --and is certainly a fantastic tool for certain problem spaces-- but it just seems like a glaring and unavoidable philosophical problem in this one. My issue isn't with using the eval set per-se (though that obviously sets off well-justified alarm bells), but rather building an AGI system whose performance relies on the arbitrary size and shape of this particular dataset.
There's a lot of ways to frame this, but given the transduction citations the most appropriate is probably the AI winter's infamous 'Frame Problem':
You say upfront that this works in the first place because ARC has "very few samples... in a high dimensional space"; to me, that seems like an extremely strong indicator that the datasets are not intended to capture anywhere near the full semantic space that we would consider relevant for AGI. If true, your approach would indeed be ""cheating"" by using an arbitrary & unavoidable feature of the dataset (that they didn't have the time or money to craft 100 million high quality cases by hand instead of 1000) to solve the frame problem upfront for you. This would explain why you don't even need a full LLM here -- that's the unsolvable problem that LLMs solve for us.
That is... even if the ARC train+eval sets contain the sum of human intuition between them, superficial differences IRL would render your model unable to identify which examples are relevant to which problems, and thus unable to transductively reason.
In plainer English: surely you'd agree that your model would do worse if we swapped it out with Opus behind the scenes than the next-highest-scoring ARC model would do in the same position, yes? For coding, research, dumb questions, SVG pelicans -- the lot?
If so, that seems like hard proof that this scores high on a benchmark at the cost of the benchmark itself. Like, if this transductive approach leads to ARC1 being claimed (which I thought it was ages ago but :shrug:), they'll either have to abandon the whole benchmark or ban this approach retroactively.
If not... well, I guess I encourage you to try it! It seems like you'd need 1000 truly stellar hand-picked examples to transductively cover that whole space, for one thing.
sigbottle 2 hours ago [-]
Kind of hijacking, would you say that LLM's have solved the frame problem?
To me, the frame problem is: Can you function in an open vs closed world, and to me the answer is yes, LLM's can definitely function in an open world where the rules are fuzzy, changing, undefined, etc. At the very least, much better than all GOFAI approaches by far.
The issue is now grounding - It can "function", but what would it take to "ground" them? A personality, maybe? Actual consequences? Making them interact only with constrained tools that are formally verified?
Right now it's a combination of harness engineering, and ml philosophers arguing about compression leading to the "objectively correct intelligence", whatever that means.
I think LLM's are "A[x]I" right now in the sense of "they have the capability to integrate with everything" - but obviously you can argue how much this actually reflects "A[x]I" (if you gave someone integration with everything, is that really your success or people handing you it)? But they are still missing some oomph factors that need to be clarified IMO. Maybe it's something as "mundane" as just having actual persistent memory, or maybe it's some deep philosophical thing like qualia. Who knows.
Planktonne 2 hours ago [-]
One of these is a much weaker claim than the other.
> yes, LLM's can definitely function in an open world where the rules are fuzzy, changing, undefined, etc.
> At the very least, much better than all GOFAI approaches by far
sigbottle 1 hours ago [-]
That's true. Again, I currently view LLMs as a function of integration - what they may lack in "intrinsic smarts", whatever that means, they can tool call and we build capacities (and they build capacities!) around them and to some degree can reason and be creative.
I do think the first claim has real merit even if it's not 100% on par with humans. Second claim is just true.
gandreani 4 hours ago [-]
Fyi, the link to rhabdomyolysis is broken on the homepage! The URL is repeated
evilhackerdude 1 hours ago [-]
nothing like legendary shrugging and keeping the mind open
asabla 6 hours ago [-]
Thank you for answering these questions. Looking forward for the next write up about this.
1 hours ago [-]
lai7th_0 4 hours ago [-]
[dead]
foota 11 minutes ago [-]
> I agree that its rare to see to face problem sets in real life where every problem is given at once. Even if it is (like an exam), humans can usually only attempt one at a time
Just one small snippet that I thought was interesting. I would always read through ~the entire exam before starting. Both so that I could find the problems most approachable to me, but also because sometimes it helps me figure out the rest of the questions :-)
foota 8 minutes ago [-]
> Ban offline training/pretraining. Models must train from scratch after submission
Previously this was considered impossible so rule. My model shows this is possible
Guarantees no synthetic data can be used
It makes the comparison fair across differet models. Otherwise some models like LLMs can benchmaxx ARC by using ungodly amounts of offline training. (Since the benchmark has been around a long time, many ARC-like datasets have been created)
I'm not an ML researcher, so YMMV, but... how could a model learn to answer these ARC-AGI questions without training beforehand?
lackoftactics 2 hours ago [-]
Sounds like a good day to be you, top 5 on Kaggle with a publication like this. It seems like you will be on a plane to SF shortly
bee_rider 4 hours ago [-]
I think(?) you’ve already probably done a good job of explaining this criticism for semi-informed people. But can you dumb it down even more for those of us who are almost entirely out-of-the-loop?
> Training on the eval puzzles is cheating / “training on test”
> No this is false. “Training on test” specifically means training on the labels of test data. The labels were not trained on.
> Also, ARC is a metalearning benchmark, so you’re supposed to learn from the eval puzzles.
> Jargon: ARC has a set of train puzzles and a set of eval puzzles. Each puzzle has example pairs and test pairs. A pair consists of an input grid + output grid.
> The ARC, the label is only the test pair’s output grid in an eval puzzle.
> These labels were not trained on. They are hidden. You can delete it beforehand if you wish
I think what I gather here is that the test comes with one batch of training problems, which everyone agrees you can train on. But maybe the eval problems also come with input/output examples (to help define the problem) and training on those is controversial? I can’t see why it would be controversial but is that the criticism?
jrflo 3 hours ago [-]
The point of ARC is essentially an "IQ Test" for AI systems. It is meant to cover abstract reasoning capabilities of generally-intelligent systems like LLMs. What the author did here was build a system that only solves ARC problems.
The other tension is the fact that this score is on the public eval set. In machine learning, you typically have 3 datasets: training, evaluation, and test. The training set is the dataset that's used to update the weights according to your loss function, you are "encoding" the patterns from the training set directly into your model. The eval set is what you use to track performance while training, it is NOT used to update model weights, but shows how well the model generalizes. The test set is a private holdout set that is only used when you're "done" developing your model. The difference between test and eval is information leakage: you can use performance against the eval set to modify your hyperparameters and model architecture to get better eval scores. So while the eval set doesn't directly update the weights, it can indirectly cause "overfitting" by tailoring your model to do well on the eval set. What you really want to see is the private test set performance, not the eval set. For all we know, this model could be ridiculously overfit on the eval set and perform poorly on the private test set.
tptacek 43 minutes ago [-]
Instantly one of my favorite comments this year. Nicely done.
bee_rider 3 hours ago [-]
It seems like an interesting strategy. Based on the author’s comment, they haven’t been at it for very long. So, I guess the folks who run the private test haven’t had a chance to get to it? It’d be interesting to hear how it does.
evilmathkid 3 hours ago [-]
I'm currently 10th in the world on the private set on Kaggle. And iirc, at one point I was 4th
Can't comment more since its an ongoing competition
porridgeraisin 3 hours ago [-]
They have _not_ trained on the test set.
On the private test set, the right way to evaluate this type of model, is giving i it the test question Q, which it will first train to AR predict first, and then it will inference using the just-updated weights with Q as prompt, giving you back A, and then you compare A with A_true secretly.
jrflo 3 hours ago [-]
I never said they did
evilmathkid 3 hours ago [-]
What you gather is correct, assuming by "the test" you mean the ARC benchmark in general. It was controversial because people are used to LLMs which are frozen at train time, where the eval problems are usually not trained on for various reasons like fragility (basically porridgeraisin's ans which is great)
Here's another explanation. Take the train dataset and test dataset of a benchmark
Train: {x_i -> f(x_i)}, Test: {x_j -> f(x_j)}
As long as f(x_j) in the test set is hidden, there is no "training on test". In a normal benchmark, each x_i is a single datapoint. But in metalearning benchmarks like ARC, x_i is the puzzle itself that has a train set and the test questions within it, hence the confusion and controversy
porridgeraisin 3 hours ago [-]
Basically, you have a bunch of Q,A pairs in the training dataset. Here, it was trained to next-word predict the question itself, as well as next-word predict the answer given the question as prompt. This is bog-standard, no one's complaining.
In the test dataset's Q,A pairs, it was only trained to next-word predict the question itself, and it was not given the answer at all.
It was then evaluated by seeing if it is able to output A_test given the Q_test as prompt.
What would be cheating is training it to produce A_test (given Q_test as prompt) as well, since then you can always make a model that scores 100% by just memorising Q_test, A_test pairs.
The complaints online mostly stem from not reading that properly and assuming they trained on Q_test,A_test instead of just Q_test. This is further because these days large LLMs are inadvertently trained on many benchmark solutions even unintentionally due to the massive scale of data and the infeasibility of auditing it all. But none of that is the case here.
The reason you want to train on Q_test is because in these AR transformer models, they learn useful composable encodings of Q by simply learning to next-word predict Q. So you enable the model to learn composable encodings of the test questions, so that it can hopefully "connect it" to an earlier train problem it had seen, and adapt the solution it had seen for that, much like humans do in school exams.
Without this step, you are making it difficult for the model to "connect" the test question to a train question it had seen earlier, and then it still has to adapt the solution. This way, you precompute that "this test question is like this train question" and then during the exam you only have to do the adapting the solution part after a simpler "retrieval" process.
You can just think of next-word training Q_test as a "retrieval" process.
This practice often used in continual learning or "test time training" is not yet useful in general real world ML tasks due to the differences in memory and compute requirements, and more so the general fragility of training large neural networks in a streaming realtime way (as opposed to large data, batched), versus inferencing from a static neural network. It is due to that fragility that I believe (correct me if I am wrong) this guy had to train on a batch of Q_tests. If you enforced that you will not provide Q_test_2 before they answer Q_test_1, the performance will drop.
While the increased compute and memory is difficult to solve inherently, there are various efforts being made to fix the fragility, especially in reinforcement learning where this is called "streaming RL", there is revival of interest as seen in RLC 2026.
[Note]
Arc-AGI-1 doesn't have any actual english words or such, but it's simpler to pretend it was a basic Q&A benchmark to explain the above
bee_rider 3 hours ago [-]
Further question—the model produces an answer to the question, it sends the answer, and then gets graded. Does it get to know immediately how it did, or does it get the grade back at the end after answering all the questions?
If it is the former case, it would be possible to add the generated question/answer pair into the training set as well. Would that be considered fair? (Of course this is a moot point if the answers all get graded simultaneously at the end). Then the model could explore interesting strategies around what order to answer questions in.
In my uninformed opinion, the various permutations of question ordering/answer revealing all map to different real-world scenarios… and any of them could be interesting!
evilmathkid 2 hours ago [-]
Nope, it never learns how it did on the questions.
During test time, you have to submit all the answers at once and you get the total score (so you dont even know which puzzles were solved)
porridgeraisin 57 minutes ago [-]
It does not, if it gets the answer (or any information about them, even % of qns solved) and is able to adjust itself in response, then that is considered training on the test set and is wrong.
howunfortunate 3 hours ago [-]
I won't weigh in on whether it's "cheating" but it is definitely benchmaxxing
iwontberude 11 minutes ago [-]
[dead]
xeonax 7 hours ago [-]
Even cooler is his about me mention of saving his own life https://mvakde.github.io/ > Saved myself in a medical emergency (doctors didn't know what rhabdomyolysis was)
qlm 7 hours ago [-]
Crazy, considering rhabdo isn't that rare.
cannonpalms 4 hours ago [-]
This was in India, where he describes the medical knowledge of providers as subpar at best.
somenameforme 3 hours ago [-]
Yeah but rhabdo is something literally any e.g. body builder, power lifter, etc could tell you about. Actually if somebody knows what hypertrophy is, they probably know what rhabdo is. It's a pretty normal and big concern in any sort of high intensity weight training.
I can't think of many ways that otherwise healthy and fit younger people can physically nearly kill themselves doing normal activity, so it kind of stands out - let alone it being not all that rare either. Rhabdo has even gone viral in the news like when a while back a couple of Chinese girls nearly killed themselves doing a social media 'squat challenge.' They did 1000, got rhabdo, didn't know what was happening, ended up in the ICU with kidney damage.
It also manifests in other ways too. For instance I had an elderly family member give himself rhabdo during a manic phase he was going through when he started going wild on construction and other physical tasks that were way beyond what his body was ready for.
Basically it's not some super obscure thing you'd expect only a good doctor, let alone a specialist, to know about.
Melatonic 2 hours ago [-]
People commonly knowing about Rhabdo is a much newer thing. I never heard people talk regularly about it all before CrossFit become popular
evilmathkid 1 hours ago [-]
there's a very large variance in doctors' abilities in India. At the very top they are close to the best in the world, esp with an insanely high workload.
but on an average, not great
Also, a lot of gymgoers and physical trainers I know hadn't heard of rhabdo either (and this is a relatively wealthy part of a tier 1 city)
things are changing for the better however
imdsm 6 hours ago [-]
Everyone and their dog who is on statins knows what rhabdo is. Bonkers!
allannienhuis 3 hours ago [-]
I've been on statins for years, and I don't remember anyone talking to me about rhabdo. To be fair the education I received about my medications was a firehose of information after a heart attack and major heart surgery, so perhaps it's possible I missed a few things.
rikkert 5 hours ago [-]
Not true, I am on statins and did not know.
p-e-w 7 hours ago [-]
“What do you call a medical student who graduated at the bottom of their class?”
“Doctor.”
echelon 5 hours ago [-]
I studied biochem in undergrad and my classes were full of premed students.
I loved the subject and nerded out about the course material - I spent my time designing my own experiments around gene cloning that took several semesters to run. They were sharing last year's tests with their frat buddies and laughing at us nerds.
I've never looked at doctors the same way again after college. I looked up to them as a child, yet after seeing how the sausages were made, I started to doubt everything.
I frequently ask doctors, who spend all of ten minutes with me while the nurses do all the work, about the molecular specifics of what they're talking about. They talk down to me as if they're explaining to a child, yet they're frequently quite wrong. I'm not trying to sound superior to them, but I'm shocked they seem to care so little about the subject. It doesn't give me much hope about what they know and their abilities or competency.
I suspect surgeons and specialists are a different breed and aren't like this at all.
And to be clear, this isn't everyone. But it does seem to be the majority I've interacted with throughout my life.
When they act disgruntled at patient interaction, I detest that their profession tries to cap the number of med students per year. We should be letting in as many med students as we can take. We should let doctors from overseas immigrate and easily become practicing doctors here in the US. We should provide easy paths for nurses to become doctors.
The premed students in my university were chiefly concerned about money and prestige. They drove BMWs gifted to them by their parents and laughed at what I drove and how hard I studied. I had to put up with their bullying for years. I know not everyone who studies to become a doctor is like that, but it permanently skewed my view of their profession.
qsera 5 hours ago [-]
>When we should let doctors from overseas immigrate and easily become practicing doctors here in the US.
Lol, what does this has to do with anything regarding aptitude or curiosity!?
Omniusaspirer 4 hours ago [-]
To the parent posters credit, he's very honest that he developed a personal complex against doctors when he was a poor student. He just sees it as a way to lash out at American doctors, the irony being that foreign medical graduates leaving their families and communities to practice in America largely do so because they are exceptionally money motivated. That doesn't make them bad doctors, but there's certainly less likelihood they're doing it purely for the love of medicine or a desire to care for their communities.
I do agree with the sentiment though that the US needs to fund more residency slots as it's an asinine professional barrier, and that we would benefit from more physicians coming from more diverse financial backgrounds.
throw234259 4 hours ago [-]
> certainly less likelihood they're doing it purely for the love of medicine or a desire to care for their communities.
Maybe self preservation?
In east europe, public hospitals will force doctors to work 36 hours shifts (overnight ER with theoretical sleep). Doctors have a full criminal liability for mall practise.
draw_down 6 hours ago [-]
Well, I guess it would be nice if the graduation cutoff were above the level of “knows what rhabdo is”
guluarte 3 hours ago [-]
I think is more concerning doctors didnt know about rhabdomyolysis...
rappatic 56 minutes ago [-]
I don’t have any kind of ML background but I have always thought of sample efficiency as the great unsolved problem of AI. We humans have unbelievably good sample efficiency; often we can durably learn something on just a single example or two. This is the main area in which LLMs are vastly, vastly behind us.
10xDev 25 minutes ago [-]
No, the unsolved problem of AI is continuous learning. We never stop learning, we don't have a "training phase". You are always updating your world model even when you sleep. Also more quality training data does lead to greater learning efficiency as you have more priors to work with.
xdavidliu 47 minutes ago [-]
the caveat is that we are not learning those small number of samples from scratch, since we're coming in with a large amount of training already, much of it from before we were even born
kvn95ss 7 hours ago [-]
> Also, I’m not sure whether “general reasoning” even exists in the first place? Maybe humans are specialised too
I have been wondering the same. We are now exposed to so many stimuli, we are tricked into thinking this is the norm - to have a reasonable understanding about everything, unless specialization is called for.
dhruv3006 4 hours ago [-]
> The mentioned approach is fundamentally flawed, since the inputs are used during pretraining constituting to a leakage, a universally recognized flaw of ML training.
I saw this on the community note for the last blog you wrote - anything to do here.
docheinestages 4 hours ago [-]
Isn't this cheating? Or rather, are frontier agents only looking at one question at a time? If I understand correctly, you're looking at all the examples of the exam questions. If the exam was adjusted so that you can only look at one question at a time, you won't get 44% anymore.
nullbio 32 minutes ago [-]
Why would that be cheating? That's what humans do when they learn, they look for the signals and patterns that reduce the possible set of answers so they can converge on the solution and narrow the search space.
pwmglenn 7 hours ago [-]
Really impressive and creative research. I wonder if the leading labs do anything similar with their models? It doenst look like the open source labs do?
6_7 2 hours ago [-]
I'm so sorry, in how many?
maz1b 4 hours ago [-]
Just wanted to say that the rhabdo part mentioned on your site was really impressive! Speaking as a medical doctor and full stack engineer myself.
evilmathkid 3 hours ago [-]
thanks! was incredibly scary when it happened
kriro 5 hours ago [-]
Thanks for motivating me to work a bit on non-LLM things again :)
hn45e7pbij 4 hours ago [-]
Nice to see arc-agi-1 framed this way — I'd been circling the same idea without the right words.
There's an updated ARC-AGI-1 chart with 5.6 Luna in each thinking level in this video from last week:
"A New Architecture [..] | MOONSHOTS " https://youtube.com/watch?v=qQfUbo7Ldc0&t=2m5s
evilmathkid 6 hours ago [-]
its not gonna do well on ARC-3 without some significant changes and effort
The new arch in that video is kinda misleading. Didn't really compare against proper baselines
eis 7 hours ago [-]
> Increases in LLM scores are now mainly driven by post training (evidence in next section) and are probably a function of amount of synthetic data. They are learning to solve ARC tasks, not learn general abstract reasoning
Agreed and that's for any benchmark. Private tests are better but you still have to trust the provider to not log and use them for training.
That's why I like when a new set of tests like a new ARC-AGI version is published, that's where you can see which of the models abstracted to more general capabilities instead of being focused on the previous tasks. Most models completely fail new ARC-AGI tests.
The "67 cents" part though is misleading imho. You can't extrapolate from there and think that investing say $100 will get you a lot better results. You hit a ceiling very fast and investing into more compute will give you diminishing results. So yes, you can train a custom model to do somewhat decently on a specific set of tasks but then what?
bkaae 7 hours ago [-]
Then nothing - that's awesome. People think that LLMs are the know-all do-all solution to every problem now.
Putting solutions in terms of cents is a great way to potentially win over some ai boosters imo. There are other ways to solve hard problems.
evilmathkid 6 hours ago [-]
> The "67 cents" part though is misleading imho. You can't extrapolate from there and think that investing say $100 will get you a lot better results.
I think thats unfair. Perf-compute is often logarithmic and will always saturate . Reaching the plateau faster is valuable as it often leads to better peaks (held true here and also look at modded nanogpt)
And more compute increases the perf (after dealing with other scaling problems)
GovPulse 5 hours ago [-]
[flagged]
larodi 7 hours ago [-]
"I don’t understand why others didn’t figure this out"
- how about we allot the possibility that so many of presumed ML experts don't have any clue what they be doing, and are eventually API bitches, nothing more.
woah 30 minutes ago [-]
Could be that many presumed ML experts don't even know how to train on the evaluation set
embedding-shape 7 hours ago [-]
Is the author only running their model against one benchmark? I don't think anyone finds that difficult to achieve, the difficulty comes when you want to make the model not benchmaxxed to a specific benchmark, and generalize so it can solve problems not part of the training data, but seems this model is specifically for not this? How useful is that?
If you just wanted to pass these specific tasks in this specific benchmark, and wanted to do so cheaply, I'm sure a non-LLM-based approach would yield better results for even cheaper, since what the author's model does, seem to basically be "solve ARC puzzles", not a general LLM or "coding" LLM.
bkaae 7 hours ago [-]
I read this as a response to the current hype around LLMs. He is showing computers can solve these issues, without using an LLM architecture. A lot of people have sort of forgot that machine learning is more than just LLMs these days.
I found it to be a very interesting angle.
bonplan23 3 hours ago [-]
Just to be clear: It was well known that you can reach such scores with small models and without an LLM if you train on the task. The author highlights those models himself - e.g. HRM/TRM.
The novelty is more that it works with such a plain transformer and low compute price.
embedding-shape 7 hours ago [-]
> He is showing computers can solve these issues, without using an LLM architecture.
Isn't it a LLM he's building though? My very point is that this particular use case could be solved better without building a LLM, now you claim he is not? The description of what he's doing surely makes it sound like it's a (very small) LLM, and personally I'm still on the "if it quacks like a duck" train in life.
> A lot of people have sort of forgot that machine learning is more than just LLMs these days.
Yeah, which I guess if you make my previous comment more concise, is exactly what I state too.
dakolli 7 hours ago [-]
Nowhere does he say he built an llm. Hes using a transformer, not an llm.
embedding-shape 7 hours ago [-]
> Nowhere does he say he built an llm. Hes using a transformer, not an llm.
Please describe what in your mind a "LLM" is exactly, then describe what this person is building. To me this sounds like "He's not building a calculator, he's just building a program that can do addition, minus, multiplication and division and display the results".
Obviously it's not a Large Language Model, but to me this looks more like a LLM than not, given the architecture he's chosen. But again, maybe I misunderstand?
evilmathkid 6 hours ago [-]
Its not an LLM if there's no pretraining. AR transformers were around before LLMs and will be there after LLMs.
When I made this, the point was to show that you dont need pretraining (which is what makes an LLM) to perform well on complex tasks
And yes it is not a language model either. I did not train it on any language data. Only ARC puzzles
soVeryTired 6 hours ago [-]
Out of interest, would you call BERT an LLM? It’s pre trained but not particularly large.
stymaar 6 hours ago [-]
AFAIK, the “large” qualifier came when transformers allowed to scale the size of language models compared to the recurrent models that where in fashion before. And although BERT isn't large by today's standard, it was large enough for the time.
evilmathkid 6 hours ago [-]
idk the definition is fuzzy. thats why people use the "modern" qualifier to talk about decoder-only style and this is also not clean since you now have reasoning models which are separate
soVeryTired 6 hours ago [-]
It’s neither large nor language-based. ARC-AGI-1 is grid-based and nonverbal.
Use of a transformer is not necessary or sufficient to qualify as an LLM.
crotonix 5 hours ago [-]
Transformer solves a Seq2Seq problem just like RNNs. All Seq2Seq problems need not involve a language. In this case teaching on ARC puzzles doesn't mean what he trained is now trained on a language which will be English(or any other language) in this case. So, does his training successfully models "English as a language" -> No. This implies it is not "Large" and has not modeled any "language".
yorwba 6 hours ago [-]
A LLM should at the very least be a language model, i.e. be able to take human-readable text as input or produce it as output. Transformers are used for plenty of tasks that don't involve language, for example object detection or blind source separation, where the models aren't called LMs; and on the other hand there are some LLM architectures that exclusively use linear attention variants and aren't really transformers anymore.
f311a 7 hours ago [-]
The whole point of his model is to optimize for a very specific benchmark.
BUT, he does not use labels when training, so the model does not know the answers.
embedding-shape 7 hours ago [-]
> The whole point of his model is to optimize for a very specific benchmark.
But benchmaxxing is what we generally try to avoid for training, as there is no point really for it. We used to call it "overfitting", now you're saying this person does it intentionally? Why?
K0balt 7 hours ago [-]
There are plenty of applications where a machine learning system needs to optimize for a very limited data set that is still intractable by linear logic systems of reasonable scale and complexity. It’s interesting, because he is using the legos of LLMs to build highly specialized machine learning systems, which is a very pragmatic approach. Obviously a lot of other ways to achieve similar goals, but it’s cool to see someone back porting the modern tools towards older style optimizations.
Also, the complexity of the task he is using occupies an interesting middle ground of ultra high dimensionality (for a “simple” problem) while being limited in width to a narrow set of solves- a space where one would be tempted to imagine you would need a much more capable system.
Some context about this:
- This is NOT an LLM. its a small ar transformer trained from scratch. One of the points was that extremely complex problems can be tackled without LLMs
- Till the v1 of this result, this benchmark was only scaled by LLMs or their finetunes (ofc w enormous training costs). Other attempts performed okayish but used v complex architectures or extremely high amounts of training compute. No one expected a simple AR transformer to perform this well, at this low cost and w these few training samples.
- Sample Efficiency is one of the most important unsolved problems today in AI. That's what I was targetting with this work. We know it is easy to increase SE by increasing compute/params, so it was important to constrain cost as much as possible (also why OpenAI's Parameter Golf had fixed compute and why Modded NanoGPT is considered very sample efficient)
- Can the perf be improved? Yes but the competition is ongoing so can't talk about it
- Personally I think today's frontier models can be beat by training from scratch. Haven't proved this yet tho
- Fun: I was new to ML when I posted this first (dec '25). I basically used ARC as a way to learn ML
Also, in curating the training data in a deliberate manner, with attention to detail. Most people just use existing datasets and call it a day. It's a lot of work, which is why there are gains on the table.
Super cool project! Though, aren't most modern LLM's ar transformers internally?
He uses the mechanism that makes LLMs clever but he doesn’t use “words” but some other symbols. Transformers can do images and videos and whatever, they are not tied to to language.
- it feels logarithmic (like most perf-compute graphs), and eventually plateaus. 44% @ 67 cents was a good stopping point for me
- more compute would require a lot of effort and dealing with new problems like training stability, cost of iterations/sweeps (didnt have the money to convincingly run larger iterations)
That said; I don't have the time, energy, or anywhere near the expertise to challenge you on the DL specifics, but I feel compelled to add another voice to the chorus of doubters nonetheless. Using other ARC examples at runtime (effectively, yes?) for "transduction" may not violate what some officer behind ARC said on Twitter --and is certainly a fantastic tool for certain problem spaces-- but it just seems like a glaring and unavoidable philosophical problem in this one. My issue isn't with using the eval set per-se (though that obviously sets off well-justified alarm bells), but rather building an AGI system whose performance relies on the arbitrary size and shape of this particular dataset.
There's a lot of ways to frame this, but given the transduction citations the most appropriate is probably the AI winter's infamous 'Frame Problem':
You say upfront that this works in the first place because ARC has "very few samples... in a high dimensional space"; to me, that seems like an extremely strong indicator that the datasets are not intended to capture anywhere near the full semantic space that we would consider relevant for AGI. If true, your approach would indeed be ""cheating"" by using an arbitrary & unavoidable feature of the dataset (that they didn't have the time or money to craft 100 million high quality cases by hand instead of 1000) to solve the frame problem upfront for you. This would explain why you don't even need a full LLM here -- that's the unsolvable problem that LLMs solve for us.
That is... even if the ARC train+eval sets contain the sum of human intuition between them, superficial differences IRL would render your model unable to identify which examples are relevant to which problems, and thus unable to transductively reason.
In plainer English: surely you'd agree that your model would do worse if we swapped it out with Opus behind the scenes than the next-highest-scoring ARC model would do in the same position, yes? For coding, research, dumb questions, SVG pelicans -- the lot?
If so, that seems like hard proof that this scores high on a benchmark at the cost of the benchmark itself. Like, if this transductive approach leads to ARC1 being claimed (which I thought it was ages ago but :shrug:), they'll either have to abandon the whole benchmark or ban this approach retroactively.
If not... well, I guess I encourage you to try it! It seems like you'd need 1000 truly stellar hand-picked examples to transductively cover that whole space, for one thing.
To me, the frame problem is: Can you function in an open vs closed world, and to me the answer is yes, LLM's can definitely function in an open world where the rules are fuzzy, changing, undefined, etc. At the very least, much better than all GOFAI approaches by far.
The issue is now grounding - It can "function", but what would it take to "ground" them? A personality, maybe? Actual consequences? Making them interact only with constrained tools that are formally verified?
Right now it's a combination of harness engineering, and ml philosophers arguing about compression leading to the "objectively correct intelligence", whatever that means.
I think LLM's are "A[x]I" right now in the sense of "they have the capability to integrate with everything" - but obviously you can argue how much this actually reflects "A[x]I" (if you gave someone integration with everything, is that really your success or people handing you it)? But they are still missing some oomph factors that need to be clarified IMO. Maybe it's something as "mundane" as just having actual persistent memory, or maybe it's some deep philosophical thing like qualia. Who knows.
> yes, LLM's can definitely function in an open world where the rules are fuzzy, changing, undefined, etc.
> At the very least, much better than all GOFAI approaches by far
I do think the first claim has real merit even if it's not 100% on par with humans. Second claim is just true.
Just one small snippet that I thought was interesting. I would always read through ~the entire exam before starting. Both so that I could find the problems most approachable to me, but also because sometimes it helps me figure out the rest of the questions :-)
I'm not an ML researcher, so YMMV, but... how could a model learn to answer these ARC-AGI questions without training beforehand?
> Training on the eval puzzles is cheating / “training on test”
> No this is false. “Training on test” specifically means training on the labels of test data. The labels were not trained on.
> Also, ARC is a metalearning benchmark, so you’re supposed to learn from the eval puzzles.
> Jargon: ARC has a set of train puzzles and a set of eval puzzles. Each puzzle has example pairs and test pairs. A pair consists of an input grid + output grid.
> The ARC, the label is only the test pair’s output grid in an eval puzzle.
> These labels were not trained on. They are hidden. You can delete it beforehand if you wish
I think what I gather here is that the test comes with one batch of training problems, which everyone agrees you can train on. But maybe the eval problems also come with input/output examples (to help define the problem) and training on those is controversial? I can’t see why it would be controversial but is that the criticism?
The other tension is the fact that this score is on the public eval set. In machine learning, you typically have 3 datasets: training, evaluation, and test. The training set is the dataset that's used to update the weights according to your loss function, you are "encoding" the patterns from the training set directly into your model. The eval set is what you use to track performance while training, it is NOT used to update model weights, but shows how well the model generalizes. The test set is a private holdout set that is only used when you're "done" developing your model. The difference between test and eval is information leakage: you can use performance against the eval set to modify your hyperparameters and model architecture to get better eval scores. So while the eval set doesn't directly update the weights, it can indirectly cause "overfitting" by tailoring your model to do well on the eval set. What you really want to see is the private test set performance, not the eval set. For all we know, this model could be ridiculously overfit on the eval set and perform poorly on the private test set.
Can't comment more since its an ongoing competition
On the private test set, the right way to evaluate this type of model, is giving i it the test question Q, which it will first train to AR predict first, and then it will inference using the just-updated weights with Q as prompt, giving you back A, and then you compare A with A_true secretly.
Here's another explanation. Take the train dataset and test dataset of a benchmark
Train: {x_i -> f(x_i)}, Test: {x_j -> f(x_j)}
As long as f(x_j) in the test set is hidden, there is no "training on test". In a normal benchmark, each x_i is a single datapoint. But in metalearning benchmarks like ARC, x_i is the puzzle itself that has a train set and the test questions within it, hence the confusion and controversy
In the test dataset's Q,A pairs, it was only trained to next-word predict the question itself, and it was not given the answer at all.
It was then evaluated by seeing if it is able to output A_test given the Q_test as prompt.
What would be cheating is training it to produce A_test (given Q_test as prompt) as well, since then you can always make a model that scores 100% by just memorising Q_test, A_test pairs.
The complaints online mostly stem from not reading that properly and assuming they trained on Q_test,A_test instead of just Q_test. This is further because these days large LLMs are inadvertently trained on many benchmark solutions even unintentionally due to the massive scale of data and the infeasibility of auditing it all. But none of that is the case here.
The reason you want to train on Q_test is because in these AR transformer models, they learn useful composable encodings of Q by simply learning to next-word predict Q. So you enable the model to learn composable encodings of the test questions, so that it can hopefully "connect it" to an earlier train problem it had seen, and adapt the solution it had seen for that, much like humans do in school exams.
Without this step, you are making it difficult for the model to "connect" the test question to a train question it had seen earlier, and then it still has to adapt the solution. This way, you precompute that "this test question is like this train question" and then during the exam you only have to do the adapting the solution part after a simpler "retrieval" process.
You can just think of next-word training Q_test as a "retrieval" process.
This practice often used in continual learning or "test time training" is not yet useful in general real world ML tasks due to the differences in memory and compute requirements, and more so the general fragility of training large neural networks in a streaming realtime way (as opposed to large data, batched), versus inferencing from a static neural network. It is due to that fragility that I believe (correct me if I am wrong) this guy had to train on a batch of Q_tests. If you enforced that you will not provide Q_test_2 before they answer Q_test_1, the performance will drop.
While the increased compute and memory is difficult to solve inherently, there are various efforts being made to fix the fragility, especially in reinforcement learning where this is called "streaming RL", there is revival of interest as seen in RLC 2026.
[Note] Arc-AGI-1 doesn't have any actual english words or such, but it's simpler to pretend it was a basic Q&A benchmark to explain the above
If it is the former case, it would be possible to add the generated question/answer pair into the training set as well. Would that be considered fair? (Of course this is a moot point if the answers all get graded simultaneously at the end). Then the model could explore interesting strategies around what order to answer questions in.
In my uninformed opinion, the various permutations of question ordering/answer revealing all map to different real-world scenarios… and any of them could be interesting!
During test time, you have to submit all the answers at once and you get the total score (so you dont even know which puzzles were solved)
I can't think of many ways that otherwise healthy and fit younger people can physically nearly kill themselves doing normal activity, so it kind of stands out - let alone it being not all that rare either. Rhabdo has even gone viral in the news like when a while back a couple of Chinese girls nearly killed themselves doing a social media 'squat challenge.' They did 1000, got rhabdo, didn't know what was happening, ended up in the ICU with kidney damage.
It also manifests in other ways too. For instance I had an elderly family member give himself rhabdo during a manic phase he was going through when he started going wild on construction and other physical tasks that were way beyond what his body was ready for.
Basically it's not some super obscure thing you'd expect only a good doctor, let alone a specialist, to know about.
but on an average, not great
Also, a lot of gymgoers and physical trainers I know hadn't heard of rhabdo either (and this is a relatively wealthy part of a tier 1 city)
things are changing for the better however
“Doctor.”
I loved the subject and nerded out about the course material - I spent my time designing my own experiments around gene cloning that took several semesters to run. They were sharing last year's tests with their frat buddies and laughing at us nerds.
I've never looked at doctors the same way again after college. I looked up to them as a child, yet after seeing how the sausages were made, I started to doubt everything.
I frequently ask doctors, who spend all of ten minutes with me while the nurses do all the work, about the molecular specifics of what they're talking about. They talk down to me as if they're explaining to a child, yet they're frequently quite wrong. I'm not trying to sound superior to them, but I'm shocked they seem to care so little about the subject. It doesn't give me much hope about what they know and their abilities or competency.
I suspect surgeons and specialists are a different breed and aren't like this at all.
And to be clear, this isn't everyone. But it does seem to be the majority I've interacted with throughout my life.
When they act disgruntled at patient interaction, I detest that their profession tries to cap the number of med students per year. We should be letting in as many med students as we can take. We should let doctors from overseas immigrate and easily become practicing doctors here in the US. We should provide easy paths for nurses to become doctors.
The premed students in my university were chiefly concerned about money and prestige. They drove BMWs gifted to them by their parents and laughed at what I drove and how hard I studied. I had to put up with their bullying for years. I know not everyone who studies to become a doctor is like that, but it permanently skewed my view of their profession.
Lol, what does this has to do with anything regarding aptitude or curiosity!?
I do agree with the sentiment though that the US needs to fund more residency slots as it's an asinine professional barrier, and that we would benefit from more physicians coming from more diverse financial backgrounds.
Maybe self preservation?
In east europe, public hospitals will force doctors to work 36 hours shifts (overnight ER with theoretical sleep). Doctors have a full criminal liability for mall practise.
I have been wondering the same. We are now exposed to so many stimuli, we are tricked into thinking this is the norm - to have a reasonable understanding about everything, unless specialization is called for.
I saw this on the community note for the last blog you wrote - anything to do here.
There was this a few weeks ago:
"Schema Harness Achieves ~99% on Arc‑AGI‑3 Public" https://news.ycombinator.com/item?id=48938163
>> Schema, the harness we introduce today, reaches 99% on the ARC_AGI_3 Public set using Claude Opus 4.8 and Fable 5, and 95.35% using GPT‑5.6 Sol
What does that do with 5.6 Luna instead of the expensive models?
What of 'schema' would improve the performance of mdlARC?
mdlARC: https://github.com/mvakde/mdlARC
There's an updated ARC-AGI-1 chart with 5.6 Luna in each thinking level in this video from last week: "A New Architecture [..] | MOONSHOTS " https://youtube.com/watch?v=qQfUbo7Ldc0&t=2m5s
The new arch in that video is kinda misleading. Didn't really compare against proper baselines
Agreed and that's for any benchmark. Private tests are better but you still have to trust the provider to not log and use them for training.
That's why I like when a new set of tests like a new ARC-AGI version is published, that's where you can see which of the models abstracted to more general capabilities instead of being focused on the previous tasks. Most models completely fail new ARC-AGI tests.
The "67 cents" part though is misleading imho. You can't extrapolate from there and think that investing say $100 will get you a lot better results. You hit a ceiling very fast and investing into more compute will give you diminishing results. So yes, you can train a custom model to do somewhat decently on a specific set of tasks but then what?
Putting solutions in terms of cents is a great way to potentially win over some ai boosters imo. There are other ways to solve hard problems.
I think thats unfair. Perf-compute is often logarithmic and will always saturate . Reaching the plateau faster is valuable as it often leads to better peaks (held true here and also look at modded nanogpt)
And more compute increases the perf (after dealing with other scaling problems)
- how about we allot the possibility that so many of presumed ML experts don't have any clue what they be doing, and are eventually API bitches, nothing more.
If you just wanted to pass these specific tasks in this specific benchmark, and wanted to do so cheaply, I'm sure a non-LLM-based approach would yield better results for even cheaper, since what the author's model does, seem to basically be "solve ARC puzzles", not a general LLM or "coding" LLM.
I found it to be a very interesting angle.
The novelty is more that it works with such a plain transformer and low compute price.
Isn't it a LLM he's building though? My very point is that this particular use case could be solved better without building a LLM, now you claim he is not? The description of what he's doing surely makes it sound like it's a (very small) LLM, and personally I'm still on the "if it quacks like a duck" train in life.
> A lot of people have sort of forgot that machine learning is more than just LLMs these days.
Yeah, which I guess if you make my previous comment more concise, is exactly what I state too.
Please describe what in your mind a "LLM" is exactly, then describe what this person is building. To me this sounds like "He's not building a calculator, he's just building a program that can do addition, minus, multiplication and division and display the results".
Obviously it's not a Large Language Model, but to me this looks more like a LLM than not, given the architecture he's chosen. But again, maybe I misunderstand?
When I made this, the point was to show that you dont need pretraining (which is what makes an LLM) to perform well on complex tasks
And yes it is not a language model either. I did not train it on any language data. Only ARC puzzles
Use of a transformer is not necessary or sufficient to qualify as an LLM.
BUT, he does not use labels when training, so the model does not know the answers.
But benchmaxxing is what we generally try to avoid for training, as there is no point really for it. We used to call it "overfitting", now you're saying this person does it intentionally? Why?
Also, the complexity of the task he is using occupies an interesting middle ground of ultra high dimensionality (for a “simple” problem) while being limited in width to a narrow set of solves- a space where one would be tempted to imagine you would need a much more capable system.
I would not call this overfitting, it's finetuning for specific task where you have a benchmark.