Would be interesting to train a cutting edge model with a cut off date of say 1900 and then prompt it about QM and relativity with some added context.
If the model comes up with anything even remotely correct it would be quite a strong evidence that LLMs are a path to something bigger if not then I think it is time to go back to the drawing board.
Looking at the training data I don't think it will know anything.[0]
Doubt On the Connexion of the Physical Sciences (1834) is going to have much about QM. While the cut-off is 1900, it seems much of the texts a much closer to 1800 than 1900.
You would find things in there that were already close to QM and relativity. The Michelson-Morley experiment was 1887 and Lorentz transformations came along in 1889. The photoelectric effect (which Einstein explained in terms of photons in 1905) was also discovered in 1887. William Clifford (who _died_ in 1889) had notions that foreshadowed general relativity: "Riemann, and more specifically Clifford, conjectured that forces and matter might be local irregularities in the curvature of space, and in this they were strikingly prophetic, though for their pains they were dismissed at the time as visionaries." - Banesh Hoffmann (1973)
Things don't happen all of a sudden, and being able to see all the scientific papers of the era its possible those could have fallen out of the synthesis.
Yann LeCun spoke explicitly on this idea recently and he asserts definitively that the LLM would not be able to add anything useful in that scenario. My understanding is that other AI researchers generally agree with him, and that it's mostly the hype beasts like Altman that think there is some "magic" in the weights that is actually intelligent. Their payday depends on it, so it is understandable. My opinion is that LeCun is probably correct.
Chemistry would be a great space to explore. The last quarter of the 19th century had a ton of advancements in chemistry. It'd be interesting the see if an LLM could propose fruitful hypotheses, made predictions of the science of thermodynamics.
A rigorous approach to predicting the future of text was proposed by Li et al 2024, "Evaluating Large Language Models for Generalization and Robustness via Data Compression" (https://ar5iv.labs.arxiv.org/html//2402.00861) and I think that work should get more recognition.
They measure compression (perplexity) on future Wikipedia, news articles, code, arXiv papers, and multi-modal data. Data compression is intimately connected with robustness and generalization.
I would love to ask such a model to summarise the handful of theories or theoretical “roads” being eyed at the time and to make a prediction with reasons as to which looks most promising. We might learn something about blind spots in human reasoning, institutions, and organisations that are applicable today in the “future”.
That would be an interesting experiment. It might be more useful to make a model with a cut off close to when copyrights expire to be as modern as possible.
Then, we have a model that knows quite a bit in modern English. We also legally have a data set for everything it knows. Then, there's all kinds of experimentation or copyright-safe training strategies we can do.
Project Gutenberg up to the 1920's seems to be the safest bet on that.
I think it would be fun to see if an LLM would reframe some scientific terms from the time in a way that would actually fit in our current theories.
I imagine if you explained quantum field theory to a 19th century scientists they might think of it as a more refined understanding of luminiferous aether.
Or if an 18th century scholar learned about positive and negative ions, it could be seen as an expansion/correction of phlogiston theory.
I'm trying to work towards that goal by training a model on mostly German science texts up to 1904 (before the world wars German was the lingua franca of most sciences).
Training data for a base model isn't that hard to come by, even though you have to OCR most of it yourself because the publicly available OCRed versions are commonly unusably bad. But training a model large enough to be useful is a major issue. Training a 700M parameter model at home is very doable (and is what this TimeCapsuleLLM is), but to get that kind of reasoning you need something closer to a 70B model. Also a lot of the "smarts" of a model gets injected in fine tuning and RL, but any of the available fine tuning datasets would obviously contaminate the model with 2026 knowledge.
I think it would raise some interesting questions, but if it did yield anything noteworthy, the biggest question would be why that LLM is capable of pioneering scientific advancements and none of the modern ones are.
The development of QM was so closely connected to experiments that it's highly unlikely, even despite some of the experiments having been performed prior to 1900.
I like this, it would be exciting (and scary) if it deduced QM, and informative if it cannot.
But I also think we can do this with normal LLMs trained on up-to-date text, by asking them to come up with any novel theory that fits the facts. It does not have to be a groundbreaking theory like QM, just original and not (yet) proven wrong ?
I like this idea. I think I'd like it more if we didn't have to prompt the LLM in the first place. If it just had all of this information and decided to act upon it. That's what the great minds of history (and even average minds like myself) do. Just think about the facts in our point of view and spontaneously reason something greater out of them.
>.If the model comes up with anything even remotely correct it would be quite a strong evidence that LLMs are a path to something bigger if not then I think it is time to go back to the drawing board.
In principle I see your point, in practice my default assumption until proven otherwise here -- is that a little something slipped through post-1900.
A much easier approach would be to just download some model, whatever model, today. Then 5 years from now, whatever interesting discoveries are found - can the model get there.
We've thought of doing this sort of exercise at work but mostly hit the wall of data becoming a lot more scare the further back in time we go. Particularly high quality science data - even going pre 1970 (and that's already a stretch) you lose a lot of information. There's a triple whammy of data still existing, being accessible in any format, and that format being suitable for training an LLM. Then there's the complications of wanting additional model capabilities that won't leak data causally.
It's already been done, without the model being aware of it, see https://arxiv.org/abs/2512.09742. They also made it think it was Hitler (not MechaHitler, the other guy), and other craziness.
It's a relief to think that we're not trusting these things for stuff like financial advice, medical advice, mental health counselling, ...
Yes, I don't understand how such an experiment could work. You either:
A). contaminate the model with your own knowledge of relativity, leading it on to "discover" what you know, or
B). you will try to simulate a blind operation but without the "competent human physicist knowledgeable up to the the 1900 scientific frontier" component prompting the LLM, because no such person is alive today nor can you simulate them (if you could, then by definition you can use that simulated Einstein to discover relativity, so the problem is moot).
So in both cases you would prove nothing about what a smart and knowledgeable scientist can achieve today from a frontier LLM.
Could be an interesting experiment, but its not conclusive proof one way or another. So much of what makes LLMs so great today (vs gpt 3.5) would not be in that dataset. The training to turn these models into coding savants has generalized to other areas just as one example.
It's a base model. It hasn't been instruction tuned to "solve problems" necessarily. All it can do is attempt to complete text given some starting text.
I wonder about something similar - leeches. Apparently they really work to eat away only decaying tissue. Could they make a comeback in modern times with better weighting?
Anyone seen a low-friction way to run prompts through this yet, either via a hosted API or chat UI or a convenient GGML or MLX build that runs in Ollama or llama.cpp or LM Studio?
Currently running it using LM Studio. It can download it from Hugging Face. It generates incoherent text though
===
You:
I pray you, who is this Master Newton?
timecapsulellm-v2-1800-1875-mlx:
TI offer to pay you the very same fee as you did before.
It was not in the power of your master to deliver the letter to your master. He did. I will be with you as soon as I can keep my word.
It is not at all clear, whether the letter has been sent or not.
It is not at all clear: but it is clear also that it was written by the person who gave it.
"No," I said, "I cannot give it to you."
There, the letter was sent to me.
"The letter is yours, I believe," I said.
"But, I hope, you will not refuse to give it to me?
It would be interesting if there's enough data to train a model capable enough to converse with and ask about contemporary views on issues of the day, or what it thought about "potential" future events/technologies yet to happen.
Could this be an experiment to show how likely LLMs are to lead to AGI, or at least intelligence well beyond our current level?
If you could only give it texts and info and concepts up to Year X, well before Discovery Y, could we then see if it could prompt its way to that discovery?
Oh I have really been thinking long about this. The intelligence that we have in these models represent a time.
Now if I train a foundation models with docs from library of Alexandria and only those texts of that period, I would have a chance to get a rudimentary insight on what the world was like at that time.
This kind of technique seems like a good way to test model performance against benchmarks. I'm too skeptical that new models are taking popular benchmark solutions into their training data. So-- how does e.g. ChatGPT's underlying architecture perform on SWE-bench if trained only on data prior to 2024.
Charles DarwinECCEMACY. Sir, — The following case is interesting to me : — I was in London a fortnight, and was much affected with an attack of rheumatism. The first attack of rheumatism was a week before I saw you, and the second when I saw you, and the third when I saw you, and the third in the same time. The second attack of gout, however, was not accompanied by any febrile symptoms, but was accompanied with an increased flow of urine, and with a more copious discharge of urine. The third attack was an hour after I saw you, and was succeeded by a return of a paroxysm of gout, and a more rapid return of the gout. The fourth attack was also accompanied by a fever, but was not always accompanied by any febrile symptoms. The third attack of gout was a fortnight after you had been ill, and the fourth was followed by a paroxysm of gout. The fourth attack was a fortnight after you were attacked, and was accompanied by a sense
I found it particularly thought-inspiring how a model with training from that time period completely lacks context/understanding of what it is itself, but then I realized that we are the same (at least for now).
Can you confidently say that the architure of the LLM doesn't include any a priori bias that might effect the integrity of this LLM?
That is, the architectures of today are chosen to yield the best results given the textual data around today and the problems we want to solve today.
I'd argue that this lack of bias would need to be researched (if it hasn't been already) before this kind of model has credence.
LLMs aren't my area of expertise but during my PhD we were able to encode a lot of a priori knowledge through the design of neural network architectures.
I'd love to see an LLM equivalent, but I don't think that's enough data to train from scratch. Could a LoRA or similar be used in a way to get speech style to strictly follow a few megabytes worth of training data?
Fascinating idea. There was another "time-locked" LLM project that popped up on HN recently[1]. Their model output is really polished but the team is trying to figure out how to avoid abuse and misrepresentation of their goals. We think it would be cool to talk to someone from 100+ years ago but haven't seriously considered the many ways in which it would be uncool. Interesting times!
Suppose two models with similar parameters trained the same way on 1800-1875 and 1800-2025 data. Running both models, we get probability distributions across tokens, let's call the distributions 1875' and 2025'. We also get a probability distribution finite difference (2025' - 1875'). What would we get if we sampled from 1.1*(2025' - 1875') + 1875'? I don't think this would actually be a decent approximation of 2040', but it would be a fun experiment to see. (Interpolation rather than extrapolation seems just as unlikely to be useful and less likely to be amusing, but what do I know.)
If the output of this is even somewhat coherent, it would disprove the argument that mass amounts of copyrighted works are required to train an LLM. Unfortunately that does not appear to be the case here.
Mm. I'm a bit sceptical of the historical expertise of someone who thinks that "Who art Henry" is 19th century language. (It's not actually grammatically correct English from any century whatever: "art" is the second person singular, so this is like saying "who are Henry?")
89 comments
[ 3.3 ms ] story [ 72.6 ms ] threadIf the model comes up with anything even remotely correct it would be quite a strong evidence that LLMs are a path to something bigger if not then I think it is time to go back to the drawing board.
[0] https://github.com/haykgrigo3/TimeCapsuleLLM/blob/main/Copy%...
At least the model part. Although others made same thought as you afaik none tried it.
Things don't happen all of a sudden, and being able to see all the scientific papers of the era its possible those could have fallen out of the synthesis.
They measure compression (perplexity) on future Wikipedia, news articles, code, arXiv papers, and multi-modal data. Data compression is intimately connected with robustness and generalization.
Then, we have a model that knows quite a bit in modern English. We also legally have a data set for everything it knows. Then, there's all kinds of experimentation or copyright-safe training strategies we can do.
Project Gutenberg up to the 1920's seems to be the safest bet on that.
I imagine if you explained quantum field theory to a 19th century scientists they might think of it as a more refined understanding of luminiferous aether.
Or if an 18th century scholar learned about positive and negative ions, it could be seen as an expansion/correction of phlogiston theory.
Training data for a base model isn't that hard to come by, even though you have to OCR most of it yourself because the publicly available OCRed versions are commonly unusably bad. But training a model large enough to be useful is a major issue. Training a 700M parameter model at home is very doable (and is what this TimeCapsuleLLM is), but to get that kind of reasoning you need something closer to a 70B model. Also a lot of the "smarts" of a model gets injected in fine tuning and RL, but any of the available fine tuning datasets would obviously contaminate the model with 2026 knowledge.
Special relativity however seems possible.
But I also think we can do this with normal LLMs trained on up-to-date text, by asking them to come up with any novel theory that fits the facts. It does not have to be a groundbreaking theory like QM, just original and not (yet) proven wrong ?
In principle I see your point, in practice my default assumption until proven otherwise here -- is that a little something slipped through post-1900.
A much easier approach would be to just download some model, whatever model, today. Then 5 years from now, whatever interesting discoveries are found - can the model get there.
1900 era humans are not available to do this so I'm not sure how this experiment is supposed to work.
It's a relief to think that we're not trusting these things for stuff like financial advice, medical advice, mental health counselling, ...
A). contaminate the model with your own knowledge of relativity, leading it on to "discover" what you know, or
B). you will try to simulate a blind operation but without the "competent human physicist knowledgeable up to the the 1900 scientific frontier" component prompting the LLM, because no such person is alive today nor can you simulate them (if you could, then by definition you can use that simulated Einstein to discover relativity, so the problem is moot).
So in both cases you would prove nothing about what a smart and knowledgeable scientist can achieve today from a frontier LLM.
===
You:
I pray you, who is this Master Newton?
timecapsulellm-v2-1800-1875-mlx:
TI offer to pay you the very same fee as you did before. It was not in the power of your master to deliver the letter to your master. He did. I will be with you as soon as I can keep my word. It is not at all clear, whether the letter has been sent or not. It is not at all clear: but it is clear also that it was written by the person who gave it. "No," I said, "I cannot give it to you." There, the letter was sent to me. "The letter is yours, I believe," I said. "But, I hope, you will not refuse to give it to me?
If you could only give it texts and info and concepts up to Year X, well before Discovery Y, could we then see if it could prompt its way to that discovery?
Now if I train a foundation models with docs from library of Alexandria and only those texts of that period, I would have a chance to get a rudimentary insight on what the world was like at that time.
And maybe time shift further more.
PROMPT:Charles Darwin
Charles DarwinECCEMACY. Sir, — The following case is interesting to me : — I was in London a fortnight, and was much affected with an attack of rheumatism. The first attack of rheumatism was a week before I saw you, and the second when I saw you, and the third when I saw you, and the third in the same time. The second attack of gout, however, was not accompanied by any febrile symptoms, but was accompanied with an increased flow of urine, and with a more copious discharge of urine. The third attack was an hour after I saw you, and was succeeded by a return of a paroxysm of gout, and a more rapid return of the gout. The fourth attack was also accompanied by a fever, but was not always accompanied by any febrile symptoms. The third attack of gout was a fortnight after you had been ill, and the fourth was followed by a paroxysm of gout. The fourth attack was a fortnight after you were attacked, and was accompanied by a sense
https://news.ycombinator.com/item?id=46319826
I found it particularly thought-inspiring how a model with training from that time period completely lacks context/understanding of what it is itself, but then I realized that we are the same (at least for now).
That is, the architectures of today are chosen to yield the best results given the textual data around today and the problems we want to solve today.
I'd argue that this lack of bias would need to be researched (if it hasn't been already) before this kind of model has credence.
LLMs aren't my area of expertise but during my PhD we were able to encode a lot of a priori knowledge through the design of neural network architectures.
Back in the pre-LLM days, someone trained a Markov chain off the King James Bible and a programming book: https://www.tumblr.com/kingjamesprogramming
I'd love to see an LLM equivalent, but I don't think that's enough data to train from scratch. Could a LoRA or similar be used in a way to get speech style to strictly follow a few megabytes worth of training data?
[1] https://news.ycombinator.com/item?id=46319826
But on various data (i.e., separate model per source): the Bible, Don Quixote and Franz Kafka. (As well as a (bad!) lyrics generator, and translator.)