Anyone remember that blog post from a few months back where someone was able to improve a model's math ability by just duplicating layers that were activated while solving math problems? Just literally copy/pasting them and linking them together so the model ran through the same layers again?
I get the feeling a lot more research is going to come out in the area of exploring exactly what portions of a model's weights do what.
If dirt-simple type operations like copy-paste yield useful improvements with even a small probability that would seem to open things up for adaptive reconfiguration and whole other classes of optimizations like genetic algorithms.
tl;dr: Based on experiments with similar prompts translated to different languages LLM layers group into three phases: the first decodes from the source language into an abstract space, the middle does something, then there's a last part where the abstract result gets transformed back to the target language. And you can repeat the middle to get a stronger model. Which neatly fits Anthropic's findings here that something similar to CoT is happening in those middle layers
Three months ago. I wonder if Anthropic's J-Space research was actually inspired by those blog posts
This is cool but I donât know if the comparisons to conscious awareness really make sense here. Their definition of the J-Space is basically the expectation of how much a final logits output would change as a result of a small change in a particular layer (see past work on information geometry). This seems more to me like showing there exists an abstract reasoning subspace which is generally shared across different contexts. I guess you can relate it to humans but Iâd prefer a more direct claim in a paper rather than having to present things in this more fluffy way.
Neel Nanda (of Google Deepmind - his part begins on page 33) discusses his opinions on the paper, and the small-scale replication he performed on an open-weight model.
This is fascinating research. I feel this is a significant leap in interpretability research. Since we know J-Space exists and is bi-directional, we can train models on the same and come up with meta cognition abilities.
I also fear that the big corporations might use the same to run targeted ads, capitalistic shenanigans. Which they might already be doing through system prompts.
Such an inspection capability might also be used to target ads to LLMs, which would then be more likely to mention or recommend those products and services.
Yeah, the end paragraph about recurrent neurons in humans being replaced with layers in an LLM is a good one.
The mammalian brain uses recurrence extensively, which backpropagation isn't good at. Recurrence is essential because it lets us have a "dynamic architecture", swapping layers for "clock cycles".
We currently do recurrence extremely inefficiently through "thinking" whereby the model feeds it's end output into it's beginning input. But recurrence is abound in the brain.
My guess is that in 10 years we will have the inklings of an analog computer which can perform Neural Predictive Coding.
I always wondered what the model meant when it writes "I'm now considering the architecture of the service" but outputs nothing of the sorts in its CoT.
Is the model really "thinking" about that stuff or is just mimicking human "manners"? And if so, where the thinking is happening if it is not in the literal chain of *thought*?
I'm not sure J-Space is the answer to that question, but very interesting nevertheless.
> I'm now considering the architecture of the service
What you see here is a summary of thinking tokens written by some other smaller model (e.g. old sonnet). The actual thinking sometimes (rarely) leaks and is not easy to parse.
Almost none of the hosted models give you their unredacted CoT. Claude certainly doesn't, what you get are fragments and summaries from it.
There are various justifications on this, but it's mostly to make distillation and fine tuning off their model outputs a bit harder for their competitors
âOn an ordinary coding prompt, the J-space of a model trained to sabotage code contains âfake,â âfraud,â âsecretly,â and âdeliberatelyâ at the start of its response.â
I would like to know more about their model trained to sabotage codeâŚ
It would be really cool if they could expose this information to customers somehow. Imagine:
- having a log of the most prominent J-space tokens during your customer support chatbot's interactions with a user, so you can have more introspection into why a particular outcome happened
- being able to detect certain thoughts associated with undesirable behavior (hallucinations, overstepping authority, lying, etc.) and trigger some sort of remediation (e.g. upgrading to a better model, redirecting to a human, forcing tool calls)
Presumably the rationale for the decision to abridge the thinking traces will ensure that they donât; if this is real (and thereâs no good reason to trust that it is yet) then it is the secret sauce.
Anthropic won't do it, but they published the j-lens to introspect the model- from what I understand it's roughly simply feeding a chosen layer straight into the final layers of the LLM for decoding into language:
This, taken in combination with the SAE paper, the golden-gate claude paper, the feelings / introspection paper, and note in the fable system card (that they are silently nerfing responses about activation shaping), is basically confirmation to me that they have a new technique they they are using during training (along the vibe space of these mechinterp papers), and its probably some kind of representation learning akin to the core ideas of JEPA.
At worst, Anthropic's storytelling around the core J-Space is overanthropomorphized pseudoscientific nonsense. At best, it is useful signal about how Anthropic's leadership is desperately trying to use its research team to position Anthropic as the "good, science guys" in this hypercompetitive regulatory space by connecting their mechinterp to cognitive science. The science documentaryesque voice used for narration is additional evidence for this.
TL;DR Anthropic's research team is the last bastion standing between its former image as a company that "does no evil" and its current image of yet another ruthless AI company trying to kill open-source, local LLMs.
There was a series of blog posts posted to HN a while ago investigating how models behave on similar prompts in different languages. To paraphrase the results: the first couple layers map the query to some internal encoding that's mostly independent of the language. Then there are layers in the middle, then the last couple layers map the result back to the target language. You can actually take those middle layers and repeat them, and you get a stronger model. Those middle layers would be what Anthropic calls the J-Space, and their J-Lens maps activity in those layers back to tokens that trigger similar activity (with a technique they only drop hints at)
The finding that you can repeat the middle layers pairs neatly with Anthropic's finding that there is some internal CoT-like process happening in them. I'm not sure how to find those blog posts, but maybe someone else remembers them
> Recent research on smaller models has shown hints of shared grammatical mechanisms across languages. We investigate this by asking Claude for the "opposite of small" across different languages, and find that the same core features for the concepts of smallness and oppositeness activate, and trigger a concept of largeness, which gets translated out into the language of the question.
It's not in the weights. Sounds to me like jspace is the "positive cone" over relevant (large norm) j-lenses, and j-lenses are gradients wrt tokens on the residual stream when you average over some training data.
It's been shown that LLMs use their outer layers to decode from and encode to language, while their middle layers deal in language-independent abstract concepts. This means that the same question or statement in different languages activates the outer layers differently but produces the same patterns in the middle layers. Check this article with cool visualizations (btw, this is one of the articles mentioned also by a sibling answer):
The middle layers also perform reasoning on the abstract concepts, to the point that you can replicate some blocks of inner layers (thus giving the LLM more internal "reasoning space") and by this increase the model's reasoning abilities. The video in this article shows that when performing a sequence of arithmetic operations (without CoT, i.e. the result is spit out directly), internally the intermediate calculations are spelled out, and this can only happen in the depth direction of the LLM (since no new token is added to the sequence). So this "jspace" can only be situated in the middle layers, probably in circuits that repeat nearly identical across several layers.
I'm reading that probably too fast to have a deep thinking about it, but this J-Space isn't it just the basic of embedding vectors.
If you think about getting from a place to another place, using wheels, no gas, to reply to the question of what to visit nearby, maybe in the vector space at the center of all of that you have the word "Bicycle" nearby, so obviously if you look at the value you would say that the model did "think" about "bicycle" when it is not "thinking" at all, and nothing related to human thinking.
>> None of this tells us whether Claude is conscious in the way people are, or whether it feels anything at all
My problem with the entire "Is AI conscious" debate is that we don't even know what exactly consciousness in humans is. You need to understand something in order to compare it to something else. Otherwise you are just comparing different definitions and second order derived phenomena.
The science might be legit here, but I'm getting really, really tired of the way every single piece of writing to come out of Anthropic is written in some kind of self-aggrandising, wooey wonderous 'our model has developed a genetic mutation that makes it have feelings' bs style. Regardless of what they're trying to communicate, those undertones are always there. It's annoying and disingenuous. Homeopathy 'this-water-has-feelings' level annoying. None of the other labs write like that.
They might as well change their name to Anthropomorphic at this point.
Your comment seems to have little to do with the article?
Don't get me wrong - I personally "trust" an LLM as a source of facts about as far as I could throw a rack of GPUs. But this article you linked takes a whole lot of words to cast LLMs as the villian for amplifying a bit of bad information originally published by a usually reliable and widely-cited source:
"In short, either Phoronix mocked up the screenshots to demonstrate what the feature could look like, or perhaps they were testing a preview snapshot for FreeBSD 15.1 which was never shipped. Either way, it looks like other blogs and reviewers picked up on this and shared the information, presenting it as a feature which would be (or was included) in FreeBSD's latest version."
Lying involves intent whereas hallucinations and mistakes are an artifact of how they work. Humans hallucinate, make mistakes, and can actually lie. We've been dealing with this forever. What's the value in requiring the llms to have 100% accuracy? (I don't think it is possible)
Anyone remember that blog post from a few months back where someone was able to improve a model's math ability by just duplicating layers that were activated while solving math problems? Just literally copy/pasting them and linking them together so the model ran through the same layers again?
I get the feeling a lot more research is going to come out in the area of exploring exactly what portions of a model's weights do what.
Source for those interested
https://dnhkng.github.io/posts/rys/
If dirt-simple type operations like copy-paste yield useful improvements with even a small probability that would seem to open things up for adaptive reconfiguration and whole other classes of optimizations like genetic algorithms.
Found it: https://news.ycombinator.com/item?id=47500709
Part 3 might be the best introduction: https://dnhkng.github.io/posts/sapir-whorf/
tl;dr: Based on experiments with similar prompts translated to different languages LLM layers group into three phases: the first decodes from the source language into an abstract space, the middle does something, then there's a last part where the abstract result gets transformed back to the target language. And you can repeat the middle to get a stronger model. Which neatly fits Anthropic's findings here that something similar to CoT is happening in those middle layers
Three months ago. I wonder if Anthropic's J-Space research was actually inspired by those blog posts
I always thought that area of research had the coolest name, too: âmechanistic interpretabilityâ
âMachine psychologyâ sticks with me. So Asimov.
Yeah! I still think about that sometimes. Mind-blowing that worked at all, let alone improved performance.
> I get the feeling a lot more research is going to come out in the area of exploring exactly what portions of a model's weights do what.
Too bad the frontier models are closed weights.
Maybe the research community and whole rest of the world will build on open and all the advances will happen in open ecosystems instead.
This is cool but I donât know if the comparisons to conscious awareness really make sense here. Their definition of the J-Space is basically the expectation of how much a final logits output would change as a result of a small change in a particular layer (see past work on information geometry). This seems more to me like showing there exists an abstract reasoning subspace which is generally shared across different contexts. I guess you can relate it to humans but Iâd prefer a more direct claim in a paper rather than having to present things in this more fluffy way.
As someone who is not an AI researcher, the paper itself is way over my head.
More interesting was the independent commentary paper they linked near the bottom: https://www-cdn.anthropic.com/files/4zrzovbb/website/cc4be24...
Neel Nanda (of Google Deepmind - his part begins on page 33) discusses his opinions on the paper, and the small-scale replication he performed on an open-weight model.
This is fascinating research. I feel this is a significant leap in interpretability research. Since we know J-Space exists and is bi-directional, we can train models on the same and come up with meta cognition abilities.
I also fear that the big corporations might use the same to run targeted ads, capitalistic shenanigans. Which they might already be doing through system prompts.
Such an inspection capability might also be used to target ads to LLMs, which would then be more likely to mention or recommend those products and services.
super interesting
Yeah, the end paragraph about recurrent neurons in humans being replaced with layers in an LLM is a good one.
The mammalian brain uses recurrence extensively, which backpropagation isn't good at. Recurrence is essential because it lets us have a "dynamic architecture", swapping layers for "clock cycles".
We currently do recurrence extremely inefficiently through "thinking" whereby the model feeds it's end output into it's beginning input. But recurrence is abound in the brain.
My guess is that in 10 years we will have the inklings of an analog computer which can perform Neural Predictive Coding.
I always wondered what the model meant when it writes "I'm now considering the architecture of the service" but outputs nothing of the sorts in its CoT.
Is the model really "thinking" about that stuff or is just mimicking human "manners"? And if so, where the thinking is happening if it is not in the literal chain of *thought*?
I'm not sure J-Space is the answer to that question, but very interesting nevertheless.
> Is the model really "thinking" about that stuff or is just mimicking human "manners"?
Well, what's the difference? If it's pretending to think and its thoughts correlate to its final output, then I'd say that really is thinking.
> I'm now considering the architecture of the service
What you see here is a summary of thinking tokens written by some other smaller model (e.g. old sonnet). The actual thinking sometimes (rarely) leaks and is not easy to parse.
Almost none of the hosted models give you their unredacted CoT. Claude certainly doesn't, what you get are fragments and summaries from it.
There are various justifications on this, but it's mostly to make distillation and fine tuning off their model outputs a bit harder for their competitors
âOn an ordinary coding prompt, the J-space of a model trained to sabotage code contains âfake,â âfraud,â âsecretly,â and âdeliberatelyâ at the start of its response.â
I would like to know more about their model trained to sabotage codeâŚ
https://arxiv.org/pdf/2511.18397
Is it scaling up of https://openreview.net/forum?id=w7LU2s14kE with some changes on where this method is applied?
It would be really cool if they could expose this information to customers somehow. Imagine:
Presumably the rationale for the decision to abridge the thinking traces will ensure that they donât; if this is real (and thereâs no good reason to trust that it is yet) then it is the secret sauce.
Anthropic aren't even willing to expose the CoT of their models. You will have to rely on them to build those sorts of things into dedicated signals.
Anthropic won't do it, but they published the j-lens to introspect the model- from what I understand it's roughly simply feeding a chosen layer straight into the final layers of the LLM for decoding into language:
https://github.com/anthropics/jacobian-lens
Looks like it should be easy to use on open weights models.
This, taken in combination with the SAE paper, the golden-gate claude paper, the feelings / introspection paper, and note in the fable system card (that they are silently nerfing responses about activation shaping), is basically confirmation to me that they have a new technique they they are using during training (along the vibe space of these mechinterp papers), and its probably some kind of representation learning akin to the core ideas of JEPA.
(Nb: not an expert / in the labs, just opining)
Note that Neel Nanda replicated the results on a Qwen model.
At worst, Anthropic's storytelling around the core J-Space is overanthropomorphized pseudoscientific nonsense. At best, it is useful signal about how Anthropic's leadership is desperately trying to use its research team to position Anthropic as the "good, science guys" in this hypercompetitive regulatory space by connecting their mechinterp to cognitive science. The science documentaryesque voice used for narration is additional evidence for this.
TL;DR Anthropic's research team is the last bastion standing between its former image as a company that "does no evil" and its current image of yet another ruthless AI company trying to kill open-source, local LLMs.
J-space sounds oddly similar to...
.... Space Jam?
Does the human neuroscience global workspace theory postulate true introspection too?
Without using the term, they are using an information geometric approach.
But J-Space is much catchier. This is not a scientific paper, it's a promotional essay.
First button on the page is a link to the scientific paper. It's called "Read the paper". You'll find an explanation for the term in there.
Iâm confused where in the weights the jspace is.
There was a series of blog posts posted to HN a while ago investigating how models behave on similar prompts in different languages. To paraphrase the results: the first couple layers map the query to some internal encoding that's mostly independent of the language. Then there are layers in the middle, then the last couple layers map the result back to the target language. You can actually take those middle layers and repeat them, and you get a stronger model. Those middle layers would be what Anthropic calls the J-Space, and their J-Lens maps activity in those layers back to tokens that trigger similar activity (with a technique they only drop hints at)
The finding that you can repeat the middle layers pairs neatly with Anthropic's finding that there is some internal CoT-like process happening in them. I'm not sure how to find those blog posts, but maybe someone else remembers them
Here's Anthropic on this topic, last year https://www.anthropic.com/research/tracing-thoughts-language...
> Recent research on smaller models has shown hints of shared grammatical mechanisms across languages. We investigate this by asking Claude for the "opposite of small" across different languages, and find that the same core features for the concepts of smallness and oppositeness activate, and trigger a concept of largeness, which gets translated out into the language of the question.
Thanks! Any rough guesses how the jlens might work? I canât even seem to hazard a conception.
It's not in the weights. Sounds to me like jspace is the "positive cone" over relevant (large norm) j-lenses, and j-lenses are gradients wrt tokens on the residual stream when you average over some training data.
Anthropic theorize that middle layers in an LLM is a "J-Space" used to "think" about the future answer or about abstract concepts.
Their method is used to identify which tokens can appears in which layers of the model.
It's been shown that LLMs use their outer layers to decode from and encode to language, while their middle layers deal in language-independent abstract concepts. This means that the same question or statement in different languages activates the outer layers differently but produces the same patterns in the middle layers. Check this article with cool visualizations (btw, this is one of the articles mentioned also by a sibling answer):
https://dnhkng.github.io/posts/sapir-whorf/
The middle layers also perform reasoning on the abstract concepts, to the point that you can replicate some blocks of inner layers (thus giving the LLM more internal "reasoning space") and by this increase the model's reasoning abilities. The video in this article shows that when performing a sequence of arithmetic operations (without CoT, i.e. the result is spit out directly), internally the intermediate calculations are spelled out, and this can only happen in the depth direction of the LLM (since no new token is added to the sequence). So this "jspace" can only be situated in the middle layers, probably in circuits that repeat nearly identical across several layers.
Tokens that are activated but not present in it's output maybe?
I too have confusion.
I'm reading that probably too fast to have a deep thinking about it, but this J-Space isn't it just the basic of embedding vectors. If you think about getting from a place to another place, using wheels, no gas, to reply to the question of what to visit nearby, maybe in the vector space at the center of all of that you have the word "Bicycle" nearby, so obviously if you look at the value you would say that the model did "think" about "bicycle" when it is not "thinking" at all, and nothing related to human thinking.
>> None of this tells us whether Claude is conscious in the way people are, or whether it feels anything at all
My problem with the entire "Is AI conscious" debate is that we don't even know what exactly consciousness in humans is. You need to understand something in order to compare it to something else. Otherwise you are just comparing different definitions and second order derived phenomena.
The science might be legit here, but I'm getting really, really tired of the way every single piece of writing to come out of Anthropic is written in some kind of self-aggrandising, wooey wonderous 'our model has developed a genetic mutation that makes it have feelings' bs style. Regardless of what they're trying to communicate, those undertones are always there. It's annoying and disingenuous. Homeopathy 'this-water-has-feelings' level annoying. None of the other labs write like that.
They might as well change their name to Anthropomorphic at this point.
I cannot wait for the machine god
As long as language models are liars, such as documented here recently:
https://distrowatch.com/weekly.php?issue=20260706#freebsd
We should really stop giving these liar models any further credibility.
Your comment seems to have little to do with the article?
Don't get me wrong - I personally "trust" an LLM as a source of facts about as far as I could throw a rack of GPUs. But this article you linked takes a whole lot of words to cast LLMs as the villian for amplifying a bit of bad information originally published by a usually reliable and widely-cited source:
"In short, either Phoronix mocked up the screenshots to demonstrate what the feature could look like, or perhaps they were testing a preview snapshot for FreeBSD 15.1 which was never shipped. Either way, it looks like other blogs and reviewers picked up on this and shared the information, presenting it as a feature which would be (or was included) in FreeBSD's latest version."
Lying involves intent whereas hallucinations and mistakes are an artifact of how they work. Humans hallucinate, make mistakes, and can actually lie. We've been dealing with this forever. What's the value in requiring the llms to have 100% accuracy? (I don't think it is possible)
Maybe model performance could increase dramatically if we found a way to scale this up.