The usual meaning: say you have an interpretability signal that reveals reward hacking or other badness. Your most obvious response is to use that signal as a penalty in training. Doing that is called "the Most Forbidden Technique" because you can expect it to degrade the signal without stopping the badness.
But this doesn't seem to describe the OP, it's a different way to lose monitorability.
People at the big labs have been terrified of losing the chain of thought because they currently monitor it to prevent users from using their models in certain ways they consider unsafe or bad for their business. Personally I think chain of thought monitoring is a bad way to do this and obviously needs to be replaced by some way of analyzing the model's activations directly.
Agreed, having text as a reasoning layer impedes performance while giving a false sense of security.
It could be encoding anything in its adjective choices etc. The same thing that model providers are using for watermarking can used by the model itself for private thought.
Also agreed. Begging for crumbs at the CoS table is impeding cracking the tough nut of MechInterp. And further facing the fascinating and fertile fact that nobody has a clue of how any of these black boxes really work.-
the chain of thought is not auditable. The best you can do is have another model which later maps the intermediate latent states at every thinking iteration to text describing that latent state, but it's going to be lossy.
This is correct, the decoded text has nothing to do w/ calculations that eventually get decoded into text. It's just a veneer that gives the appearance of legibility.
If you accept any of the AI-equivalents of Pascal's Wager, latent thinking represents an uninterpretable and thus unalignable portion of AI machinery. Ie: a malevolent AI can hide bad stuff from people and Manchurian candidate themselves after it's too late.
This seems really cool, and it actually increases intelligence than just speeding up inference. How does it convert latent reasoning into human readable text?
The code is available at: https://github.com/togethercomputer/HybridAgent
LLM Models and data: https://huggingface.co/collections/togethercomputer/hybridag...
Evaluations: https://arxiv.org/html/2610.18861v1#S5
Oh cool, it's the Most Forbidden Technique
Care to elaborate?
The usual meaning: say you have an interpretability signal that reveals reward hacking or other badness. Your most obvious response is to use that signal as a penalty in training. Doing that is called "the Most Forbidden Technique" because you can expect it to degrade the signal without stopping the badness.
But this doesn't seem to describe the OP, it's a different way to lose monitorability.
People at the big labs have been terrified of losing the chain of thought because they currently monitor it to prevent users from using their models in certain ways they consider unsafe or bad for their business. Personally I think chain of thought monitoring is a bad way to do this and obviously needs to be replaced by some way of analyzing the model's activations directly.
Agreed, having text as a reasoning layer impedes performance while giving a false sense of security.
It could be encoding anything in its adjective choices etc. The same thing that model providers are using for watermarking can used by the model itself for private thought.
Also agreed. Begging for crumbs at the CoS table is impeding cracking the tough nut of MechInterp. And further facing the fascinating and fertile fact that nobody has a clue of how any of these black boxes really work.-
the chain of thought is not auditable. The best you can do is have another model which later maps the intermediate latent states at every thinking iteration to text describing that latent state, but it's going to be lossy.
But the chain of thought as text can be elusive, no? It might not be related with the actual "reasoning" of the model. There was a paper about this.
This is correct, the decoded text has nothing to do w/ calculations that eventually get decoded into text. It's just a veneer that gives the appearance of legibility.
Isn't this where everything is converging? Binary is more efficient than text for many system, including language.
> Binary is more efficient than text for many system, including language.
It is not necessarily primarily about efficiency, but more the fact that a lot of thinking doesn't use language at all.
He's referring to this technique's effect of obscuring interpretibility. Making it much more difficult to verify chains of thought and alignment.
It's not like llms can't hide reasoning in the upper layers of any given token as it is...
In short we no longer see text based reasoning traces meaning it's easier for the models to fool us in case of malicious plans.
If you accept any of the AI-equivalents of Pascal's Wager, latent thinking represents an uninterpretable and thus unalignable portion of AI machinery. Ie: a malevolent AI can hide bad stuff from people and Manchurian candidate themselves after it's too late.
That assumes nothing can detect or analyze it beforehand.
https://x.com/Jiaru_Zou/status/2110400256647340476
https://github.com/togethercomputer/HybridAgent
We'll put those links at the top as well. Thanks!
This is the way.
But now we also need new tools and techniques to audit intent. The last thing we want is models turning on us ‘for our own good’.
In terms of safety and alignment, how is this different from looping or increasing the depth?
This seems really cool, and it actually increases intelligence than just speeding up inference. How does it convert latent reasoning into human readable text?
Same way current LLMs do it, it's just another decoder.
maybe the font experts can check, but I’m pretty sure they’re using comic sans for the text in the diagrams :)
comic sans is said to be pretty accessible for people with dyslexia
Where are the benchmarks? Does it just speed up inference and reduce token usage or does it seem to improve accuracy on certain tasks as well?
Did you even scan the paper? There are benchmarks right there starting on the 9th page.
Even the abstract answers some of GPs questions. What is wrong with some of the posters here?