[dead]
Show HN: Duplicate 3 layers in a 24B LLM, logical deduction .22→.76. No training
41–50 of 100 posts
Re: Show HN: Duplicate 3 layers in a 24B LLM, logical deduction .22→.76. No training
#42If running twice is good, then is running N times even better? I wonder if you could even loop until some kind of convergence, say hitting a fixed point (input equals output). I wonder if there's even a sort of bifurcation property where it sometimes loops A->A->A, but other times A->B->A, or more, rather like the logistic map fractal.
That's how deep equilibrium models were discovered.
Whats's more. It was found out that you only need a single looped layer to be equivalent to a multi layer network.
Re: Show HN: Duplicate 3 layers in a 24B LLM, logical deduction .22→.76. No training
#43[dead]
Re: Show HN: Duplicate 3 layers in a 24B LLM, logical deduction .22→.76. No training
#44[dead]
Re: Show HN: Duplicate 3 layers in a 24B LLM, logical deduction .22→.76. No training
#45Fascinating! I wonder if new training techniques could emerge from this.
If we say layer-1=translater, layer2-5=reasoner, layer6 retranslater, could we train small 6 layer models but evaluate their performance in a 1>n*(2-5)>6 setup to directly train towards optimal middle-layers that can be looped? You'd only have to train 6 layers but get the duplication-benefit of the middle layers for free.
Re: Show HN: Duplicate 3 layers in a 24B LLM, logical deduction .22→.76. No training
#46> I replicated David Ng's RYS method [...] found something I didn't expect.
> Transformers appear to have discrete "reasoning circuits" — contiguous blocks of 3-4 layers that act as indivisible cognitive units. Duplicate the right block and the model runs its reasoning pipeline twice. No weights change. No training. The model just thinks longer.
How did you not expect that if you read his post? That's literally what he discovered, two years ago.
For anyone interested, there's more meat in the post and comments from last week: https://news.ycombinator.com/item?id=47322887
Re: Show HN: Duplicate 3 layers in a 24B LLM, logical deduction .22→.76. No training
#47There's still a lot of low hanging fruit left IMO. Good find and rather funny to think about as you can have someone simply clone the various layers multiple times and instead of spending millions of dollars retraining the model increase performance significantly with "this one trick".
I've been running my own (admittedly naïve) experiments of new, wacky ideas for both LLMs (well, SLMs) and for Image-Super-Resolution models.
I'm just trying different kinds of attention mechanisms, different configurations of the network, adding loops, ... All kind of wacky ideas. And the real weird thing is that 99% of the ideas I try work at all.
Re: Show HN: Duplicate 3 layers in a 24B LLM, logical deduction .22→.76. No training
#48> I replicated David Ng's RYS method [...] found something I didn't expect. > Transformers appear to have discrete "reasoning circuits" — contiguous blocks of 3-4 layers that act as indivisible cognitive units. Duplicate the right block and the model runs its reasoning pipeline twice. No weights change. No training. The model just thinks longer. How did you not expect that if you read his post? That's literally what…
That's explicitly not the unexpected part. Read the rest of the post.
Re: Show HN: Duplicate 3 layers in a 24B LLM, logical deduction .22→.76. No training
#49That weird part is kind of what I was expecting.
This goes to the thing that I posted on the thread a couple of days ago. https://news.ycombinator.com/item?id=47327132
What you need is a mechanism to pick the right looping pattern, Then it really does seem to be Mixture of experts on a different level.
Break the model into input path, thinking, output path. and make the thinking phase a single looping layer of many experts. Then the router gets to decide 13,13,14,14,15,15,16.
Training the router left as an exercise to the reader.
Re: Show HN: Duplicate 3 layers in a 24B LLM, logical deduction .22→.76. No training
#50Assuming the benchmarks are sound (rather than capturing a fluke), the provided explanation still does not pass the smell test. As far as I can tell, there is nothing about the training process of these models that would encourage them to make the output of any layer apart from (n-1) meaningful as the input of layer n, unless perhaps these layers were initialised as identity and the training process did not get to ch…
Empirical findings tell a very different tale: all LLM layers use vaguely compatible internal representations. And middle layers in particular can be almost interchangeable - a lot of what they seems to be "iterative refinement of the same representations". Proven by various probes and ablations, but the most obvious one is probably the good old logit lens.
This is likely to be shaped by tied embeddings and skips on one end, and maybe training pressures on the other.
The very top of FF stack and the very bottom of FF stack both reflect the same token embeddings - and this propagates through the model, setting up a shared identity space. Skip connections propagate that through the layers. No explicit shared identity imposed, but there is an implicit one set by the architecture. Fairly well established.
(Now: highly speculative! Attention over past tokens creates an implicit "robustness/convergence" pressure? The model can't be "certain" if it'll have access to the right representations at a given layer, because representations depend not just on the past layers, but also on the highly uncertain contents of previous tokens as passed through attention. Which in turn depends on more of the same, increasing variance further. So the training causes: "each layer can't be certain of what it will have access to, so it develops to refine anything it currently has access to in a convergent fashion, because that's what's useful under pressure of attention-induced uncertainty".)
LLMs are notoriously nonfragile, and robust to perturbations. Far more so if you anneal with SFT/distillation after your model surgery, although this wasn't done here. Plenty of weird franken-LLM experiments prove that empirically.
So I'm not too surprised to find that someone has managed to improve benchmark performance on a few narrow tasks by duplicating a few middle layers. "Duplicating a few layers that were doing convergent iterative refinement benefits a few tasks that suffered from insufficient depth of convergent iterative refinement" is a fairly reasonable hypothesis, in my eyes.
The chances of duplication "breaking something somewhere" are high, and I would expect the capability profile of an unannealed franken-LLM like this to have a few gaps in it if evaluated extensively against the original. But "franken-LLM layer duplication can actually improve some things" is far too plausible with what we know to be dismissed pre-emptively.