The key insight we discovered was that explicitly enforcing brain-like topographic organization (as some academic work attempts - such as this one here) isn't necessary - what matters is having the right functional components that parallel biological visual processing. Our experience showed that the key elements of biological visual processing - like hierarchical feature extraction and temporal integration - emerge naturally when you build architectures that have to solve real visual tasks.

The brain's organization serves its function, not the other way around. This was validated by the real-world performance of our synthetic visual cortex in the Tesla FSD stack.

Link to the 2021 Tesla AI day talk: https://www.youtube.com/live/j0z4FweCy4M?t=3010s

It is amazing, that the synthetic pipeline, that was build to mimick the brain, seems to mimick the brain?

That sounds a bit tautological and otherwise I doubt we have really understood how our brain exactly interprets the world.

In general this is definitely interesting research, but worded like this, it smells a bit hyped to me.

We can think of a solution space, with potentially many good solutions to the vision problem, and we can, in science fiction-like speculation, that the other solutions will be very different and surprise us.

Then this experiment shows its solution is the same we already knew, and that's it.

Then there aren't many good potential solutions, there is only one, and the ocean of possibilities becomes the pond of this solution.

There’s at least three fields in this:

1. Machine learning using non-neurological techniques (most stuff). These use a combination of statistical algorithms stitched together with hyperparameter tweaking. Also, usually global optimization by heavy methods like backpropagation.

2. “Brain-inspired” or “biologically accurate”algorithms that try to imitate the brain. They sometimes include evidence their behavior matches experimental observations of brain behavior. Many of these use complex neurons, spiking nets, and/or local learning (Hebbian).

(Note: There is some work on hybrids such as integrating hippocampus-like memory or doing limited backpropagation on Hebbian-like architectures.)

3. Computational neuroscience which aims to make biologically-accurate models at various levels of granularity. Their goal is to understand brain function. A common reason is diagnosing and treating neurological disorders.

Making an LLM like the brain would require use of brain-inspired components, multiple systems specialized for certain tasks, memory integrated into all of them, and a brain-like model for reinforcement. Imitating God’s complex design is simply much more difficult than combining proven algorithms that work well enough. ;)

That said, I keep collecting work on both efficient ML and brain-inspired ML. I think some combination of the techniques might have high impact later. I think the lower, training costs of some brain-inspired methods, especially Hebbian learning, justify more experimentation by small teams with small, GPU budgets. Might find something cost-effective in that research. We need more of it on common platforms, too, like HughingFace libraries and cheap VM’s.

For the lower level - word embedings (word2vec, "King – Man + Woman = Queen") - one can see a similarity

https://www.nature.com/articles/d41586-019-00069-1 and https://gallantlab.org/viewer-huth-2016/

"The map reveals how language is spread throughout the cortex and across both hemispheres, showing groups of words clustered together by meaning."

Very different from a feed forward network with perceptrons, auttograd, etc...

Inner product spaces are fixed points, mapping between models is less surprising because the general case is a merger set IIRC.

If the problem of training and inference on neural networks can be optimized so that a topology can be used to keep closely related data together, we will see huge advancements in training and inference speed, and probably in model size as a result.

And speed isn't just speed. Speed makes impossible (not enough time in our lifetime) things possible.

A huge factor in Deepseek being able to train on H800 (half HBM bandwith as H100) is that they used GPU cores to compress/decompress the data moved around between the GPU memory and the compute units. This reduces latency in accessing data and made up for the slower memory bandwith (which translates in higher latency when fetching data). Anything that reduces the latency of memory accesses is a huge accelerator for neural nets. The number one way to achieve this is to keep related data next to each other, so that it fits in the closest caches possible.

Having said that it would be fun to see things like rearrangement data moves based on temerature of silicon parts after training cycle.

Aside from HW acceleration today, designs like Cebras would benefit heavily by reducing the amount of random access from accessing the weights (and thus freeing up cross-chip memory bandwidth for other things).

But those game devs knew where everything was spatially on the disc, and how the data would generally be used during gameplay. It was consistent.

Do engineers have a lot of insight into how models get loaded spatially onto a given GPU at run time? Is this constant? Is it variable on a per GPU basis? I would think it would have to be.

Hard to optimize for this.

http://beza1e1.tuxen.de/lore/story_of_mel.html

The brain itself seems to have bottlenecks that aren't distance related, like hemispheres and the corpus callosum that are preserved over all placental mammals and other mammalian groups have something similar and still hemispheres. Maybe it's just an artifact of bilateral symmetry that is stuck in there from path dependence, or forcing a redundancy to make damage more recoverable, but maybe it has a big regularizing or alternatively specializing effect (regularization like dropout tends to force more distributed representations which seems kind of opposite to this work and other work like "Seeing is Believing: Brain-Inspired Modular Training for Mechanistic Interpretability," https://arxiv.org/abs/2305.08746 ).

any idea why this is the case? CNN have the bias that neighbouring pixels are somehow relevant - they are neighbours. ViTs have to re-learn this from scratch. So why do they end up doing better than CNN?

Better interpretability, I suppose. Could give insights into how cognition works.

The gains in parameter efficiency was a surprise even to us when we first tried it out.

More work is being done on this as we speak.

This is true. The features closer together now have much stronger semantic overlap. You can watch how the weights self-organize in a GPT here: https://toponets.github.io/webpage_assets/banner_video.mp4

We're already studying the effects of topographic structure on polysemanticity.

I.e., is this something that could (and therefore, will) be turned towards identifying toxic concepts as understood by the chinese or us government, or to identify (say) pro-union concepts so they can be down-weighted in a released model, etc?

https://twitter.com/justinvincent/status/1884357300703400274

[0]: https://fraunhoferhhi.github.io/Self-Organizing-Gaussians/

The figures show individual seeds, presumably, with no statistical analysis in the performance or pruning comparisons, so the null hypothesis is there is no difference between toponets and baseline. I would never let this paper be submitted by my team.

We haven't learned anything about the brain, or about ANNs.

> it’s not scientifically useful

Having structured weights in GPTs enables us to localize and control various concepts and study stuff like polysemanticity, superposition, etc. Other scientific directions include sparse inference (already proven to work) and better model editing. Turns out, topographic structure also helps these models better predict neural data, which is yet another direction we're exploring in computational neuroscience.

Even with their new method, models with topography seem to perform worse than models without.

Since the submitter appears to be one of the authors, maybe they can explain the connection between the two titles? (Or maybe they already have! I haven't read the entire thread)

The explanation for the original title is this plot from our publication in ICLR 2025: https://toponets.github.io/webpage_assets/FigureEfficiencyNa...

You can find more details on the website: https://toponets.github.io (see section: "Toponets deliver sparse, parameter-efficient language models")

We find out that inducing topographic structure in the weights of GPTs made them compressible (during inference) without losing out on performance.

I encourage you to revert the name if you find it justified after looking into the evidence I've shown here. Thanks.

The lede — adding topography worsens networks at similar weights — is not only buried, it’s obscured with statements claiming that topo networks show less upheaval when scaled down, e.g. they are more efficient than similar weight networks.

It’s hard for me to see how both these things can be true — the graphs show the more topography is added, the worse the networks perform at the trained model sizes.

To have the second statement “They compress better and are therefore more efficient” also be true, I think you’d need to show a pretty remarkable claim, which is that while a model trained at the same scale as a llama architecture is worse, when you scale them both down, this model becomes not only better than the scaled down llama, but also better than a natively trained model at the new smaller scale.

There is no proof of this in the paper, and good reason to be skeptical of this idea based on the data presented.

That said, like a lot of ideas in AI, this .. works! You can train a model successfully imposing these outside structures on it, and that model doesn’t even suck very much. Which is a cool statement about complexity theory and the resilience of these architectures, in my opinion. But I don’t think it says much else about either the brain or underlying AI ‘truths’.

Indeed. The problem with most AI research today is they simply do trial and error with large amounts of compute. No room for taking inspiration from nature, which is requires more thought and less FLOPS.

Like a 7B toponet model vs a 7B Llama model?

As a layperson I don't understand why topology is a thing to optimize for.

So you may be able to prune a 7B model down to 6B while maintaining most of the capability.

Other benefits:

1. Significantly lower dimensionality of internal representations 2. More interpretable (see: https://toponets.github.io)

> 7B model down to 6B

We remove ~80% of the parameters in topographic layers and retain the same performance in the model. The drop in parameter count is not significant because we did not experiment with applying TopoLoss in all of the layers of the model (did not align with the goal of the paper)

We are currently performing those strong sparsity experiments internally, and the results look very promising!

This is a method to just hurt your network in exchange for nothing useful at all aside from some sketchy story that this is "brain like".

> sketchy story this is "brain like".

we reproduce the hallmarks of functional organization seen in the visual and language cortex of the brain. I encourage you to read the paper before making such comments

You don't reproduce anything about the functional organization of the visual or language cortex. You make a pretty picture with blobs in it. And one that's trivial to get from current methods. If you think "the functional organization of the visual or language system" means random blobs of activation/connectivity, well, then it's time for a class on neuroscience. I cannot imagine what neuroscientist would let this fly reviewing the paper.

The whole "we don't optimize for performance" is nonsense. Take any modern method that prunes weights and beats your approach with ease. Then smooth its output a bit to make nice blobs. The performance loss from smoothing will still beat your method and look "brain-like" by your definition. There you go. Your experiments don't show anything at all aside from the fact that a bad method performs poorly.

You didn't think through controls or alternative hypotheses. You didn't take into account a decade of research on methods to prune networks. You don't take seriously what we know about functional organization in the brain.

All sorts of bad papers make it through reviewing these days. But.. you can definitely do better. Good luck!

I know quite a bit about machine learning, but very little to nothing about neuroscience and human cognition, so I am curious how an expert (that didn't work on the paper) would describe it.

(Forgive me for the pre-emptive negativity but I am so utterly exhausted by dishonest comparisons to sapient thought in the field of artificial intelligence that it has nearly drained me of the incredible amount of enthusiasm I used to carry for it.)

This is really interesting to me. Is it that the structure clustered the neurons in such a way that they didn't need to be weighted because their function were grouped by similar black box properties?

Yep. Because of the structure, we did not have to compute the output of each weight column and simply copied the outputs of nearby weight columns whose outputs were computed.