4 ms·

Perception seems to be one of the main constraints on LLMs that not much progress has been made on. Perhaps not surprising, given perception is something evolut

Perception seems to be one of the main constraints on LLMs that not much progress has been made on. Perhaps not surprising, given perception is something evolution has worked on since the inception of life itself. Likely much, much more expensive computationally than it receives credit for.

Workaccount2 10mo ago
I strongly suspect it's a tokenization problem. Text and symbols fit nicely in tokens, but having something like a single "dog leg" token is a tough problem to solve.
- stalfie 10mo ago
  The neural network in the retina actually pre-processes visual information into something akin to "tokens". Basic shapes that are probably somewhat evolutionarily preserved. I wonder if we could somehow mimic those for tokenization purposes. Most likely there's someone out there already trying. (Source: "The mind is flat" by Nick Chater)
  - machiaweliczny 10mo ago
    It's also easy to spot as when you are tired you might misrecognize objects, I caught myself with this when doing long roadtrips
    - stalfie 10mo ago
      AFAIK this is actually a separate mechanism, which is part of the visual cortex and not the retina. Essentially recognizing even a single object requires the complete attention of pretty much your entire brain in the moment of recognition. What I am referring to is a much more basic form of shape recognition that goes on at the level of the neural networks in the retina.
- recitedropper 10mo ago
  I think in this case, tokenization and percpetion are somewhat analogous. I think it is probably the case our current tokenization schemes are really simplistic compared to what nature is working with. If you allow the analogy.
orly01 10mo ago
Why should it have to be expensive computationally? How do brains do it with such a low amount of energy? I think catching the brain abilities even of a bug might be very hard, but that does not mean that there isn't a way to do it with little computational power. It requires having the correct structures/models/algorithms or whatever is the precise jargon.
- recitedropper 10mo ago
  This is the million dollar question. I'm not qualified to answer it, and I don't really think anyone out there has the answer yet. My armchair take would be that watt usage probably isn't a good proxy for computational complexity in biological systems. A good piece of evidence for this is from the C. elegans research that has found that the configuration of ions within a neuron--not just the electrical charge on the membrane--record computationally-relevant information about a stimulus. There are probably many more hacks like this that allow the brain to handle enormous complexity without it showing up in our measurements of its power consumption.
  - nick32661123 10mo ago
    Following the trend of discovering smaller and smaller phenomena that our brains use for processing, it would not be surprising if we eventually find that our brains are very nearly "room temperature" quantum computers.
  - programd 10mo ago
    My armchair is equally comfy, and I have an actual paper to point to: Jaxley: Differentiable simulation enables large-scale training of detailed biophysical models of neural dynamics [1] They basically created sofware to simulate real neurons and ran some realistic models to replicate typical AI learning tasks: "The model had nine different channels in the apical and basal dendrite, the soma, and the axon [39], with a total of 19 free parameters, including maximal channel conductances and dynamics of the calcium pumps." So yeah, real neurons are a bit more complex then ReLU or Sigmoid. [1] https://www.biorxiv.org/content/10.1101/2024.08.21.608979v2.full https://www.biorxiv.org/content/10.1101/2024.08.21.608979v2....

Workaccount2 10mo ago

I strongly suspect it's a tokenization problem. Text and symbols fit nicely in tokens, but having something like a single "dog leg" token is a tough problem to solve.

stalfie 10mo ago
The neural network in the retina actually pre-processes visual information into something akin to "tokens". Basic shapes that are probably somewhat evolutionarily preserved. I wonder if we could somehow mimic those for tokenization purposes. Most likely there's someone out there already trying. (Source: "The mind is flat" by Nick Chater)
- machiaweliczny 10mo ago
  It's also easy to spot as when you are tired you might misrecognize objects, I caught myself with this when doing long roadtrips
  - stalfie 10mo ago
    AFAIK this is actually a separate mechanism, which is part of the visual cortex and not the retina. Essentially recognizing even a single object requires the complete attention of pretty much your entire brain in the moment of recognition. What I am referring to is a much more basic form of shape recognition that goes on at the level of the neural networks in the retina.
recitedropper 10mo ago
I think in this case, tokenization and percpetion are somewhat analogous. I think it is probably the case our current tokenization schemes are really simplistic compared to what nature is working with. If you allow the analogy.

orly01 10mo ago

Why should it have to be expensive computationally? How do brains do it with such a low amount of energy? I think catching the brain abilities even of a bug might be very hard, but that does not mean that there isn't a way to do it with little computational power. It requires having the correct structures/models/algorithms or whatever is the precise jargon.

recitedropper 10mo ago
This is the million dollar question. I'm not qualified to answer it, and I don't really think anyone out there has the answer yet. My armchair take would be that watt usage probably isn't a good proxy for computational complexity in biological systems. A good piece of evidence for this is from the C. elegans research that has found that the configuration of ions within a neuron--not just the electrical charge on the membrane--record computationally-relevant information about a stimulus. There are probably many more hacks like this that allow the brain to handle enormous complexity without it showing up in our measurements of its power consumption.
- nick32661123 10mo ago
  Following the trend of discovering smaller and smaller phenomena that our brains use for processing, it would not be surprising if we eventually find that our brains are very nearly "room temperature" quantum computers.
- programd 10mo ago
  My armchair is equally comfy, and I have an actual paper to point to: Jaxley: Differentiable simulation enables large-scale training of detailed biophysical models of neural dynamics [1] They basically created sofware to simulate real neurons and ran some realistic models to replicate typical AI learning tasks: "The model had nine different channels in the apical and basal dendrite, the soma, and the axon [39], with a total of 19 free parameters, including maximal channel conductances and dynamics of the calcium pumps." So yeah, real neurons are a bit more complex then ReLU or Sigmoid. [1] https://www.biorxiv.org/content/10.1101/2024.08.21.608979v2.full https://www.biorxiv.org/content/10.1101/2024.08.21.608979v2....