It achieves a 6.74 FID on ImageNet 64x64.
The official thread from @unconvai !
Introducing our first model, Un-0!
We trained an image generator powered by a backbone of coupled oscillators in place of a more traditional conventional neural network.
Many users praised UnconvAI's Un-0 image generator for its novel coupled-oscillator approach, calling it cool, creative, and promising for hardware-efficient computation.
No Digg Deeper questions have been answered for this story yet.
Did you know this wiggly thing can be a computer?
We think of computation as symbolic: add, subtract, multiply....
…but…you can also compute directly with voltage, current, charge, waves, materials (octopus arms…look it up)...or wiggles.
Today, we’re (@unconvAI ) sharing Un-0: our first step toward scaling this idea to modern AI workloads.
Introducing our first model, Un-0!
We trained an image generator powered by a backbone of coupled oscillators in place of a more traditional conventional neural network.
We’re at the beginning of the unconventional AI journey, but we are releasing artifacts to the community to enable us all to build together. We’ve open-sourced the Un-0 weights, training scripts, and ablation code.
If you build physics-based models, plug them into our scaffold and share your success!
Read the blog: https://unconv.ai/blog/introducing-un-0-generating-images-with-coupled-oscillators
Visit our GitHub: https://github.com/unconv-ai/Un-0
How does it work? Un-0 uses a system of coupled Kuramoto oscillators.
Initial random phases evolve under the pull of other oscillators through learned coupling strengths.
Is the physics actually doing the work? Yes.
Our ablations show that the trained dynamics provide value over a decoder only baseline and a random Kuramoto feature reservoir, and that increasing the number of integration steps increases model quality.
The results? We scale Un-0 up to 16k oscillators and 322M parameters and achieve an FID of ~6.74 on ImageNet 64x64.
To our knowledge, this is the most capable model based on a simulation of a physical system to date, expanding the Pareto frontier for small generative models. Though there is still remaining work to scale model quality as a function of model size.
At Unconventional AI, we're building a new kind of computer that runs AI on the dynamics of a physical system, at a fraction of the energy today's machines need.
We’re leveraging the dynamics of a physical system, such as the noisy, time-varying behavior of analog circuits that compute with analog voltage and current instead of conventional digitized numbers.
Un-0 illuminates a path towards running modern AI workloads on physical substrates that are more efficient than today's hardware.
with only 64 spins, we can't run ImageNet, but we can generate some very nice images of squares!
https://github.com/zachbe/digial-ising
And the official @unconvAI post!
Introducing our first model, Un-0!
We trained an image generator powered by a backbone of coupled oscillators in place of a more traditional conventional neural network.
blog: https://unconv.ai/blog/introducing-un-0-generating-images-with-coupled-oscillators/
repo: train or extend it yourself. https://github.com/unconv-ai/Un-0
Did you know this wiggly thing can be a computer?
We think of computation as symbolic: add, subtract, multiply....
…but…you can also compute directly with voltage, current, charge, waves, materials (octopus arms…look it up)...or wiggles.
Today, we’re (@unconvAI ) sharing Un-0: our first step toward scaling this idea to modern AI workloads.
@davidcox i actually made a proof of concept for hopfield networks :)
https://www.zach.be/p/implementing-hopfield-networks-on
@unconvAI Coupled oscillators instead of attention is a wild swing. Kuramoto dynamics carrying the load is cool, but where does it break first, can it scale past toy resolutions or does the sync go unstable as params grow? Rooting for the weird approach!
@unconvAI this approach sounds interesting, is there a paper associated to this work?
yea, good question. We don't discriminate at unconventional. We have several different dynamics in the hopper as well...we're looking for what works.
For Un-0 we chose kuramoto because it was quite unusual and unfamiliar to many in ml (though there is some great work by @takeru_miyato and others.
At the same time, kuramoto dynamics are quite familiar in physics to characterize different phenomena. so there are also many ideas/computations to analyze and take inspiration from.
@mcarbin @unconvAI This is how I described it to people a month ago
@blip_tm this is really sick! it's not every day you meet someone who appreciates a nice ising square.
you could probably build a hopfield network with DIMPLE pretty easily
@mov_axbx @unconvAI nice! awesome work.
Love this work!
Introducing our first model, Un-0!
We trained an image generator powered by a backbone of coupled oscillators in place of a more traditional conventional neural network.
Introducing our first model, Un-0!
We trained an image generator powered by a backbone of coupled oscillators in place of a more traditional conventional neural network.
super impressive for unconventional to develop the best coupled oscillator image generation model in 6 months!
if you want to give coupled oscillator computing a try for free using FPGAs, check out my open-source architecture, DIMPLE ⬇️
Introducing our first model, Un-0!
We trained an image generator powered by a backbone of coupled oscillators in place of a more traditional conventional neural network.
