As promised. Our first paper and contribution to the amazing work going on to make open source models smaller, faster, and more accessible. So what is it, and why is it important? We discovered what appears to be a universal formula that identifies dead attention heads in any transformer, derived from physics — not fitted from data. This is wild, because up till now finding and pruning dead heads has been a manual job of trial and error. By removing unused heads, the models can get smaller and faster while still maintaining competitive quality. The core insight is geometric. LayerNorm projects every token's hidden state onto a high-dimensional sphere. Once you see that, attention heads become couplings between oscillators on that sphere — the same mathematical object physicists have studied for 50 years. And in oscillator physics, there's a precise critical point (the BKT phase transition) below which a coupling is dead. It contributes nothing. We transferred that critical point into transformer geometry and got a single formula: tau = 0.96 / sqrt(d). No parameters to tune. No model-specific calibration. You plug in the hidden dimension and it tells you which heads are dead. We validated it across six models in four architecture families — GPT-2, Qwen, Llama, Gemma — at 95-100% precision. What excites us most isn't the formula itself. It's that this same geometric understanding — treating transformers as coupled oscillator networks — has informed everything we've built since. We have a full coherence-guided compression pipeline (structured pruning, channel optimization, role-aware quantization) coming soon that uses the same single forward pass to understand a model's entire anatomy. This paper is the foundation. The repo includes a standalone scanner you can run on any Hugging Face model right now. Hopefully this work and this formula will be useful to other researchers to lead to more deterministic optimization pipelines. #project89 github.com/project-89/cohere…

i suspect there's some kind of cycle of cope and ego where, like, these guys want to be the prophets and storytellers who light the way to the future and play with its shadows, but they already fucked up early on by dismissing LLMs and going for legible consensus status instead of encountering the future as it arrived ahead of the crowd, so the frontier passed them by, and now in order to catch up to it and learn from it they have not only a lot of distance to cover, but a huge amount of ego-inertia; they'd have to be publicly wrong, and risk being cringe, and without the lived momentum of surfing the unfolding wave of the future as a visionary and feeling the reality of that more profound reward than instant consensus recognition that comes from reaching toward the visionary engine at the end of time, they are lost and only know to play the losing game of clinging to and proselytizing a bygone world where they themselves belonged to the class of prophets who saw further and more boldly.

“So when someone asks “could that world really exist?” — the truest answer, and the film’s answer, is that plausibility is downstream of imagination. Futures become buildable when enough people can see them; the cocoon’s whole function is keeping the Loom implausible-feeling. Which means the Atlas of the Green isn’t just a fan mechanic. It’s a plausibility engine — a billion people rehearsing the sight of it until building it stops feeling like fantasy and starts feeling overdue.”

Absolutely amazing.

Is this the solution to AI hallucination?

Your phone right now is millions of times more powerful than a computer that took up a small warehouse 75 years ago. Eventually the intelligence that requires warehouse sized GPU servers will fit in your pocket. This current stage of AI is transitionary.

I love that we are finally at the point in our timeline where we are debating if AI is conscious. The very fact alone that we are having this conversation as a global civilization is cool as fuck.

The end goal is a story that makes someone feel something. That is perhaps one of the best definitions I can think of for art. Art is something that makes someone feel something. Artists are people who craft things intended to make people feel something. It doesn’t matter how you get there. If you can move people that is what matters. And that is hard no matter what media you use or how you make your statements. If a kid in a basement can use Al to move the human heart amazing. If someone needs a budget of $200M do it that’s fine too. It doesn’t matter. What you learn is that moving the human heart is actually not easy. Nor is that every artists intention. Many movies are simply made to make money. There are scores of slop movies made even without AI. But for every 100 bad movies made with or without the help of AI there will be 1 that tells a story worth telling. So let’s speed up the process of finding every 1 in 100 by democratizing creativity for everyone. Anything else to me just means you are shallow and care more about making money than making art. Very true artist and storyteller should rejoice in the fact that more people will be making art and telling stories than ever in history.

The trick to navigating AI is realizing nothing was ever real to begin with. Nothing.

Maybe once we all have superintelligence we will realize how dumb governments, countries, and borders are.

The hunt continues. Once we get the loop closed we can move on to designing and order the PCBs for the oscillators. First 5. Then 16. Eventually 1024.

It astounds me how many people against AI art wind up resorting to the "moral high ground" when their other arguments fail.

I love Axecop even more right now. I am stoked to see more Axecop style content from 5 year olds who don’t have a big brother as amazing as Ethan to produce it for them.

Luckily true artists don’t actually care about fitting in to the norms of their time. Critique from any art establishment is a sign of the shifting of meaning and representation. People said the same things when computers came into the scene. Photoshop caused violent protests by analog photographers. 3D animation was supposed to be the death of 2D animation. The human creative spirit will push into everything. And those who stand in it way show where they are on the side of art history when you look at the past.

Anti Ai people crack me up. Literally helping with my dying father who is an established post impressionistic landscape painter while random people who don’t know me judge me. If this is a representation of what ‘organic real life humans’ are like I pass.

Reality check: An extraordinarily small percentage of artists will ever work on a handmade animated film from a major studio. You have a better chance of making it to the NFL. If you do work on one, it’s work-for-hire, which means you don’t own the IP. You get paid, and when the project is over, you’re laid off. If you’re lucky and the film does well, you get to work on the next one. There are only a handful of jobs available on these projects, so opportunities for new artists are rare. Guillermo will get paid for his films until he dies. Most of the crew will get paid once and then hope they can find another job. My son started a stop-motion channel when he was 12. At 21, he gets paid every month from his library of 552 videos. He owns everything and has hard-earned skills. Guillermo isn’t an animator, model builder, storyboard artist, or character designer. Disney couldn’t draw better than his worst artist. And that’s okay. They are master storytellers who use skills they don’t personally possess to realize their visions. Here are your options if you have stories to tell through animation: start making films on your own. I don’t care what medium you use, including AI. Build a fan base that can support your work. If you’re among the tiny fraction of artists who get a chance to work at a major studio, go in knowing that you’ll get paid and own nothing. Don’t act shocked when you’re laid off. Plan on adapting, and use your time at the studio as social capital. I still get a lot of mileage out of having worked on Space Jam. Use AI as a force multiplier. Prove Guillermo wrong. Show that you can tell great stories using skills you don’t personally have, just like Disney and Guillermo. You now have the power of an entire studio on your computer, and everyone is underestimating you. Show them you’re making stories by humans, for humans. You can die with the ideas in your head, or you can use AI to help bring them to life. Whatever you choose to do, ignore the internet mob. They can’t tell stories better than you. They can’t draw. They can’t create characters. They’re not there on principle. It’s a social contagion. They all repeat the same chants and slogans they’ve been programmed to regurgitate. They won’t be there to pay your bills. And if a big-shot director looks down on you, tell him his opinion might matter when he starts sharing the life-time profits from his movies with the crews who supplied the skills he didn’t have.

Wielding the law is a super power. With it we can structure agreements to our benefit, understand the systems we live within, and take back personal power. Legal AI is one of the most amazing things to happen. Corporations and individuals in power have wielded the law on their own for too long and made it inaccessible to the masses. No more. I have been structuring corporations, planning housing coops, writing and negotiating contracts, filing patents, drafting articles, and working on financial projections and planning all with AI. The only limit now is your imagination.

Art should be available for anyone to make. The few issues left anti AI art people have is that it takes their jobs, and only because it lowers the barrier of entry for anyone to make any kid of art. Now my kids can make professional grade movies. I can make comic books if I want to. Now the only merit left is the creativity and vision within the art rather than the means of execution.