Aaron Kazah

descending the gradient

[Arvind Narayanan @random_walker]

To understand and empathize with how workers in many or most fields outside software experience advances in AI capabilities, I propose a little thought experiment. https://t.co/QZG24aRiBe

[Aaron Kazah @aaronkazah]

@PimDeWitte @_offmylawn that's a fair point assuming you're 100% inferencing.

but i think a more likely outcome is that you figure out a way to have these worlds worlds + artifacts exportable to deterministic engines.

that alone would probably collapse dev time by half.

[Aaron Kazah @aaronkazah]

@PimDeWitte @_offmylawn are you afraid of backlash from gamers / game devs?

i think it's fair to say over a long horizon (5,10,20 years) world models will eventually replace (or massively accelerate) developing games (which typically takes years)

[Elon Musk @elonmusk]

@robustus Code itself will go away in favor of just making the binary directly.

The next step after that is direct, real-time pixel generation by the neural net.

[Aaron Kazah @aaronkazah]

@Scobleizer imo virtual reality only works if it's sword art online / black mirror level vr

[Aaron Kazah @aaronkazah]

i had a great time building laguna vision: a native visual input path for laguna xs.2 using siglip, a resampler, and lora adapters trained on 300k examples.

honored to get a mention alongside so many great projects.

thanks to the @poolside & @PrimeIntellect team, judges, and everyone who made the weekend happen.

[Poolside @poolsideai]

What a weekend. Around 30 teams showed up to build on Laguna XS.2, and the bar was very, very high.

Winners below 🏆

1st: Overthinking Machines Labs
@emilfristed
Pseudo-full-duplex with text-only models through dialogue modeling with silence tokens.
https://t.co/rP4BZWrtrz

2nd: Coding Kernels by the Pool
Charlie Masters, Evan O’Leary, Jessica Mak
Laguna-Dense: a ~3B fully dense distillation of Laguna XS.2 for generating CUDA kernels from PyTorch.
https://t.co/OLmGezfGqF

3rd: attnvq
@alaradirik
Attention-aware product vector quantization of KV caches.
https://t.co/SwXmcIEOhn

Honorary mention: Laguna Vision
Aaron Kazah @aaronkazah
A SigLIP vision encoder + resampler + LoRA adapters, trained on 300k examples to give Laguna XS.2 a native visual input path.
https://t.co/dgrtgI7eoj

Huge congrats to the winners, and thank you to everyone who hacked, demoed, judged, helped, and pushed Laguna XS.2 in directions we would not have found on our own!

@nvidia @PrimeIntellect @adaption_ai @huggingface

[Bowen Wang @BowenWangNLP]

RLVR has become the recipe for agentic post-training. But for Computer-Use Agents, the bottleneck is not the algorithm, it is the data. 🐌

🚀 We introduce CUA-Gym: a scalable, lightweight synthesis engine that turns arbitrary task queries into verifiable RLVR data for computer-use agents. The largest open CUA RLVR dataset to date:

🎯 32,122 verifiable RLVR tasks with programmatic setup scripts + rewards
🌐 110 environments: 16 desktop apps + 94 synthesized mock web apps
🏆 Qwen3.5-based CUA models trained with GSPO reach 72.6% on OSWorld-Verified and 56.6% on WebArena

📄 Paper: https://t.co/cdvHJPzgb1
🏠 Homepage: https://t.co/kvhaOQxNx7
🤗 Dataset: https://t.co/w5vOIRdchR
💻 Codebase: https://t.co/CcRlNTlS1c
🧩 Environments: https://t.co/fNZ6YAI8LD
🧵[1/6]

[Suhail @Suhail]

Nothing seems to replace understanding the thing you're working on if you intend to improve it. AI generating 2K lines of slop may initially work but once you want to tweak it or understand the nuance of what's happening under the hood, you end up wishing you had built it up.

[Aaron Kazah @aaronkazah]

@AdrienLE https://t.co/ioEG6zlAWl

[Aaron Kazah @aaronkazah]

how long before openai seize their mandate of heaven?

[Aaron Kazah @aaronkazah]

how long before openai seize their mandate of heaven?

[terminally onλine εngineer @tekbog]

cant believe anthropic secured a partnership with the vatican

[Aaron Kazah @aaronkazah]

it's no surprise a game like elden ring emerged from japan.

[Super PINTO @PINTO03091]

@skalskip92 You might not believe it, but I simply manually annotated over 2,000,000 human body parts with ultra-precise detail. Probably no one else could do that.

[signüll @signulll]

believing in something early is effectively learning to tolerate being alone in a position long enough for the world to catch up, if it ever does.

if you do this over & over again, you can imagine why sustaining this is difficult as hell.

[Rohan Paul @rohanpaul_ai]

Demis Hassabis on the limit in today’s AI: language can describe the world, but it cannot contain it - and why "World Models" are his "longest standing passion".

Language models absorbed far more structure about reality from text than many researchers expected, because human language quietly carries physics, psychology, culture, tools, plans, and cause-and-effect.

But text is still a compressed residue of experience, not experience itself.

A sentence can say a cup falls from a table, yet it does not fully encode weight, grip, balance, friction, timing, sound, surprise, or the tiny motor corrections a body makes before it even notices them.

The world is not only made of facts that can be named; it is made of constraints that have to be lived through, touched, predicted, violated, and repaired.

That is why world models matter.

They aim to learn the hidden grammar of physical reality: how objects persist, how forces unfold, how space changes when an agent moves, and how action creates feedback.

Language models can often reason about the world because people have written so much about it.

World models try to learn what the world is like before it becomes words.

The difference is exactly what matters because intelligence is not just answering well; it is knowing what would happen next if you moved, reached, pushed, smelled, slipped, or failed.

A mind trained only on descriptions may become brilliant at explanation.

A mind trained on experience may become better at consequence.

---

Full video from "Google DeepMind" and "Hannah Fry" YT channel (link in comment)

[Aaron Kazah @aaronkazah]

@tszzl anime characters before they crash out

[Andrew Carr 🤸 @andrew_n_carr]

animation is the past, present, and future of storytelling - but it can be painful and tedious. I'm pretty excited about this workflow

cartwheel in blender!

capture motion, edit, generate, repath, repose, retime - all in your DCC.

we're testing this out, so let me know if you're interested (DMs open), we're hoping to launch a cleaned up version early next month

[Aaron Kazah @aaronkazah]

the government wants you to forget about this

[アメリカ大使館 @usembassytokyo]

Deputy Chief of Mission Aaron Snipe recently met with Shoji Yutani, CEO of Weyland-Yutani Corporation, to discuss greater 🇺🇸 🇯🇵 coordination in deep-space exploration. With companies like Weyland-Yutani considering new large-scale terraforming and atmosphere-processing projects on distant planets, ties between government and private industry have never been stronger. #weylandyutani #LV426

[Aaron Kazah @aaronkazah]

introducing trunks: the most powerful open-source git-native filesystem for ai agents.

it gives agents a normal filesystem with git semantics built in: branches, diffs, rollback, checkpoints, and push/pull.

available today.

completely self-hosted. open source. your data stays on storage you control.
https://t.co/FFfwKc9NF1

[Aaron Kazah @aaronkazah]

@bentossell servers will never take off because they require too much maintenance, the future of self hosted git is in object storage

[Aaron Kazah @aaronkazah]

@anshnanda i'm building the primitives for this
https://t.co/a7FscqhnLl

[Aaron Kazah @aaronkazah]

the computer is the closest thing we’ve ever built to a perfectly verifiable universe.

so it would be strange if a world model didn’t emerge from one.