🇺🇦🇮🇱dmitriy samsonov

github.com/d0rc/n-interac… - A hackable @PyTorch script to visualize emergent structures in n-interacting-particle systems.

chemical reaction between sodium and water

@UnoLeeb @GitaGopinath I just like to make fun of them. And it doesn’t change the fact that people on photo are criminals.

@d0rc @GitaGopinath 他是女权

A painting of the end of meritocracy: A meeting of the two largest economies and not one woman at the table.

7 roots orbit in ℂ. Perturb the degree-7 polynomial's coefficients with Gaussian noise: isolated roots stay sharp, clustered ones smear into a cloud. These are the polynomial's ε-pseudozeros: Wilkinson conditioning made visible. #Pseudospectra #MathArt

@MickRhythm russia is not a state, it’s a gang holding hostages; but, yes, any state is indeed illegitimate, as well as any politician or religious leader is a criminal.

russia is not legit. When a state uses force, deception and aggression as tools on the world stage, that state is illegitimate. This event is a farce.

@bnielson01 But the universe we live in is not a set of all possible problems:) or is it?:)

There's a mathematical proof that says no algorithm — no matter how clever, how sophisticated, or how well-designed — can outperform random guessing when averaged across all possible problems. Not A*, not neural networks, not even humans in the loop. Every advantage on one class of problems is paid for, dollar for dollar, somewhere else. It sounds like it should be false. It isn't. Please review the article: the theorem is sound. So you can't just dismiss it. Welcome to the "No Free Lunch Theorem". This raises an interesting question: how can humans be universal learners if this theorem says universal learns are impossible? Make your best arguments here. I hint at one possible resolution to this problem. mindfiretechnology.com/blog/archive/t…

We put the paper online that provides further details (beyond my ICLR keynote) on the role of spontaneous symmetry breaking and Goldstone modes in deep learning. Enjoy! (w/ Nabil Iqbal, Thomas Andy Keller, Takeru Miyato and Yue Song.) arxiv.org/abs/2605.14685

@paulg Because all politicians and all religious leaders are pure evil. Do not trust the word they say.

At one point my son and his friend kept looking for shortcuts to getting rich. Over and over I told them the way to do it is just to make something people want. If this is what I tell my own kids about getting rich, why won't politicians believe this is how a lot of people do it?

@McFaul @GUkrainska I wish all those involved in politics and religion dead or retired. Stop pretending you guys can do any good to our biological species. You are decease. May you all be damned.

I wish Trump could praise leaders of our democratic allies just half as much as he fawns over autocratic leaders like Xi and Putin.

Datadog Toto 2.0: 4M→2.5B open time-series models. Forecasting gets its foundation model era Not another chatbot in a suit. Just a model reading production noise: "Your latency is about to ruin your weekend." Useful AI. May your CRPS be low! May your incidents be forecasted!

With NormaCore Station, your robot's full operational life is the dataset. ✅Continuous capture at max hardware speed. ❌ No downsampling. ❌No "start/stop" recording. It’s all automatically compressed, encrypted, and stored. When you need a dataset, just pull it. No scripts. No pipelines. Just data. 🎯

@petite_michelle @lindermannAnon Give me the reason, why all members of all governments and all religious leaders should stay alive for one more day.

Give me one reason why Moscow shouldn't be in ruins?

@RTSG_Main Not heroic enough, since your stupid parents survived his rule.

✅FACT: Stalin was a HERO!

Attention @arxiv authors: Our Code of Conduct states that by signing your name as an author of a paper, each author takes full responsibility for all its contents, irrespective of how the contents were generated. 1/

Today we’re releasing Toto 2.0: a family of open-weights time series foundation models spanning 4M to 2.5B parameters. The question we set out to answer was simple (yet previously open): Do time series foundation models get reliably better as they scale? Our answer: yes! 🧵

@mhmck 100% pure truth.

Peace is not possible without destroying the Moscow Empire and liberating the captive nations.

ChipMATE achieves 75.0% and 80.1% pass@1 on VerilogEval V2 with 4B and 9B base models, outperforming all existing self-trained models and even DeepSeek V4 with 1600B parameters.

@simplifyinAI Breaking indeed:)

🚨 BREAKING: NVIDIA proved back-propagation isn't the only way to build an AI. Billion-parameter models were trained without a single gradient. No calculus, no exploding memory, no massive GPU clusters. The culprit? A long-dismissed technique called Evolution Strategies. NVIDIA and Oxford just made it scalable with EGGROLL, which replaces bloated mutation matrices with two tiny ones, enabling hundreds of thousands of parallel mutations at inference-level speed. They're pretraining models from scratch using only simple integers. No backprop. No decimals. We assumed the future of AI required endless precision hardware. Evolution had other plans.

Capitalism is the only economic system. Everything else is a violence system.