Generative Recursive Reasoning GRAM (Generative Recursive reAsoning Models) turns recursion itself into a stochastic latent trajectory. Multiple hypotheses, alternative solution strategies, and inference-time scaling not just by depth, but by width — parallel trajectory sampling.

Should a government without a majority of the vote be able to doom entire species to extinction? Proportional representation ensures environmental protections are built on broad majority consensus, not something to be ripped up when they become inconvenient.

Creating a water-sized chip has always sounded like a stupid bet to me.

Perhaps his plan is to steal their model and usage data? Until recently I would trust Anthropic not to train on enterprise data. But not with X into the picture.

Ending direct fossil fuel subsidies alone would avoid 70,000 premature deaths from air pollution every year. It’s time to tax these deadly polluting corporations and go renewable.

Anthropic finds that they can use a weaker model to supervise a more capable model from sandbagging. Paper: Removing Sandbagging in LLMs by Training with Weak Supervision (arxiv.org/abs/2604.22082)

Every guy in a videogame:

There are many perks of working at NVIDIA... including unfettered access to coding agents like Codex. It's mind-blowing how powerful these have become in recent months. With it, I made significant progress on 4 projects at once today, including...

Excited to share HOI-PAGE, to appear at #ICML2026! 🚀 @craigleili.bsky.social generates 4D human-object interactions zero-shot from text A part-affordance graph grounds interactions via LLM+video priors, enabling complex multi-person, multi-object interactions 👉 craigleili.github.io/projects/hoi...

NVIDIA researchers published "MotionBricks", real-time 3D character animation for games and robots: Project: nvlabs.github.io/motionbricks/ Check out that demo video... very impressive!

NVIDIA Launches Nemotron 3 Nano Omni Model, Unifying Vision, Audio and Language for up to 9x More Efficient AI Agents - 30B-A3B, hybrid mixture-of-experts - blogs.nvidia.com/blog/nemotro... - huggingface.co/nvidia/Nemot...

Anyone who has been to Dubai, or Abu Dhabi, knows they are places of opulence, and excess in the middle of the desert. They don't deserve "bailing out". The American People struggling to pay for essentials like housing, healthcare, and groceries, do.

Last year, we introduced FlexOlmo, a novel way to train parts of a model independently then combine them later. BAR builds on that idea for a harder problem: how to keep improving a model without having to retrain each time. 🧵

It seems, that we have failed the communication about IMC26. Let's try again. The competition this year is here: kaggle.com/competitions... No prizes, but whole year leaderboard -- similar to KITTY and other academic competitions. 3D people, please retweet and share.

Real cool project, and a fun team we collaborate with! research.nvidia.com/labs/sil/pro...

The fossil fuel industry is probably the biggest fraud apparatus in the history of mankind. It should be joyous to take that machine down. #EEODayRI

High speed rail is a classic example of first-past-the-post policy whiplash. Governments elected with 40% or less furiously undo previous projects, so nothing gets built. Proportional representation incentivises broad consensus and stable, long term planning.

It definitely looks like a grift between Putin and Trump. Trump is most likely being blackmailed. Why else could this war instantly be improving Russia’s economic fortunes, while continuing to damage the economies of the rest of the world. There are so many other options for Trump and he chose this.

Bluesky must succeed. It just might. werd.io/coming-off-t...

Can the world stop relying on oil? Can Carney stop promoting fossil fuel use around the world? www.cbc.ca/news/busines...

Gas prices (💶/litre) in Munich: Feb 2 2026 vs today

I’m sure this was the #1 reason for the war in Iran. Enrich Big Oil and oil&gas exporters like Russia. sigh

As a Canadian this matches my experience and I’m delighted with our scores… but shouldn’t they have inverted the order of that table? We should be on top and the USA in the bottom.

According to OpenAI, their contract with the US DoW locks in current law, "even if those laws or policies change in the future". Our legal analysis, with Virgil Law CEO Luke Versweyveld, shows that this is almost certainly incorrect. www.answer.ai/posts/2026-0...

This French interview with @yoshuabengio.bsky.social about the risks of AI and what we can do to safeguard against them. It woke me up. I will reach out to see if I can volunteer. youtube.com/watch?v=6Zm-...

"What exactly do you expect Carney to do / not do?" A thread of 50 specific items (with links, and not in any particular order), a large fraction of which are in line with warning signs we pointed out early on and were told to be quiet, wait and see, or trust his 4D chess acumen. 1/??

Evolver, a near-universal optimizer for code and text. It achieved SOTA (95%) on ARC-AGI-2 (last week that is 😆) and 3x’d performance of the best open model, reaching GPT-5.2-level performance.

Once again the dividing lines turn out to be “has principles” vs. “no principles“

XStreamVGGT: Extremely Memory-Efficient Streaming Vision Geometry Grounded Transformer with KV Cache Compression Zunhai Su, Weihao Ye, Hansen Feng, Keyu Fan, Jing Zhang, Dahai Yu, Zhengwu Liu, Ngai Wong tl;dr: pruning and quantization->StreamVGGT arxiv.org/abs/2602.21780

I'm late to the game but... making crazy combos in @playbalatro.com is awesome!

Very impressive!

LLMs are largely memory bound! Tokenizers can eat >30% of on‑device memory for small (<1B) models, because we reuse the huge vocabularies from big models. Our method lets you distill large models into small ones with much smaller vocabulary, reducing overall memory footprint.

We can't fall into right-wing populism’s lie that the most vulnerable in society are to blame for wealth inequality in our countries. We need to build movements that tell the truth: the story of wealth inequality is not a cultural one, but a class one.

Emails Show Epstein Scheming That Environmental Destruction Could Solve “Overpopulation” | Article by Joe Wilkins (@joeonhere.bsky.social) for @futurism.com: futurism.com/science-ener...

This is Edward Niedermeyer explaining, in plain terms, why Tesla stock stays float but above the $400 mark, despite the car sales going down. I encourage anyone to watch it from the marker onwards I've set as Edward explains that #$TSLA is built on momentum and what we can do to bring him down

Small Language Models (SLMs) don’t have the capacity to remember everything in their training data. Which tokens should they learn to predict, and when should they ask for help? We tackle this question in our new preprint. You can check it out on arxiv: arxiv.org/abs/2602.12005 🧵1/7

Absolutely breathtaking!

Billionaires who lobbied Trump to halt Canadian bridge praise EZ-Bribe App

OneVision Encoder Currently, multimodal models sample frames at a very low FPS, reducing continuous motion to a handful of static snapshots. OneVision Encoder uses video codecs - native motion vectors and residual signals to select patches,

The new ByteDance SeeDance 2.0 video model is VERY good. Each video is the very first output of the prompt. There are four, worth seeing them all to get a sense of the range (and potential issues) "A nature documentary about an otter flying an airplane"

xAI founding members are jumping ship.

You can now train MoE models 12× faster with 35% less VRAM via our new Triton kernels (no accuracy loss). Train gpt-oss locally on 12.8GB VRAM. In collab with @hf.co, Unsloth trains DeepSeek, Qwen3, GLM faster. Repo: github.com/unslothai/un... Blog: unsloth.ai/docs/new/fas...

✨ New 3D pose estimation method from my lab! #FMPose3D allows for monocular (i.e. single camera) 2D➡️3D 🔥 Led by Ti Wang & w/ Xiaohang Yu #FMPose3D is SOTA on human & animal 3D benchmarks, & will be integrated into @deeplabcut.bsky.social ⬇️ 📝 arxiv.org/abs/2602.05755 ➡️ xiu-cs.github.io/FMPose3D/

Google DeepMind's Mining Generalizable Activation Functions They argue that evolutionary search is a powerful framework for discovering new activation functions, showing that LLM-driven pipelines like AlphaEvolve replace manually designed search spaces with flexible,

the fact that early-stage automated vehicles that still occasionally require remote operation assistance are like several orders of magnitude safer than the average American driver should be the real wakeup call here, not the idea that these vehicles still sometimes need remote operation assistance

Pydantic's Monty A minimal, secure Python interpreter written in Rust for use by AI. github.com/pydantic/monty

RayRoPE: Projective Ray Positional Encoding for Multi-view Attention Yu Wu, Minsik Jeon, Jen-Hao Rick Chang, @onceltuzel.bsky.social Shubham Tulsiani tl;dr: even more projective geometry based PE, if you know your camera pose. arxiv.org/abs/2601.15275

They propose a new paradigm called Drifting Models, which evolve the pushforward distribution during training and naturally admit one-step inference. We introduce a drifting field that governs the sample movement and achieves equilibrium when the distributions match.

In case it wasn’t ALREADY blatantly clear!

They develop TinyLoRA, a new ft method. with TinyLoRA + RL, models learn well with dozens or hundreds of params. For example, they use only 13 parameters to train 7B Qwen model from 76 to 91% on GSM8K. "Learning to Reason in 13 Parameters" Paper: arxiv.org/abs/2602.04118