How robust can model predictive control be if we... can solve each trajectory optimization to global optimality? On the contact-rich push-T problem, we show that model-based global optimization is so robust that it never fails, even if the model is not even correct! We achieve global optimality via sparse Moment and SOS relaxations. -- Yes, we managed to solve SDPs online on a robot. Amazing work by Shucheng Kang and Guorui Liu.show more

Malaria is a complex & unforgiving disease, but one... that we can end if we work together. Fredros Okumu & JenHatmaker discuss the global impacts of malaria and how we can advance progress against it on this episode of “Make Me Care About…”show more

Evaluating policies on a real robot can be painful.... Can we use a world model to get a rough estimate of how good a policy is? Checkout "Evaluating Robot Policies in a World Model". Paper: Demo: Code:show more

Norman Powell on the trade deadline: “We just focus... on the moment and what we can control and if moves do happen we support and if they don’t — we lock in with the group that we have and go out there and try to get wins.” (via Zachary Weinberger)show more

🚨How do LLMs acquire human values?🤔 We often point... to preference optimization. However, in our new work, we trace how and when model values shift during post-training and uncover surprising dynamics. We ask: How do data, algorithms, and their interaction shape model values?🧵show more

#Habs Mike Matheson on officiating "That is a good... question, and we are wondering ourselves. But it is out of our control, so it is not something to worry about. We need to keep playing and keep focusing on the things we can control." #GoHabsGo #NHL #Hockeyshow more

“Last night BTS’ fans ever since we announced it.... The ARMY have already raised over $10k just like that just by going online & fundraising. We didn’t even ask them to do so. The ARMY are so amazing.” - Hugh Evans (Founder & CEO of Global Citizen) via Billboard Newsshow more

Mark Carney, "Canada can’t solve all the world’s problems"... "But we can show that another way is possible" "That the arc of history isn’t destined to be warped towards authoritarianism and exclusion" "It can still bend towards progress and justice"show more

we're never getting a model that can do UX,... are we?show more

Judging by this pattern... Are y'all telling me that... if America leaves the UN... we could literally achieve global peace?show more

This is amazing! Someone trained a 1.5B model (with... GRPO) to solve mazes! model:show more

Maduro: The petroleum stream deformed everything. We did not... choose to have a model that depended on petroleum. We did not decide that. When I was born in the year ’62, they had already imposed the model of capitalism that depended on petroleum, and we were a petroleum colony of the United States.show more

Blackrock CEO Larry Fink moments ago: "One of the... main foundations of the US is that we are a global stabilizer. Obviously in the last week we have not been a global stabilizer. Is this going to persist? Are there going to be new global stabilizers in the world?"show more

I put a lens after the single slit and... see how aperture size affects the focusing of light, and we can observe that a larger aperture can achieve a better focus. That got me thinking, if we want to pinpoint a problem, it's better to view it from multiple perspectives. #opticsshow more

ICML 2026: Latent Reasoning in TRMs is Secretly a... Policy Improvement Operator Why does recursive reasoning, especially latent reasoning, actually work? The theory is still young, and even mechanistic explanations are limited. We close part of this gap by showing that latent reasoning is secretly doing policy improvement. Each recursion pushes the model steadily toward the target. Based on this view, we propose an algorithm that boosts learning and inference efficiency by up to 18x.show more

Model-Free Reinforcement Learning (MFRL) has been alluring, especially with... supercharged compute with physics on GPU. However, the methods use 0-th order gradients, and are often not the best optimizers. Can we do better than PPO in continuous control for robotics? Turns out yes! 🥳 tl;dr: Faster, better RL than PPO in continuous control 💪 The answer lies in using more information from the simulation. We are juicing the simulation on GPU as it is, why not use it for gradients as well? This has been a driving question in a series of our works. We first studied this problem in ICLR 2022 paper on Short Horizon Actor Critic Naive gradient based methods are stuck in local minima and have exploding/vanishing gradients. SHAC solved this problem truncated rollouts and model based value estimation, where the model is Differentiable Sim. This boosted sample efficiency and wall-clock time immensely especially in high dimensional systems such as humanoids Yet, given enough compute PPO often caught up. Our follow up paper on on Adaptive Horizon Actor Critic at ICML 2024 discovers the cause and provides a fix. However, we find that even when given ground-truth dynamics, not all gradients are useful due to sample error. 1st-Order Model-Based Reinforcement Learning methods employing differentiable simulation provide gradients with reduced variance but are susceptible to bias in scenarios involving stiff dynamics, such as physical contact. We find that back-propagating through contact and long trajectories drastically reduces gradient accuracy. Using this insight, we propose AHAC to dynamically adapt its roll-out horizon to avoid differentiating through stiff contact. AHAC is a first-order model-based RL algorithm that learns high-dimensional tasks in minutes (wall clock) and outperforms PPO by 40%, even in the limit of data provided to PPO. This work is led by Ignat Georgiev alongside Krishnan Srinivasan, Jie Xu, Eric Heiden and ample assistance from warp team at NVIDIA Robotics (Miles Macklin)show more

Excited to release Minimalist Compliance Control! We achieve robust,... compliant robot interaction across robot arms, dexterous hands, and humanoids, with NO force sensors or learning. If you’re wondering what remains, please see the thread below😉 Website:show more

Each of these homes were built in a factory... in 2 hours, and assembled on-site in 3 hours. This is the type of innovation Canada is embracing in order to build the homes we need, faster. Canadian industry and Canadian workers have the creativity and ambition required to meet this moment. By working together, we can, and we will, solve the housing crisis.show more

Rubio on Syria: We like the trajectory. We have... to keep it on that trajectory. We've got good agreements in place. The key now is implementation, and we'll be very involved in that regard.show more

Excited to introduce Diffusion Augmented Agents (DAAGs)✨. We give... an agent control of a diffusion model, so it can create its own *synthetic experience*.🪄 The result is a lifelong agent that can learn new reward detectors and policies, much more efficiently. Here's how. 👇show more