Video yükleniyor...

Video Yüklenemedi

Bu video yüklenirken bir sorun oluştu. Bu geçici bir ağ sorunundan kaynaklanıyor olabilir veya video kullanılamıyor olabilir.

Ana Sayfaya Dön

Pretrained ViTs see the world in rich, dense detail. Most policies pool it to a single vector before acting, discarding most of it. We introduce Patch Policy: a minimal architectural extension that enables transformer-based policies to consume dense tokens directly, no billion-param VLM required. It outperforms a fine-tuned 7B... show more

Gaoyue Zhou

1,181 subscribers

54,713 görüntüleme • 1 ay önce •via X (Twitter)

Bilim & Teknoloji Eğitim

Anya Rossi• Live Now

Private livecam show

0 Yorum

Yorum bulunmuyor

Orijinal gönderinin yorumları burada görünecek

Benzer Videolar

By unifying prediction and control, NavWAM acts as a closed-loop policy out of the box. In evaluations, it outperforms planning-based world models without needing test-time action search, matches a much larger 7B VLA policy, and transfers successfully to real mobile robots.

By unifying prediction and control, NavWAM acts as a closed-loop policy out of the box. In evaluations, it outperforms planning-based world models without needing test-time action search, matches a much larger 7B VLA policy, and transfers successfully to real mobile robots.

Robots Digest 🤖

11,076 görüntüleme • 2 ay önce

It is really hard to get robot policies that are both precise (small margins for error) and general (robust to env variations). We just released ViSk, where skin sensing is used to train fine-grained policies with ~1 hour of data. Below is a single-take video.

It is really hard to get robot policies that are both precise (small margins for error) and general (robust to env variations). We just released ViSk, where skin sensing is used to train fine-grained policies with ~1 hour of data. Below is a single-take video.

Lerrel Pinto

17,169 görüntüleme • 1 yıl önce

Instead of asking a VLM to output progress, it reads the model’s internal belief directly from token logits. No in-context learning. No fine-tuning. No reward training. 📈 We introduce: TOPReward, a zero-shot reward modeling approach for robotics using token probabilities from pretrained video VLMs. The simplest way of doing reward modelling for robotics! Project: 🧵👇

Instead of asking a VLM to output progress, it reads the model’s internal belief directly from token logits. No in-context learning. No fine-tuning. No reward training. 📈 We introduce: TOPReward, a zero-shot reward modeling approach for robotics using token probabilities from pretrained video VLMs. The simplest way of doing reward modelling for robotics! Project: 🧵👇

Jiafei Duan

107,926 görüntüleme • 6 ay önce

Introducing Tether 🪢, a fun little idea to scale data by having our robot “play” in the real world for over 24 hours, throughout the day and overnight—improving policies from zero to mastery with minimal supervision! But play is messy, with out-of-distribution scenarios that are hard to anticipate. To perform autonomous functional play in the real world, from just a handful of demos, we propose a highly robust few-shot imitation method that warps demo trajectories using visual correspondences. Then, continuously running it within a multi-task VLM-guided cycle, we generate a data stream that produces 1000+ expert-level demos. This generated data is finally funneled downstream to train imitation learning policies, which improve from zero to near-perfect success rates. We’ll be presenting Tether at #ICLR2026 in just a few weeks! But before that, deep dive with me… 🧵

Introducing Tether 🪢, a fun little idea to scale data by having our robot “play” in the real world for over 24 hours, throughout the day and overnight—improving policies from zero to mastery with minimal supervision! But play is messy, with out-of-distribution scenarios that are hard to anticipate. To perform autonomous functional play in the real world, from just a handful of demos, we propose a highly robust few-shot imitation method that warps demo trajectories using visual correspondences. Then, continuously running it within a multi-task VLM-guided cycle, we generate a data stream that produces 1000+ expert-level demos. This generated data is finally funneled downstream to train imitation learning policies, which improve from zero to near-perfect success rates. We’ll be presenting Tether at #ICLR2026 in just a few weeks! But before that, deep dive with me… 🧵

Will Liang

45,671 görüntüleme • 5 ay önce

The One Child Policy wasn’t implemented for the reason most people think. It’s often framed as a response to overpopulation fears. But another way to view it is as an attempt to fundamentally reshape the family unit itself. By limiting births it disrupted one of humanity’s most enduring structures, the multi generational multi child family. Over time it weakened those deep family networks that had existed for centuries. It did not fully break the family but it significantly altered its foundation. And it remains one of the most consequential social engineering policies under CCP rule. Heritage Foundation

The One Child Policy wasn’t implemented for the reason most people think. It’s often framed as a response to overpopulation fears. But another way to view it is as an attempt to fundamentally reshape the family unit itself. By limiting births it disrupted one of humanity’s most enduring structures, the multi generational multi child family. Over time it weakened those deep family networks that had existed for centuries. It did not fully break the family but it significantly altered its foundation. And it remains one of the most consequential social engineering policies under CCP rule. Heritage Foundation

Jan Jekielek

41,490 görüntüleme • 4 ay önce

$A $351 MINI PC IS RUNNING 26-BILLION-PARAMETER AI MODELS AT 20 TOKENS/SEC AND HERMES AGENT ON TOP OF IT This is the Minisforum UM790 Pro. AMD Ryzen 9 7940HS, Radeon 780M iGPU, 48GB DDR5. The BIOS reports the GPU has 4GB of VRAM Here's the part people get wrong. The 780M has no dedicated VRAM at all it borrows from system RAM. Vulkan ignores the BIOS number and reads the full 48GB pool directly That's the whole trick. 21+ GB allocated to model weights on a "4GB" GPU. Because the models are MoE, only 4 billion of the parameters activate per token a fraction of the reads a dense model needs. That's why 20 tok/s works here Gemma 4 26B MoE holds 19.5 tok/s with 196K context. Qwen3.5-35B-A3B holds 20.8. Nemotron Cascade 2 clears 24.8. A dense 31B, by contrast, drops to 4 tok/s it reads the entire model every token, no way around it On top: Hermes Agent, full agentic workflows terminal, file ops, web, 40+ tools against local models only. No API keys. The wall between you and a usable local agent used to be a GPU you couldn't afford. Now it's a BIOS setting most people never check Bookmark this & Try it yourself ↓$

A $351 MINI PC IS RUNNING 26-BILLION-PARAMETER AI MODELS AT 20 TOKENS/SEC AND HERMES AGENT ON TOP OF IT This is the Minisforum UM790 Pro. AMD Ryzen 9 7940HS, Radeon 780M iGPU, 48GB DDR5. The BIOS reports the GPU has 4GB of VRAM Here's the part people get wrong. The 780M has no dedicated VRAM at all it borrows from system RAM. Vulkan ignores the BIOS number and reads the full 48GB pool directly That's the whole trick. 21+ GB allocated to model weights on a "4GB" GPU. Because the models are MoE, only 4 billion of the parameters activate per token a fraction of the reads a dense model needs. That's why 20 tok/s works here Gemma 4 26B MoE holds 19.5 tok/s with 196K context. Qwen3.5-35B-A3B holds 20.8. Nemotron Cascade 2 clears 24.8. A dense 31B, by contrast, drops to 4 tok/s it reads the entire model every token, no way around it On top: Hermes Agent, full agentic workflows terminal, file ops, web, 40+ tools against local models only. No API keys. The wall between you and a usable local agent used to be a GPU you couldn't afford. Now it's a BIOS setting most people never check Bookmark this & Try it yourself ↓

slash1s

35,742 görüntüleme • 2 ay önce

Most of us judge a book by its cover before we've read a single page. That makes cover design a pretty special craft—and one of the many reasons it was so fun to have a Cheeky Chat with Pablo about his work designing the covers of Stripe Press books!

Most of us judge a book by its cover before we've read a single page. That makes cover design a pretty special craft—and one of the many reasons it was so fun to have a Cheeky Chat with Pablo about his work designing the covers of Stripe Press books!

Katie Dill

15,994 görüntüleme • 1 ay önce

Meet VLA-JEPA, the first world model policy in LeRobot 🌍 Most VLAs just learn observation → action. VLA-JEPA also learns action-relevant dynamics: during training, a JEPA world model has to predict upcoming frames in latent space from the model's own actions. Then, at inference, the world model disappears entirely, leaving a standard, fast VLA: Qwen backbone, action head, nothing extra. In our demo it was fine-tuned on just 13 examples and ran in real time on an NVIDIA DGX Spark. On a consumer RTX 3080, it hits 10Hz using under 6GB of VRAM. World-model supervision during training, zero extra cost at inference. That's the whole trick.

Meet VLA-JEPA, the first world model policy in LeRobot 🌍 Most VLAs just learn observation → action. VLA-JEPA also learns action-relevant dynamics: during training, a JEPA world model has to predict upcoming frames in latent space from the model's own actions. Then, at inference, the world model disappears entirely, leaving a standard, fast VLA: Qwen backbone, action head, nothing extra. In our demo it was fine-tuned on just 13 examples and ran in real time on an NVIDIA DGX Spark. On a consumer RTX 3080, it hits 10Hz using under 6GB of VRAM. World-model supervision during training, zero extra cost at inference. That's the whole trick.

LeRobot

42,389 görüntüleme • 1 ay önce

This is a 20-minute unedited on/off-gantry run from today. What's special about it is that no gains were tuned on the robot before we started running the policy. This is the DVBF-based motion policy we've been working on, which is turning out to be quite special. The goal of using DVBF is to ensure policies can be deployed on many robots at once without manual gain adjustment before the policy runs reliably on each robot.

This is a 20-minute unedited on/off-gantry run from today. What's special about it is that no gains were tuned on the robot before we started running the policy. This is the DVBF-based motion policy we've been working on, which is turning out to be quite special. The goal of using DVBF is to ensure policies can be deployed on many robots at once without manual gain adjustment before the policy runs reliably on each robot.

Sankaet

36,570 görüntüleme • 9 ay önce

Based on the latest features of #TermiX, seamless integration with Anthropic and BNB Chain enables a variety of customized functions. Unlock the potential of #DeFAI ! Watch our demo & tutorial video to see how easy it is to launch BEP20 tokens, swap, transfer, and check wallets. Exciting news: In the coming days, we will release a series of demo videos to introduce more features! Join us and let's build the future of DeFAI together! Start now 👇 Check how it works → Fork it → Build your own GitHub: Learn more: #DeFAI #BNBHack #BNBChain

Based on the latest features of #TermiX, seamless integration with Anthropic and BNB Chain enables a variety of customized functions. Unlock the potential of #DeFAI ! Watch our demo & tutorial video to see how easy it is to launch BEP20 tokens, swap, transfer, and check wallets. Exciting news: In the coming days, we will release a series of demo videos to introduce more features! Join us and let's build the future of DeFAI together! Start now 👇 Check how it works → Fork it → Build your own GitHub: Learn more: #DeFAI #BNBHack #BNBChain

TermiX

31,110 görüntüleme • 1 yıl önce

Every major decision in human history has been made the same way. Guess. Execute. Hope it doesn’t break the world. That era is ending. Simile CEO Joon Sung Park just revealed what the most serious AI researchers are actually building toward. Park: “Our goal eventually is to ask ourselves what would it mean if we can create simulations of 8 billion people? The entire Earth.” Not a metaphor. A literal, bottom-up digital replica of civilization. We assumed the ultimate goal of AI was to predict the future. Park says that’s thinking too small. He points to The Oracle in The Matrix to explain the real paradigm shift. Park: “You’re not necessarily here to make a decision. You’ve already made that decision. You’re here to understand the decision and why you made that.” AI isn’t a crystal ball to tell us what happens next. It’s an x-ray to show us why. Think about how civilization actually operates right now. We guess. We pass a law. We shift an economic policy. We launch a product. Then we wait a decade to find out if it broke the world. The simulation ends that. Before a government passes a policy, they run a dry-run on 8 billion digital agents. Before a company ships a product, they trace the exact moment the market breaks. Park: “Creating these kind of bottom up simulations where we can actually go back and trace through the audit logs of how society might unfold is an amazing way to gain that interpretability layer of our reality.” Let that phrase land. The interpretability layer of our reality. We spent the last twenty years extracting data from the physical world. We are about to spend the next twenty years using that data to build a mirror world. One where every policy gets stress-tested before it’s enacted. Every product before it ships. Every decision before its consequences become irreversible. Park: “If you want to shape new policies, new products that would actually serve the people, the best way is actually to understand those people and communicating with them at a scalable way to basically create those policies.” We are not moving from ignorance to prediction. We are moving from blind execution to total societal foresight. And once that capability exists, everything changes. Wars modeled before they’re fought. Economies stress-tested before they collapse. Policies run through billions of digital lives before a single real one is affected. Reality becomes the place we execute what the mirror world already proved would work. The ultimate application of artificial intelligence was never about replacing human judgment. It was always about perfecting it. We aren’t just predicting the future anymore. We are rehearsing it. And when you can rehearse the future before you live it, the most catastrophic word in human history finally becomes extinct. That word is surprise.

Every major decision in human history has been made the same way. Guess. Execute. Hope it doesn’t break the world. That era is ending. Simile CEO Joon Sung Park just revealed what the most serious AI researchers are actually building toward. Park: “Our goal eventually is to ask ourselves what would it mean if we can create simulations of 8 billion people? The entire Earth.” Not a metaphor. A literal, bottom-up digital replica of civilization. We assumed the ultimate goal of AI was to predict the future. Park says that’s thinking too small. He points to The Oracle in The Matrix to explain the real paradigm shift. Park: “You’re not necessarily here to make a decision. You’ve already made that decision. You’re here to understand the decision and why you made that.” AI isn’t a crystal ball to tell us what happens next. It’s an x-ray to show us why. Think about how civilization actually operates right now. We guess. We pass a law. We shift an economic policy. We launch a product. Then we wait a decade to find out if it broke the world. The simulation ends that. Before a government passes a policy, they run a dry-run on 8 billion digital agents. Before a company ships a product, they trace the exact moment the market breaks. Park: “Creating these kind of bottom up simulations where we can actually go back and trace through the audit logs of how society might unfold is an amazing way to gain that interpretability layer of our reality.” Let that phrase land. The interpretability layer of our reality. We spent the last twenty years extracting data from the physical world. We are about to spend the next twenty years using that data to build a mirror world. One where every policy gets stress-tested before it’s enacted. Every product before it ships. Every decision before its consequences become irreversible. Park: “If you want to shape new policies, new products that would actually serve the people, the best way is actually to understand those people and communicating with them at a scalable way to basically create those policies.” We are not moving from ignorance to prediction. We are moving from blind execution to total societal foresight. And once that capability exists, everything changes. Wars modeled before they’re fought. Economies stress-tested before they collapse. Policies run through billions of digital lives before a single real one is affected. Reality becomes the place we execute what the mirror world already proved would work. The ultimate application of artificial intelligence was never about replacing human judgment. It was always about perfecting it. We aren’t just predicting the future anymore. We are rehearsing it. And when you can rehearse the future before you live it, the most catastrophic word in human history finally becomes extinct. That word is surprise.

Dustin

33,220 görüntüleme • 6 ay önce

“You have to mine it out of the ground. You then turn it into a gas for enrichment. You then enrich it — which is essentially a refining process. So enrichment is separating the element of uranium into its two or a couple of different types based on its isotope, which is how many neutrons it has. It’s really a refining, separation process. A lot of people hear ‘enrichment’ and they think there’s going to be a lot of radiation or dangerous chemical reactions. It’s actually just separation." “Cold War era. We were the world leader in enrichment. We had these plants spread out throughout the country that used an old technique that was expensive. It required a lot of energy to do it. But we had the most capacity. Then we built up a lot of enriched uranium that we stockpiled, and those stockpiles..." Scott Nolan

“You have to mine it out of the ground. You then turn it into a gas for enrichment. You then enrich it — which is essentially a refining process. So enrichment is separating the element of uranium into its two or a couple of different types based on its isotope, which is how many neutrons it has. It’s really a refining, separation process. A lot of people hear ‘enrichment’ and they think there’s going to be a lot of radiation or dangerous chemical reactions. It’s actually just separation." “Cold War era. We were the world leader in enrichment. We had these plants spread out throughout the country that used an old technique that was expensive. It required a lot of energy to do it. But we had the most capacity. Then we built up a lot of enriched uranium that we stockpiled, and those stockpiles..." Scott Nolan

Shawn Ryan

104,536 görüntüleme • 1 yıl önce

ALL THIS CONSTRUCTION OF THE VISIBLE WORLD IS AT ITS ROOTS DESIGNED IN THE INVISIBLE WORLD Before language, before numbers, before thought, there was form. Not arbitrary, but precise. Not invented, but revealed. Sacred Geometry is the blueprint beneath all becoming—the hidden grammar of light, sound, and matter. It is the spiral of galaxies and the whorl of a shell. It is the vesica that births light, The triangle that gives rise to stability, The square that anchors space, And the pentagram that mirrors life’s golden proportion. Sacred Geometry does not merely decorate creation, it structures it. It governs the growth of leaves, the pulse of waves, the rhythm of your breath. It reveals that harmony is not a concept, it is a code. A flower blooms not by chance, But by a silent obedience to phi. A crystal aligns not at random, But by listening to the quiet song of symmetry. This is not mysticism, it is order. This is not dogma, it is design. Sacred Geometry is where the visible and the invisible meet: Where math becomes art, Where number becomes movement, Where space becomes spirit. To study it is not only to learn, it is to remember. That beauty is not a luxury. It is the signature of coherence. And when we align with it, We don’t just build structures. We become them. Everything that has form in this dimension is made up in its essence of sacred geometry. All this construction of the visible world is at its root designed in the invisible world.

ALL THIS CONSTRUCTION OF THE VISIBLE WORLD IS AT ITS ROOTS DESIGNED IN THE INVISIBLE WORLD Before language, before numbers, before thought, there was form. Not arbitrary, but precise. Not invented, but revealed. Sacred Geometry is the blueprint beneath all becoming—the hidden grammar of light, sound, and matter. It is the spiral of galaxies and the whorl of a shell. It is the vesica that births light, The triangle that gives rise to stability, The square that anchors space, And the pentagram that mirrors life’s golden proportion. Sacred Geometry does not merely decorate creation, it structures it. It governs the growth of leaves, the pulse of waves, the rhythm of your breath. It reveals that harmony is not a concept, it is a code. A flower blooms not by chance, But by a silent obedience to phi. A crystal aligns not at random, But by listening to the quiet song of symmetry. This is not mysticism, it is order. This is not dogma, it is design. Sacred Geometry is where the visible and the invisible meet: Where math becomes art, Where number becomes movement, Where space becomes spirit. To study it is not only to learn, it is to remember. That beauty is not a luxury. It is the signature of coherence. And when we align with it, We don’t just build structures. We become them. Everything that has form in this dimension is made up in its essence of sacred geometry. All this construction of the visible world is at its root designed in the invisible world.

🧬Maxpein🧬

13,405 görüntüleme • 9 gün önce

Historically, RL policies for robots have been trained in synthetic, untextured environments, limiting perceptive policies to depth images where the sim-to-real gap is manageable. RGB has always had more potential, but leveraging it to train policies in simulation remained an open problem. Partnering with Niantic Spatial 🌎 and NVIDIA Robotics, we built a pipeline that addresses exactly that. We can now scan a real deployment site with off-the-shelf hardware, reconstruct it into a photorealistic Gaussian splat, and run massively parallel RL training. The policies trained in our Gym environment then transfer zero-shot to the real robot and environments they were trained for. This enables faster deployment of more capable and robust policies for the end user. The new resulting capabilities are a big step towards solving sim-to-real and also apply well beyond navigation. Read the full technical breakdown on our blog; link in the comments. #HumanoidRobots #Flexion #NianticSpatial #NVIDIA

Historically, RL policies for robots have been trained in synthetic, untextured environments, limiting perceptive policies to depth images where the sim-to-real gap is manageable. RGB has always had more potential, but leveraging it to train policies in simulation remained an open problem. Partnering with Niantic Spatial 🌎 and NVIDIA Robotics, we built a pipeline that addresses exactly that. We can now scan a real deployment site with off-the-shelf hardware, reconstruct it into a photorealistic Gaussian splat, and run massively parallel RL training. The policies trained in our Gym environment then transfer zero-shot to the real robot and environments they were trained for. This enables faster deployment of more capable and robust policies for the end user. The new resulting capabilities are a big step towards solving sim-to-real and also apply well beyond navigation. Read the full technical breakdown on our blog; link in the comments. #HumanoidRobots #Flexion #NianticSpatial #NVIDIA

Flexion

71,981 görüntüleme • 1 ay önce

One Nation has the most detailed policies published out of any party. Just ask the opposition or the government what their immigration policy actually is and you won't get an answer. One Nation is clear. Our policies have always been available at - have a look and tell me what you think. To be clear, I won't just be offering a "coalition" with anyone. If the voters believe in One Nation and choose to send more of us to Parliament, we would be responsible and guarantee supply to the government of the day so that the country keeps running. While that happens, we will keep pushing One Nation policies and checking the details of every single bill for whether it is in the best interest of the country.

One Nation has the most detailed policies published out of any party. Just ask the opposition or the government what their immigration policy actually is and you won't get an answer. One Nation is clear. Our policies have always been available at - have a look and tell me what you think. To be clear, I won't just be offering a "coalition" with anyone. If the voters believe in One Nation and choose to send more of us to Parliament, we would be responsible and guarantee supply to the government of the day so that the country keeps running. While that happens, we will keep pushing One Nation policies and checking the details of every single bill for whether it is in the best interest of the country.

Pauline Hanson 🇦🇺

37,120 görüntüleme • 6 ay önce

Footage of the RS-28 'Sarmat' (Sarmatian) ICBM test The liquid-fueled heavy RS-28 is designed to replace the world’s most powerful silo-based missile, the RS-20V Voevoda (SS-18 Satan according to NATO classification). It is believed to be to be able to carry up to ten heavy warheads with multiple re-entry capability. It is capable of delivering a payload four times bigger than that of any Western missile. It also has a range of over 35,000 kilometers and is twice as precise as its predecessor.

Footage of the RS-28 'Sarmat' (Sarmatian) ICBM test The liquid-fueled heavy RS-28 is designed to replace the world’s most powerful silo-based missile, the RS-20V Voevoda (SS-18 Satan according to NATO classification). It is believed to be to be able to carry up to ten heavy warheads with multiple re-entry capability. It is capable of delivering a payload four times bigger than that of any Western missile. It also has a range of over 35,000 kilometers and is twice as precise as its predecessor.

Hawkeye1812Z

15,348 görüntüleme • 3 ay önce

Kochi is one the most unique cities in our country A city defined by its waterways, so no wonder it hosts a Water Metro. Not the first water-based transport but the first to cleverly name it a ‘Metro’ and capture the imagination of both its citizens and tourists. And a good looking, comfortable and emission-free boat to top it all. We need more of our cities to make such breakthroughs in citizen friendly infrastructure. Great job, Kochi!

Kochi is one the most unique cities in our country A city defined by its waterways, so no wonder it hosts a Water Metro. Not the first water-based transport but the first to cleverly name it a ‘Metro’ and capture the imagination of both its citizens and tourists. And a good looking, comfortable and emission-free boat to top it all. We need more of our cities to make such breakthroughs in citizen friendly infrastructure. Great job, Kochi!

anand mahindra

332,359 görüntüleme • 1 yıl önce