Video yükleniyor...

Video Yüklenemedi

Ana Sayfaya Dön

This guy connected a computer vision model to dual robotic manipulators on his desk and the system now folds shirts in 47 seconds per garment without any human intervention after loading Automated laundry folding is one of those problems that sounds trivial until you realize fabric has no rigid...

browomo's profile picture

Blaze

9,733 subscribers

24,799 görüntüleme • 4 ay önce •via X (Twitter)

LIVE

Anya Rossi• Live Now

Private livecam show

0 Yorum

Yorum bulunmuyor

Orijinal gönderinin yorumları burada görünecek

Benzer Videolar

Robotics is almost solved! Pun intended. Sunday published a "Solve" standard. They declared the scope (garment types, lighting, surfaces, starting configurations) and adaptation cost (zero—no fine-tuning per home) before evaluating. The same model weights run everywhere. This might be the differentiation between a good demo and reliable work of robots. They presented ACT-2, a robot foundation model that folds laundry at 99.1% success across unseen homes. Different houses, lighting, bed types, garment sizes. According to them scaling pre-training closes the generalization gap. Train on massive, high-quality human data. When you do that right, what you learn in-house actually transfers to unseen homes. Even wilder is that the model learns new folding techniques from a single example and generalizes them to new environments. Memo goes to homes later this year. Would you adopt this cutie? 🥹 Read more here: Great work Tony Zhao and team! 🧡 ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →
1:21

Robotics is almost solved! Pun intended. Sunday published a "Solve" standard. They declared the scope (garment types, lighting, surfaces, starting configurations) and adaptation cost (zero—no fine-tuning per home) before evaluating. The same model weights run everywhere. This might be the differentiation between a good demo and reliable work of robots. They presented ACT-2, a robot foundation model that folds laundry at 99.1% success across unseen homes. Different houses, lighting, bed types, garment sizes. According to them scaling pre-training closes the generalization gap. Train on massive, high-quality human data. When you do that right, what you learn in-house actually transfers to unseen homes. Even wilder is that the model learns new folding techniques from a single example and generalizes them to new environments. Memo goes to homes later this year. Would you adopt this cutie? 🥹 Read more here: Great work Tony Zhao and team! 🧡 ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

Lukas Ziegler

37,023 görüntüleme • 3 ay önce

A robot that drives straight up a wall! 🧱 VertiGo is a robot built by Disney Research with ETH Zürich. Most wall-climbing robots hold on. Suction cups, magnets, dry adhesives. All of them need a surface that cooperates, and none of them solve the genuinely hard part, which is getting from the floor onto the wall in the first place. Two tiltable propellers generate thrust into the wall, pressing four wheels against it hard enough to drive. Each propeller has two degrees of freedom for aiming that thrust, and one pair of wheels steers. The rear propeller pushes the robot toward the wall while the front propeller pushes upward, and the combination flips it onto the vertical surface. One propeller can hold you on a wall. It takes two to get you there. This is from the end of 2015, which is worth saying out loud. A decade later it's still an awesome project in robotics! :) ~~~ ♻️ Join the weekly robotics newsletter, and never miss any news →
0:33

A robot that drives straight up a wall! 🧱 VertiGo is a robot built by Disney Research with ETH Zürich. Most wall-climbing robots hold on. Suction cups, magnets, dry adhesives. All of them need a surface that cooperates, and none of them solve the genuinely hard part, which is getting from the floor onto the wall in the first place. Two tiltable propellers generate thrust into the wall, pressing four wheels against it hard enough to drive. Each propeller has two degrees of freedom for aiming that thrust, and one pair of wheels steers. The rear propeller pushes the robot toward the wall while the front propeller pushes upward, and the combination flips it onto the vertical surface. One propeller can hold you on a wall. It takes two to get you there. This is from the end of 2015, which is worth saying out loud. A decade later it's still an awesome project in robotics! :) ~~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

Lukas Ziegler

121,319 görüntüleme • 26 gün önce

How to undo hours of sitting in 60 seconds: Physical therapist Dr. John Rusin uses this tool with clients as a reset between long stretches at a desk or on a phone, and it requires no equipment. Get into what he calls a Superman position: 1. Stand up, put the phone down, take one step away from the desk 2. Extend the upper back, open the chest, lift through the sternum 3. Arms out and back, rotated so the palms turn outward 4. Pull the shoulder blades down and back 5. Look up, letting the neck extend 6. Take a deep breath in 7. Squeeze the glutes hard (Rusin calls this the most important step) Hold, release, and run it a couple of times. That's the whole thing. Why it works is mostly what it's the opposite of. Sitting flexes the spine, rounds the shoulders, drops the head forward, and shuts the glutes off. Every piece of this position reverses one of those. Rusin's framing is that it resets both the mechanical system and the nervous system in under a minute. The practical result is the same: you stand differently afterward, and the next hour at the desk starts from a better place. The best mobility tool is usually the one you'll actually do at 2pm on a Tuesday. This one takes a minute.
0:47

How to undo hours of sitting in 60 seconds: Physical therapist Dr. John Rusin uses this tool with clients as a reset between long stretches at a desk or on a phone, and it requires no equipment. Get into what he calls a Superman position: 1. Stand up, put the phone down, take one step away from the desk 2. Extend the upper back, open the chest, lift through the sternum 3. Arms out and back, rotated so the palms turn outward 4. Pull the shoulder blades down and back 5. Look up, letting the neck extend 6. Take a deep breath in 7. Squeeze the glutes hard (Rusin calls this the most important step) Hold, release, and run it a couple of times. That's the whole thing. Why it works is mostly what it's the opposite of. Sitting flexes the spine, rounds the shoulders, drops the head forward, and shuts the glutes off. Every piece of this position reverses one of those. Rusin's framing is that it resets both the mechanical system and the nervous system in under a minute. The practical result is the same: you stand differently afterward, and the next hour at the desk starts from a better place. The best mobility tool is usually the one you'll actually do at 2pm on a Tuesday. This one takes a minute.

Andy Galpin, PhD

70,541 görüntüleme • 28 gün önce

WTF, GROK BOT GALAXY STARTS TOMORROW AND THEY'RE BUILDING AN ENTIRE COMPANY FROM ZERO, LIVE ON STAGE. I'M NOT WAITING THREE DAYS. THE GROK BOT CONTENT SYSTEM IS ALREADY BUILT – AND YOU CAN HAVE IT RUNNING TONIGHT. Twelve sessions across three days. 101, engineering, product, founders, sales engineering, sales, SDRs, support, marketing ops, post-sales, marketing, final showcase. A company assembled in public, workflow by workflow. → Watch all of it. Register if you haven't. It's going to be good. But I'm not bringing a concept to this. I'm bringing infrastructure that already runs. The Grok Bot Content System is finished and in use. Seven agents, one narrow job each, a fixed handoff order. A chief of staff that routes. A researcher that sources and verifies. A writer that works in my voice. A visualiser that renders in my style. An analyst that reads what performed and tells the other two what to change. A scheduler that owns timing. A publisher that ships and confirms. Zero code. Seven prompts. Ten minutes to stand up. Then it runs without me in the loop. Here's why that matters if you're trying to make money from content: content isn't one job, it's six. One person doing all six ships once a week and quits by month four. And nothing on any platform – audience, sponsors, a product, a token community – compounds off one post a week. The thing standing between most people and revenue was never the idea. It was throughput and consistency, and that is exactly the part this removes. I'm not going to tell you what you'll earn. I'm handing you the desk, not a number. What you put through it is yours. Nothing publishes without my approval – that gate I kept on purpose. I wrote the whole build up in one article below ↓
0:30

WTF, GROK BOT GALAXY STARTS TOMORROW AND THEY'RE BUILDING AN ENTIRE COMPANY FROM ZERO, LIVE ON STAGE. I'M NOT WAITING THREE DAYS. THE GROK BOT CONTENT SYSTEM IS ALREADY BUILT – AND YOU CAN HAVE IT RUNNING TONIGHT. Twelve sessions across three days. 101, engineering, product, founders, sales engineering, sales, SDRs, support, marketing ops, post-sales, marketing, final showcase. A company assembled in public, workflow by workflow. → Watch all of it. Register if you haven't. It's going to be good. But I'm not bringing a concept to this. I'm bringing infrastructure that already runs. The Grok Bot Content System is finished and in use. Seven agents, one narrow job each, a fixed handoff order. A chief of staff that routes. A researcher that sources and verifies. A writer that works in my voice. A visualiser that renders in my style. An analyst that reads what performed and tells the other two what to change. A scheduler that owns timing. A publisher that ships and confirms. Zero code. Seven prompts. Ten minutes to stand up. Then it runs without me in the loop. Here's why that matters if you're trying to make money from content: content isn't one job, it's six. One person doing all six ships once a week and quits by month four. And nothing on any platform – audience, sponsors, a product, a token community – compounds off one post a week. The thing standing between most people and revenue was never the idea. It was throughput and consistency, and that is exactly the part this removes. I'm not going to tell you what you'll earn. I'm handing you the desk, not a number. What you put through it is yours. Nothing publishes without my approval – that gate I kept on purpose. I wrote the whole build up in one article below ↓

SCOTTY BEAM

71,749 görüntüleme • 1 ay önce

Tired of teleoperation? One human video → 1,000s of robot demos. (📍GitHub ) Scaling Robot Data Without Dynamics Simulation or Robot Hardware Real2Render2Real (R2R2R) is a new way to scale robot data without physics simulation or hardware. You take a phone scan + a single monocular human demo. It tracks the motion, renders photorealistic scenes, and generates diverse, robot-agnostic trajectories ready for training. > No teleop, no sim, no robot, just a phone and a video > Train VLA models and diffusion policies directly on the output > Supports multiple robot embodiments with kinematic consistency > 1000s of demos in 1/27 the time of real-world collection Thank you, Max Fu, for sharing!! Project: Paper: Code coming soon: It shows that with the right pipeline, you can scale robot learning data without touching a robot. One of the most interesting directions in scalable robotics today. —— Weekly robotics and AI insights. Subscribe free:
0:29

Tired of teleoperation? One human video → 1,000s of robot demos. (📍GitHub ) Scaling Robot Data Without Dynamics Simulation or Robot Hardware Real2Render2Real (R2R2R) is a new way to scale robot data without physics simulation or hardware. You take a phone scan + a single monocular human demo. It tracks the motion, renders photorealistic scenes, and generates diverse, robot-agnostic trajectories ready for training. > No teleop, no sim, no robot, just a phone and a video > Train VLA models and diffusion policies directly on the output > Supports multiple robot embodiments with kinematic consistency > 1000s of demos in 1/27 the time of real-world collection Thank you, Max Fu, for sharing!! Project: Paper: Code coming soon: It shows that with the right pipeline, you can scale robot learning data without touching a robot. One of the most interesting directions in scalable robotics today. —— Weekly robotics and AI insights. Subscribe free:

Ilir Aliu

42,864 görüntüleme • 8 ay önce

AI BASKETBALL ANALYSIS. A FULL COMPUTER VISION SYSTEM. BUILT ON YOLO, OPENCV, AND PYTHON. Take any NBA broadcast. Any camera angle. Any resolution. Feed it into the system. YOLO finds every player and the ball. Frame by frame. No manual annotation. No pre-labeled data.The model just sees the court and understands it. Zero-shot classification looks at jersey pixels and decides who plays for whom. Two teams separated in milliseconds. Without ever being told the team names. But here's where it gets wild. Court keypoint detection identifies the geometry of the playing surface. Homography transforms the broadcast camera into a top-down tactical map. Real meters. Real coordinates. From a flat 2D video. Now every player has a position on a real court. Every movement becomes a data point. Every pass becomes a vector. Speed. Distance. Possession time. Pass networks. All extracted from nothing but pixels. No GPS. No chips in the ball. No million-dollar Second Spectrum setup. A Python script and a GPU. The full pipeline is open source. The tutorial walks through every line of code. From detection to transformation to analytics output. Teams used to pay six figures for this data. Now you build it in an afternoon. Computer vision didn't just enter basketball. It democratized it.
5:58:26

AI BASKETBALL ANALYSIS. A FULL COMPUTER VISION SYSTEM. BUILT ON YOLO, OPENCV, AND PYTHON. Take any NBA broadcast. Any camera angle. Any resolution. Feed it into the system. YOLO finds every player and the ball. Frame by frame. No manual annotation. No pre-labeled data.The model just sees the court and understands it. Zero-shot classification looks at jersey pixels and decides who plays for whom. Two teams separated in milliseconds. Without ever being told the team names. But here's where it gets wild. Court keypoint detection identifies the geometry of the playing surface. Homography transforms the broadcast camera into a top-down tactical map. Real meters. Real coordinates. From a flat 2D video. Now every player has a position on a real court. Every movement becomes a data point. Every pass becomes a vector. Speed. Distance. Possession time. Pass networks. All extracted from nothing but pixels. No GPS. No chips in the ball. No million-dollar Second Spectrum setup. A Python script and a GPU. The full pipeline is open source. The tutorial walks through every line of code. From detection to transformation to analytics output. Teams used to pay six figures for this data. Now you build it in an afternoon. Computer vision didn't just enter basketball. It democratized it.

zostaff

231,628 görüntüleme • 5 ay önce

166,700 fruit fly neurons are walking a robot cat across a desk right now. In September 2026, Google Research and HHMI Janelia mapped every neuron and every connection in an adult male fruit fly's full nervous system, brain, optic lobes, and ventral nerve cord together, and published it in Cell. The number is staggering on its own: 166,700 neurons, roughly 125 million synapses, the most complete nervous system map of any animal ever built. Now that exact wiring is running live, simulated neuron for neuron, inside a small robotic quadruped equipped with a camera and motion sensors. Movement comes in through the sensors, the biological circuitry fires the way it always did, and the output drives the robot's legs directly. No one wrote the balance logic. No one coded the gait. The insect's own nervous system does that job, just on legs it never had. To be clear about what this is and isn't: nothing biological is alive in this loop. It's a full computational reconstruction of real neural wiring, controlling hardware instead of an insect body. The brain is real. The body is borrowed. Remarkable proof that a mapped nervous system can drive any body you give it, or a preview of something people aren't ready to think through yet?
0:24

166,700 fruit fly neurons are walking a robot cat across a desk right now. In September 2026, Google Research and HHMI Janelia mapped every neuron and every connection in an adult male fruit fly's full nervous system, brain, optic lobes, and ventral nerve cord together, and published it in Cell. The number is staggering on its own: 166,700 neurons, roughly 125 million synapses, the most complete nervous system map of any animal ever built. Now that exact wiring is running live, simulated neuron for neuron, inside a small robotic quadruped equipped with a camera and motion sensors. Movement comes in through the sensors, the biological circuitry fires the way it always did, and the output drives the robot's legs directly. No one wrote the balance logic. No one coded the gait. The insect's own nervous system does that job, just on legs it never had. To be clear about what this is and isn't: nothing biological is alive in this loop. It's a full computational reconstruction of real neural wiring, controlling hardware instead of an insect body. The brain is real. The body is borrowed. Remarkable proof that a mapped nervous system can drive any body you give it, or a preview of something people aren't ready to think through yet?

Zentrix⌚️

232,128 görüntüleme • 1 ay önce

HERMES JUST GOT A CHATGPT INTERFACE AND IT IS COMPLETELY FREE. No terminal. No command line. No technical setup that takes half a day to figure out. Here is what just became possible. You can now run Hermes, DeepSeek, Ollama, and OpenClaw from a single clean chat interface. Switch between them in one click. Upload files directly. Attach knowledge bases and notes. Preview code without touching the terminal once. Use it on mobile. Make voice and video calls. All of it. Free. Open source. The setup took one conversation with Hermes itself. Pasted the GitHub link in. Hermes installed Open Web UI in Docker on its own. Connected the API server and gateway automatically. A clean ChatGPT-level interface running entirely on local hardware was live in minutes. No subscription. No data leaving your machine. No API bill at the end of the month. The people who set this up this weekend will have a local AI stack that rivals what most companies are paying $2,000 a month for in cloud infrastructure. The people who wait will keep paying for tools that send their data somewhere they cannot see. Both tools are free. Both are open source. The only thing between you and this setup is the next 30 minutes. Follow CyrilXBT for the exact config, setup steps, and local AI workflows I use to run a full stack without spending a dollar on cloud inference.
9:10

HERMES JUST GOT A CHATGPT INTERFACE AND IT IS COMPLETELY FREE. No terminal. No command line. No technical setup that takes half a day to figure out. Here is what just became possible. You can now run Hermes, DeepSeek, Ollama, and OpenClaw from a single clean chat interface. Switch between them in one click. Upload files directly. Attach knowledge bases and notes. Preview code without touching the terminal once. Use it on mobile. Make voice and video calls. All of it. Free. Open source. The setup took one conversation with Hermes itself. Pasted the GitHub link in. Hermes installed Open Web UI in Docker on its own. Connected the API server and gateway automatically. A clean ChatGPT-level interface running entirely on local hardware was live in minutes. No subscription. No data leaving your machine. No API bill at the end of the month. The people who set this up this weekend will have a local AI stack that rivals what most companies are paying $2,000 a month for in cloud infrastructure. The people who wait will keep paying for tools that send their data somewhere they cannot see. Both tools are free. Both are open source. The only thing between you and this setup is the next 30 minutes. Follow CyrilXBT for the exact config, setup steps, and local AI workflows I use to run a full stack without spending a dollar on cloud inference.

CyrilXBT

37,120 görüntüleme • 5 ay önce

🚨David Friedberg: AI is starting to identify and solve problems on its own “I'll give you a science corner example: there's this Evo 2 model that they publish at the Arc Institute, which Patrick Collison, you know, is the main funder and chairman.” “So that Evo 2 model, they just ingested all the DNA data they could find in the world.” “Trillions and trillions of base paired data that they ingested and then they looked at patterns in DNA. And that's it.” “They had no context for what the DNA represented, they had no context for the concept of genes, none of the structured understanding of what that DNA does, what it is, and you know what it did?” “They fed in the BRCA gene variant and the thing output a warning saying, ‘I think that this is a pathogenic variant to DNA,’ without having any context.” “This is the breast cancer allele.” “And it didn't have any knowledge and it wasn't trained on that at all.” “It had no knowledge that there are pathogenic variants for cancer, and it identified that this was a genetic variant that can cause some sort of pathogenic outcome in the organism.” “That's a great example where there's a lack of understanding at the human level on what really drives some of the patterns in nature, the patterns in society, the patterns in behavior that are kind of emergent phenomena perhaps, that these AI models are starting to identify.”
1:03

🚨David Friedberg: AI is starting to identify and solve problems on its own “I'll give you a science corner example: there's this Evo 2 model that they publish at the Arc Institute, which Patrick Collison, you know, is the main funder and chairman.” “So that Evo 2 model, they just ingested all the DNA data they could find in the world.” “Trillions and trillions of base paired data that they ingested and then they looked at patterns in DNA. And that's it.” “They had no context for what the DNA represented, they had no context for the concept of genes, none of the structured understanding of what that DNA does, what it is, and you know what it did?” “They fed in the BRCA gene variant and the thing output a warning saying, ‘I think that this is a pathogenic variant to DNA,’ without having any context.” “This is the breast cancer allele.” “And it didn't have any knowledge and it wasn't trained on that at all.” “It had no knowledge that there are pathogenic variants for cancer, and it identified that this was a genetic variant that can cause some sort of pathogenic outcome in the organism.” “That's a great example where there's a lack of understanding at the human level on what really drives some of the patterns in nature, the patterns in society, the patterns in behavior that are kind of emergent phenomena perhaps, that these AI models are starting to identify.”

The All-In Podcast

79,718 görüntüleme • 1 yıl önce

A robotic ballet! 🩰 Coordinating multiple robot arms on a busy factory floor is notoriously complex. Each arm needs to move without colliding with its neighbors or the surrounding equipment, and today that planning is still mostly done by hand, a process that takes specialists hundreds of hours. Researchers at UCL, Google DeepMind, and Intrinsic have introduced RoboBallet, a new AI system that tackles this challenge head-on. Using a combination of graph neural networks and reinforcement learning, RoboBallet learns how to coordinate many robots in real time, generating smooth, collision-free motion plans in seconds instead of days. In tests, the system handled up to 40 tasks with eight arms working together, far beyond the limits of traditional planners. More importantly, it showed strong generalization: RoboBallet could adapt instantly to new factory layouts or recover if one robot failed, something previous methods struggled with. For manufacturers, this could mean faster deployment of automation, less downtime, and the ability to reconfigure production lines on the fly. Applications range from automotive welding to electronics assembly, and even large-scale construction. The current system focuses on reaching tasks, but the team plans to expand it to more complex operations like pick-and-place or painting. Here's the paper:
0:54

A robotic ballet! 🩰 Coordinating multiple robot arms on a busy factory floor is notoriously complex. Each arm needs to move without colliding with its neighbors or the surrounding equipment, and today that planning is still mostly done by hand, a process that takes specialists hundreds of hours. Researchers at UCL, Google DeepMind, and Intrinsic have introduced RoboBallet, a new AI system that tackles this challenge head-on. Using a combination of graph neural networks and reinforcement learning, RoboBallet learns how to coordinate many robots in real time, generating smooth, collision-free motion plans in seconds instead of days. In tests, the system handled up to 40 tasks with eight arms working together, far beyond the limits of traditional planners. More importantly, it showed strong generalization: RoboBallet could adapt instantly to new factory layouts or recover if one robot failed, something previous methods struggled with. For manufacturers, this could mean faster deployment of automation, less downtime, and the ability to reconfigure production lines on the fly. Applications range from automotive welding to electronics assembly, and even large-scale construction. The current system focuses on reaching tasks, but the team plans to expand it to more complex operations like pick-and-place or painting. Here's the paper:

Lukas Ziegler

26,083 görüntüleme • 1 yıl önce

TESLA’S OPTIMUS JUST CARRIED A $10,000+ GPU LIKE IT WAS A GROCERY BAG! The robot is carrying NVIDIA’s RTX PRO 6000 Blackwell workstation GPU while walking and keeping the package stable in its hands. That sounds simple until you remember what’s happening underneath: 2 legs balancing, 2 arms controlling the load, cameras tracking the environment, and motors constantly making tiny corrections. A task that looks completely ordinary to a human is actually a serious test of real-world robotic control. This is where humanoid robots get interesting. They don’t need to learn 1,000 completely different machines if the same pair of hands can eventually pick up a GPU, move a box, open a door, carry equipment, and perform dozens of other tasks. One general-purpose robot could potentially replace an entire collection of specialized machines. The hardware is impressive, but the bigger story is versatility. The moment a robot can reliably handle arbitrary objects instead of one predefined task, the economics of automation start to change.
0:21

TESLA’S OPTIMUS JUST CARRIED A $10,000+ GPU LIKE IT WAS A GROCERY BAG! The robot is carrying NVIDIA’s RTX PRO 6000 Blackwell workstation GPU while walking and keeping the package stable in its hands. That sounds simple until you remember what’s happening underneath: 2 legs balancing, 2 arms controlling the load, cameras tracking the environment, and motors constantly making tiny corrections. A task that looks completely ordinary to a human is actually a serious test of real-world robotic control. This is where humanoid robots get interesting. They don’t need to learn 1,000 completely different machines if the same pair of hands can eventually pick up a GPU, move a box, open a door, carry equipment, and perform dozens of other tasks. One general-purpose robot could potentially replace an entire collection of specialized machines. The hardware is impressive, but the bigger story is versatility. The moment a robot can reliably handle arbitrary objects instead of one predefined task, the economics of automation start to change.

Future Memo

36,252 görüntüleme • 28 gün önce

Former Meta Chief AI Scientist Yann LeCun on the three paradigms of machine learning — and why the third is what made ChatGPT possible: Here's each one, and where it breaks. First, supervised learning. You tell the machine the answer. "You show it a picture, let's say of a table, and you tell it this is a table. So it's supervised because you tell it what the correct answer is." Get it wrong, and the machine rewrites itself: "The system computes its output, and if it says something else than table, then it's going to adjust its parameters, its internal structure, so that the output it produces gets closer to the output you want." Repeat at scale and something more than memorisation appears: "Eventually the system will find a way to recognize every image you trained it on, but also images it's never seen that are similar to the one you train it on. This is called a generalization ability." The limit: a human has to supply every single answer. That doesn't scale to the size of the internet. Second, reinforcement learning. You don't give the answer, only a verdict. "You don't tell the system what the correct answer is. You only tell it whether the answer it produced was good or bad." Learning to ride a bike, essentially: "You try to ride a bike and you don't know how to ride the bike and after a while you fall. So you know you did something bad and so you change your strategy a little bit. And eventually you learn how to ride a bike." For years the field assumed this was the closest thing to how animals actually learn. Yann LeCun's verdict: "Now it turns out reinforcement learning is extremely inefficient." It dominates wherever failure is free: "It works really well if you want to train a system to play chess or play go or poker, because you can have the system play millions and millions of games against itself and basically fine-tune itself. But it doesn't really work in the real world." The limit, in one image: "If you want to train a car to drive itself, you're not going to do it with reinforcement learning. It's going to crash thousands of times." On robotics he's careful rather than dismissive: "Reinforcement learning can be part of the solution, but it's not the complete answer. It's not sufficient." Third, self-supervised learning. You tell the machine nothing at all. "And this is what has enabled the recent progress in natural language understanding and chatbots." The strange part is that you stop asking for a task: "You don't train the system to accomplish any particular task. You just train it to basically capture the structure..." The method is deliberate sabotage: "You take a piece of text, you corrupt it in some way, by for example removing some words, and then you train a big neural net to predict the words that are missing." And one narrow version of that trick runs every chatbot on Earth: "A special case of this is that you take a piece of text and the last word in that text is not visible, and so you train the system to predict the last word in that text — and this is the way large language models are trained on." So why did the third one win? Supervised learning needs a human. Reinforcement learning needs a crash. Self-supervised learning needs neither — because the missing word and the correct answer are the same thing. The data grades itself.
2:58

Former Meta Chief AI Scientist Yann LeCun on the three paradigms of machine learning — and why the third is what made ChatGPT possible: Here's each one, and where it breaks. First, supervised learning. You tell the machine the answer. "You show it a picture, let's say of a table, and you tell it this is a table. So it's supervised because you tell it what the correct answer is." Get it wrong, and the machine rewrites itself: "The system computes its output, and if it says something else than table, then it's going to adjust its parameters, its internal structure, so that the output it produces gets closer to the output you want." Repeat at scale and something more than memorisation appears: "Eventually the system will find a way to recognize every image you trained it on, but also images it's never seen that are similar to the one you train it on. This is called a generalization ability." The limit: a human has to supply every single answer. That doesn't scale to the size of the internet. Second, reinforcement learning. You don't give the answer, only a verdict. "You don't tell the system what the correct answer is. You only tell it whether the answer it produced was good or bad." Learning to ride a bike, essentially: "You try to ride a bike and you don't know how to ride the bike and after a while you fall. So you know you did something bad and so you change your strategy a little bit. And eventually you learn how to ride a bike." For years the field assumed this was the closest thing to how animals actually learn. Yann LeCun's verdict: "Now it turns out reinforcement learning is extremely inefficient." It dominates wherever failure is free: "It works really well if you want to train a system to play chess or play go or poker, because you can have the system play millions and millions of games against itself and basically fine-tune itself. But it doesn't really work in the real world." The limit, in one image: "If you want to train a car to drive itself, you're not going to do it with reinforcement learning. It's going to crash thousands of times." On robotics he's careful rather than dismissive: "Reinforcement learning can be part of the solution, but it's not the complete answer. It's not sufficient." Third, self-supervised learning. You tell the machine nothing at all. "And this is what has enabled the recent progress in natural language understanding and chatbots." The strange part is that you stop asking for a task: "You don't train the system to accomplish any particular task. You just train it to basically capture the structure..." The method is deliberate sabotage: "You take a piece of text, you corrupt it in some way, by for example removing some words, and then you train a big neural net to predict the words that are missing." And one narrow version of that trick runs every chatbot on Earth: "A special case of this is that you take a piece of text and the last word in that text is not visible, and so you train the system to predict the last word in that text — and this is the way large language models are trained on." So why did the third one win? Supervised learning needs a human. Reinforcement learning needs a crash. Self-supervised learning needs neither — because the missing word and the correct answer are the same thing. The data grades itself.

Big Brain AI

49,293 görüntüleme • 2 ay önce

FOUR BOXES ON A DESK ARE NOW A FRONTIER AI CLUSTER This is four NVIDIA DGX Spark units wired together, sitting on a single desk and running Qwen3-235B end to end. Each box holds 128GB of unified memory. Stacked together, that is half a terabyte of memory pooled across the cluster, all from hardware you could fit in a backpack. A year ago a setup like this lived in a data center behind a badge reader. Now it runs next to a monitor and a coffee cup. Here is the part people get wrong. The four boxes do not magically merge into one giant GPU. The model is sharded across all of them, and they pass data over a fast link while every box does its slice of the work. That is the whole trick. No single box could hold a 235B model, but split four ways, it runs. The machines stay quiet and barely warm while doing it, sipping a fraction of the power a rack of server GPUs would burn. The wall between you and a frontier-class model used to be a cloud bill. Now it is just desk space. Bookmark this & Watch it run ↓
1:42

FOUR BOXES ON A DESK ARE NOW A FRONTIER AI CLUSTER This is four NVIDIA DGX Spark units wired together, sitting on a single desk and running Qwen3-235B end to end. Each box holds 128GB of unified memory. Stacked together, that is half a terabyte of memory pooled across the cluster, all from hardware you could fit in a backpack. A year ago a setup like this lived in a data center behind a badge reader. Now it runs next to a monitor and a coffee cup. Here is the part people get wrong. The four boxes do not magically merge into one giant GPU. The model is sharded across all of them, and they pass data over a fast link while every box does its slice of the work. That is the whole trick. No single box could hold a 235B model, but split four ways, it runs. The machines stay quiet and barely warm while doing it, sipping a fraction of the power a rack of server GPUs would burn. The wall between you and a frontier-class model used to be a cloud bill. Now it is just desk space. Bookmark this & Watch it run ↓

slash1s

19,473 görüntüleme • 4 ay önce

This guy spent several days teaching a tabletop robot arm to roll a burrito and when one could not do it he did not rewrite the controller he printed 2 more and launched a 3-arm setup for about $1,000. He does not rewrite the software, does not wait for a smarter model, does not update the imitation weights, he just prints another arm and connects it to the existing leader-follower through LeRobot, nonstop. And it got more interesting: the hardware of one arm is a DIY kit SO-101 for about $300 to $400 on STS3215 bus servos, coordination between the arms goes through leader-follower teleoperation in Hugging Face LeRobot, and replication goes through a Bambu Lab A1 for $399 that prints a copy in a day. It knows the position of the tortilla by the leader-arm coordinates at the start, knows the handoff moment between the arms by the fold phase, and knows the final rolling from the sequence of demonstrations from teleop sessions. And it even distributes the work between the arms, one does the initial folds, the second holds the tortilla still, the third performs the final rolling, depending on which phase the burrito is currently in. In one 57-second demonstration 3 arms rolled a burrito for the first time without human involvement, and the total stack cost less than $1,000 in hardware versus a UR5 at $25,000 or an industrial burrito machine Solbern BR-1500 that costs about $50,000. The viewer is not watching 3 robots rolling a burrito. The viewer is watching permission to believe in a future where a kitchen task on $1,000 of hardware is done by the same loop as an industrial one at $50,000. Here is what happens when the bottleneck stops being the intelligence of the controller and the quality of the imitation model, and becomes the number of arms in the setup, and for a maker with a 3D printer the number of arms is limited only by print time. 3 arms do not get tired between sessions, do not require retraining when a new one is added, do not degrade from repetition, and every next burrito goes through the same teleop pipeline at the same quality as the first. Imagine that multi-arm DIY setups are no longer built for one kitchen task, but printed for each one, burritos, sushi, tacos, pizza dough, pour-over coffee. We just watched hobby robotics shift from "retrain the controller" to "print 2 more": when one robot can not handle it, you do not make it smarter, you print 2 more. The viewer thinks they are watching a DIY experiment. They are watching a multi-agent robotics stack that in one 57-second demonstration rolled a burrito for the first time without human hands, and whose filling partially falls out on the final rolling. What will improve the final rolling, a softer silicone gripper, a 4th follower arm in the setup, or a different filling composition and a more moist tortilla?
0:57

This guy spent several days teaching a tabletop robot arm to roll a burrito and when one could not do it he did not rewrite the controller he printed 2 more and launched a 3-arm setup for about $1,000. He does not rewrite the software, does not wait for a smarter model, does not update the imitation weights, he just prints another arm and connects it to the existing leader-follower through LeRobot, nonstop. And it got more interesting: the hardware of one arm is a DIY kit SO-101 for about $300 to $400 on STS3215 bus servos, coordination between the arms goes through leader-follower teleoperation in Hugging Face LeRobot, and replication goes through a Bambu Lab A1 for $399 that prints a copy in a day. It knows the position of the tortilla by the leader-arm coordinates at the start, knows the handoff moment between the arms by the fold phase, and knows the final rolling from the sequence of demonstrations from teleop sessions. And it even distributes the work between the arms, one does the initial folds, the second holds the tortilla still, the third performs the final rolling, depending on which phase the burrito is currently in. In one 57-second demonstration 3 arms rolled a burrito for the first time without human involvement, and the total stack cost less than $1,000 in hardware versus a UR5 at $25,000 or an industrial burrito machine Solbern BR-1500 that costs about $50,000. The viewer is not watching 3 robots rolling a burrito. The viewer is watching permission to believe in a future where a kitchen task on $1,000 of hardware is done by the same loop as an industrial one at $50,000. Here is what happens when the bottleneck stops being the intelligence of the controller and the quality of the imitation model, and becomes the number of arms in the setup, and for a maker with a 3D printer the number of arms is limited only by print time. 3 arms do not get tired between sessions, do not require retraining when a new one is added, do not degrade from repetition, and every next burrito goes through the same teleop pipeline at the same quality as the first. Imagine that multi-arm DIY setups are no longer built for one kitchen task, but printed for each one, burritos, sushi, tacos, pizza dough, pour-over coffee. We just watched hobby robotics shift from "retrain the controller" to "print 2 more": when one robot can not handle it, you do not make it smarter, you print 2 more. The viewer thinks they are watching a DIY experiment. They are watching a multi-agent robotics stack that in one 57-second demonstration rolled a burrito for the first time without human hands, and whose filling partially falls out on the final rolling. What will improve the final rolling, a softer silicone gripper, a 4th follower arm in the setup, or a different filling composition and a more moist tortilla?

Blaze

61,030 görüntüleme • 5 ay önce

Playing #VR games "without" a headset... ... Absolute genius or total abomination? This is called PortalVR and it lets you play any SteamVR game or app on your Windows PC without wearing a headset. Just plug in your #MetaQuest via USB, put it on your desk facing you, and use your controllers while facing your screen. The setup and connection is quite seamless and the team came up with a clever locomotion system to simulate turning the user's head by just tapping the top of the thumbstick. Now the biggest question is: does this make any sense? In my opinion, to play "standard" VR #games, it doesn't. You lose the beauty of being immersed and reaching out to grab objects or looking around can be tricky. On the other hand, I think there might be experiences specifically designed with this interaction in mind like interactive installations or educational experiences that could greatly benefit from a more tactile interface than a mouse and a keyboard
1:10

Playing #VR games "without" a headset... ... Absolute genius or total abomination? This is called PortalVR and it lets you play any SteamVR game or app on your Windows PC without wearing a headset. Just plug in your #MetaQuest via USB, put it on your desk facing you, and use your controllers while facing your screen. The setup and connection is quite seamless and the team came up with a clever locomotion system to simulate turning the user's head by just tapping the top of the thumbstick. Now the biggest question is: does this make any sense? In my opinion, to play "standard" VR #games, it doesn't. You lose the beauty of being immersed and reaching out to grab objects or looking around can be tricky. On the other hand, I think there might be experiences specifically designed with this interaction in mind like interactive installations or educational experiences that could greatly benefit from a more tactile interface than a mouse and a keyboard

Gabriele Romagnoli

67,051 görüntüleme • 9 ay önce