Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

Codex iterated a pure NumPy + cv2 closed-loop heuristic policy for VizDoom D3 Battle. No neural network training, no map, no object coordinates, no seed-specific routes. Just screen pixels plus public game variables, roughly the same signals a human player gets. It works surprisingly well. Notes and videos are... show more

Jiayi Weng

12,307 subscribers

2,778,279 views • 2 months ago •via X (Twitter)

Gaming Science & Technology

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

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.

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

230,194 views • 3 months ago

I handed Claude a creator contract clause and asked for a vetted human expert to check it......without leaving the chat. Tendem now plugs into Claude, ChatGPT, Cursor and Codex via MCP. You mention Tendem, approve the quote, and the work comes back in the same conversation. The AI caught the obvious risks....uncapped liability, no kill fee, payment a client could withhold on a whim. Then a real expert does a second pass on the same clauses. AI scopes it. A human backs it. No new tab, no copy-paste. Toloka Tendem AI Live today →

I handed Claude a creator contract clause and asked for a vetted human expert to check it......without leaving the chat. Tendem now plugs into Claude, ChatGPT, Cursor and Codex via MCP. You mention Tendem, approve the quote, and the work comes back in the same conversation. The AI caught the obvious risks....uncapped liability, no kill fee, payment a client could withhold on a whim. Then a real expert does a second pass on the same clauses. AI scopes it. A human backs it. No new tab, no copy-paste. Toloka Tendem AI Live today →

Ace

16,900 views • 29 days ago

Justin Bieber Vows To Hunt Down and Kill the Pedophiles Who Abused Him as a Child Justin Bieber just declared war. The same creeps and pedophiles who preyed on him as a child are now in his crosshairs. He's done playing games. No more coded messages in his lyrics. No more subtle signals. He’s saying it out loud: he's coming for them.

Justin Bieber Vows To Hunt Down and Kill the Pedophiles Who Abused Him as a Child Justin Bieber just declared war. The same creeps and pedophiles who preyed on him as a child are now in his crosshairs. He's done playing games. No more coded messages in his lyrics. No more subtle signals. He’s saying it out loud: he's coming for them.

TPV Sean

729,703 views • 3 months ago

If you’re playing on MonkeyTilt, the outcome might already be decided BEFORE you bet “The cryptography behind every bet has always been sound” We tested it It isn’t... we sent two different client seeds for the same nonce same result both times the server is ignoring the client seed entirely the player has ZERO influence on outcomes — what we did — opened the MonkeyTilt provably fair verifier with a revealed server seed entered nonce 1 → got 3.04x changed the client seed same server seed, same nonce → 3.04x changed it again → 3.04x the client seed has ZERO effect on the outcome — why this matters — in a provably fair system the outcome is derived from: -server seed -client seed -nonce the client seed exists so the player contributes randomness the server CAN’T predict if it’s ignored the server knows every outcome BEFORE you bet the hash commitment proves nothing because the server already controls all the inputs — not a frontend bug — we checked the network traffic the verifier sends a WebSocket request with: -client seed -server seed -nonce the server receives different client seeds and returns the same result anyway request 1 → client_seed “Clientseed123456789” → 3.04x request 2 → client_seed “NEWSEED12345” → 3.04x this is the server ignoring input not a display issue — live bet traffic — placing a Limbo bet returns: -random_multiplier -total_payout -round_closed that’s it no hash no seed reference no nonce no proof seeds are fetched from a separate endpoint game engine and seed system are DISCONNECTED — nonce verification — you can only verify nonces you’ve already played in a real system all outcomes are predetermined once the server seed is committed - they’re deterministic blocking future nonces suggests results don’t exist until you bet — summary — -client seed ignored by server -no cryptographic proof in bet responses -game engine and seed system DISCONNECTED -nonce verification restricted this isn’t provably fair it looks like provably fair UI on top of server side RNG MonkeyTilt should probably address this anyone can verify this themselves in 60 seconds - open the verifier, change the client seed, watch the result stay the same

If you’re playing on MonkeyTilt, the outcome might already be decided BEFORE you bet “The cryptography behind every bet has always been sound” We tested it It isn’t... we sent two different client seeds for the same nonce same result both times the server is ignoring the client seed entirely the player has ZERO influence on outcomes — what we did — opened the MonkeyTilt provably fair verifier with a revealed server seed entered nonce 1 → got 3.04x changed the client seed same server seed, same nonce → 3.04x changed it again → 3.04x the client seed has ZERO effect on the outcome — why this matters — in a provably fair system the outcome is derived from: -server seed -client seed -nonce the client seed exists so the player contributes randomness the server CAN’T predict if it’s ignored the server knows every outcome BEFORE you bet the hash commitment proves nothing because the server already controls all the inputs — not a frontend bug — we checked the network traffic the verifier sends a WebSocket request with: -client seed -server seed -nonce the server receives different client seeds and returns the same result anyway request 1 → client_seed “Clientseed123456789” → 3.04x request 2 → client_seed “NEWSEED12345” → 3.04x this is the server ignoring input not a display issue — live bet traffic — placing a Limbo bet returns: -random_multiplier -total_payout -round_closed that’s it no hash no seed reference no nonce no proof seeds are fetched from a separate endpoint game engine and seed system are DISCONNECTED — nonce verification — you can only verify nonces you’ve already played in a real system all outcomes are predetermined once the server seed is committed - they’re deterministic blocking future nonces suggests results don’t exist until you bet — summary — -client seed ignored by server -no cryptographic proof in bet responses -game engine and seed system DISCONNECTED -nonce verification restricted this isn’t provably fair it looks like provably fair UI on top of server side RNG MonkeyTilt should probably address this anyone can verify this themselves in 60 seconds - open the verifier, change the client seed, watch the result stay the same

CoinBets🔍

20,973 views • 4 months ago

PUBLIC WORKS MINISTER DEAN MACPHERSON AND HIS DEPARTMENT ARRANGE DD MABUZA FUNERAL: Funerals such as these are generally organised by the Department of Public Works. The entire infrastructure, tent, sounds, chairs etc. Dean Macpherson MP pulled it off in a week with absolutely no corruption scandal. Gave a befitting send off to DD. The department made sure the family was there with them every step of the way and having an input as to how they want the funeral. No R10 million scandal. No looting. No wastage. Well done Minister Dean Macpherson and team. No looting. No scandal. Its just GNU in action The GNU in its present form will save South Africa.

PUBLIC WORKS MINISTER DEAN MACPHERSON AND HIS DEPARTMENT ARRANGE DD MABUZA FUNERAL: Funerals such as these are generally organised by the Department of Public Works. The entire infrastructure, tent, sounds, chairs etc. Dean Macpherson MP pulled it off in a week with absolutely no corruption scandal. Gave a befitting send off to DD. The department made sure the family was there with them every step of the way and having an input as to how they want the funeral. No R10 million scandal. No looting. No wastage. Well done Minister Dean Macpherson and team. No looting. No scandal. Its just GNU in action The GNU in its present form will save South Africa.

Goolam

43,789 views • 1 year ago

A YC company just shipped the closest thing yet to "type a sentence, get a real company." It's called Naïve. AI employees with their own bank accounts, emails, and credentials. They incorporate LLCs, sign up for tools, and run companies as themselves. No human in the loop. Here's how it works: 👇

A YC company just shipped the closest thing yet to "type a sentence, get a real company." It's called Naïve. AI employees with their own bank accounts, emails, and credentials. They incorporate LLCs, sign up for tools, and run companies as themselves. No human in the loop. Here's how it works: 👇

Hasan Toor

108,667 views • 2 months ago

I took Clicky and I made it 5x Faster So I saw that Farza 🇵🇰🇺🇸 was using uses Claude's vision to find UI elements on screen, send a screenshot, wait for coordinates back. It works, but it's slow. I replaced that with OmniParser V2 by Microsoft which is a local Object detection model trained specifically on UI elements. It runs on-device, detects every button, menu, and icon in 400ms, and gives me pixel-precise coordinates. No API call, no latency, no cost. The green highlights you see around the UI elements is the detection overlay and you can see as I am switching the screen it takes no time to detect and highlight which is pretty neat! With Local models improving by the day, Its the right direction for applications like these. Next up: video-synced tutorials where a YouTube tutorial pauses and waits for you to perform each action in the real app. I am not stopping!

I took Clicky and I made it 5x Faster So I saw that Farza 🇵🇰🇺🇸 was using uses Claude's vision to find UI elements on screen, send a screenshot, wait for coordinates back. It works, but it's slow. I replaced that with OmniParser V2 by Microsoft which is a local Object detection model trained specifically on UI elements. It runs on-device, detects every button, menu, and icon in 400ms, and gives me pixel-precise coordinates. No API call, no latency, no cost. The green highlights you see around the UI elements is the detection overlay and you can see as I am switching the screen it takes no time to detect and highlight which is pretty neat! With Local models improving by the day, Its the right direction for applications like these. Next up: video-synced tutorials where a YouTube tutorial pauses and waits for you to perform each action in the real app. I am not stopping!

Milind S

73,191 views • 3 months ago

THE FIRST VIDEO GAME WAS JUST DOTS, KNOBS, AND PURE NERD GENIUS In 1958, physicist William Higinbotham built Tennis for Two using an oscilloscope, a few knobs, and straight-up electrical wizardry. No story. No skins. No microtransactions - just two players, a bouncing dot, and raw analog vibes. This was pre-pixels, and somehow it kicked off everything from Space Invaders to Elden Ring. Every AAA title owes a debt to this glowing green fossil. Source: Wowfeeds

THE FIRST VIDEO GAME WAS JUST DOTS, KNOBS, AND PURE NERD GENIUS In 1958, physicist William Higinbotham built Tennis for Two using an oscilloscope, a few knobs, and straight-up electrical wizardry. No story. No skins. No microtransactions - just two players, a bouncing dot, and raw analog vibes. This was pre-pixels, and somehow it kicked off everything from Space Invaders to Elden Ring. Every AAA title owes a debt to this glowing green fossil. Source: Wowfeeds

Mario Nawfal

81,629 views • 1 year ago

My YouTube channel (22K subscribers, built in 8 months) was permanently terminated. 0 strikes. No prior warnings. No specific video cited. No evidence shown. The reason? “Spam, scam, deceptive practices.” Here’s one of my videos. Where exactly is the scam? Where is the spam? I created original, high-quality entertainment content. No spam networks. No deception of any sort. Never used a single automation tool. Now I am banned for life from making YouTube videos. How is this fair? Now hundreds of viewers are messaging me asking what happened to their favorite channel. This wasn’t just an account. It was months of work, investment, and real human connection erased in a click. If this was an automated decision, it deserves a real human review. TeamYouTube I’m asking for transparency and fairness.

My YouTube channel (22K subscribers, built in 8 months) was permanently terminated. 0 strikes. No prior warnings. No specific video cited. No evidence shown. The reason? “Spam, scam, deceptive practices.” Here’s one of my videos. Where exactly is the scam? Where is the spam? I created original, high-quality entertainment content. No spam networks. No deception of any sort. Never used a single automation tool. Now I am banned for life from making YouTube videos. How is this fair? Now hundreds of viewers are messaging me asking what happened to their favorite channel. This wasn’t just an account. It was months of work, investment, and real human connection erased in a click. If this was an automated decision, it deserves a real human review. TeamYouTube I’m asking for transparency and fairness.

Sith Daddy

1,639,958 views • 5 months ago

Seen a post today… Stats lie. Show me the film! No stat for how deep you shoot that 3 or level of difficulty. No stat for the IQ it took to let go of that ball on the trap. No stat for turning around and immediately knowing it’s time for an inbound play. No stat for communication. No stat for scoring on 2 or 3 D1 level talent players, the points are same BUT different. No stat line for a player scoring on a triple team while the stands are chanting over rated. No stat for that timely screen or that hockey pass. No stat for motor bc you played all game and still drained the deep 3…. Recruits: Stats don’t tell the whole story! They tell a piece of that story. Stay the course!

Seen a post today… Stats lie. Show me the film! No stat for how deep you shoot that 3 or level of difficulty. No stat for the IQ it took to let go of that ball on the trap. No stat for turning around and immediately knowing it’s time for an inbound play. No stat for communication. No stat for scoring on 2 or 3 D1 level talent players, the points are same BUT different. No stat line for a player scoring on a triple team while the stands are chanting over rated. No stat for that timely screen or that hockey pass. No stat for motor bc you played all game and still drained the deep 3…. Recruits: Stats don’t tell the whole story! They tell a piece of that story. Stay the course!

Amber Mathurin ™️

12,654 views • 2 years ago

ANTHROPIC LEAD ENGINEER MAKING $2.3M/YEAR JUST LEAKED A 12-PAGE DOCUMENT - AND GOT FIRED 15 MINUTES AFTER PUBLISHING most developers build loops - but almost every loop breaks in one of five ways, and each failure has a symptom visible from the outside without reading a single line of code > Blind loop - the agent waits for a human to hand it work every morning - automating execution but not discovery > Tangled loop - parallel agents share one directory and overwrite each other parallelism that destroys instead of multiplies > Nodding loop - the agent grades its own work and approves it every time a loop that has never said no to itself is broken > Amnesiac loop - results live only in the context window every morning the system wakes up with no memory of yesterday > Manual loop - no trigger a human presses the button to start it that is not a loop that is a script waiting for a person a real loop finds its own work - remembers what it did has something that can say no never lets two agents touch the same file and fires itself on a timer remove any one of the five - and the loop either breaks or never starts 12 pages that changed how I build agentic systems today

ANTHROPIC LEAD ENGINEER MAKING $2.3M/YEAR JUST LEAKED A 12-PAGE DOCUMENT - AND GOT FIRED 15 MINUTES AFTER PUBLISHING most developers build loops - but almost every loop breaks in one of five ways, and each failure has a symptom visible from the outside without reading a single line of code > Blind loop - the agent waits for a human to hand it work every morning - automating execution but not discovery > Tangled loop - parallel agents share one directory and overwrite each other parallelism that destroys instead of multiplies > Nodding loop - the agent grades its own work and approves it every time a loop that has never said no to itself is broken > Amnesiac loop - results live only in the context window every morning the system wakes up with no memory of yesterday > Manual loop - no trigger a human presses the button to start it that is not a loop that is a script waiting for a person a real loop finds its own work - remembers what it did has something that can say no never lets two agents touch the same file and fires itself on a timer remove any one of the five - and the loop either breaks or never starts 12 pages that changed how I build agentic systems today

Sprytix

153,149 views • 17 days ago

AI FOOTBALL ANALYSIS. A FULL COMPUTER VISION SYSTEM. BUILT ON YOLO, OPENCV, AND PYTHON. You upload a regular match video. No sensors, no GPS trackers, just camera footage. The neural network finds every player, referee, and ball on its own. Every frame, in real time. KMeans clustering breaks down jersey colors pixel by pixel. The system splits players into teams automatically. Without a single manual hint. Optical Flow tracks camera movement. Separates it from player movement. Perspective Transformation converts pixels into real meters. Speed of every player. Distance covered. Ball possession percentage. All calculated automatically. Four hours of tutorial from zero to a working system. The model is trained on real Bundesliga matches. Runs on a regular GPU. Python code - take it and run. Sports analytics is no longer behind closed doors. AI leveled the playing field.

AI FOOTBALL ANALYSIS. A FULL COMPUTER VISION SYSTEM. BUILT ON YOLO, OPENCV, AND PYTHON. You upload a regular match video. No sensors, no GPS trackers, just camera footage. The neural network finds every player, referee, and ball on its own. Every frame, in real time. KMeans clustering breaks down jersey colors pixel by pixel. The system splits players into teams automatically. Without a single manual hint. Optical Flow tracks camera movement. Separates it from player movement. Perspective Transformation converts pixels into real meters. Speed of every player. Distance covered. Ball possession percentage. All calculated automatically. Four hours of tutorial from zero to a working system. The model is trained on real Bundesliga matches. Runs on a regular GPU. Python code - take it and run. Sports analytics is no longer behind closed doors. AI leveled the playing field.

zostaff

1,073,241 views • 3 months ago

A girl is making $14,000 a month from a tablet and two apps. She spends one hour on it. Then closes everything and gets on with her day. Finds a kids video already pulling millions of views. Studies what makes toddlers rewatch it on loop. Then copies the concept and drops it into Claude. One message back and she has a complete video brief. Character descriptions, scene structure, color palette, the exact hook that makes small children reach for the screen. That brief goes straight into Meideo. A few minutes later a fully animated kids video is ready. Dancing alphabet characters. Glowing stars. Animals bouncing in saturated colors that toddlers cannot look away from. No studio. No camera. No editing software. No laptop. Just Claude for the creative and Meideo for the visuals. Two apps. One hour maximum. Here is why it works. Parents don’t put a kids video on once. They put it on and walk away. Kids rewatch the same video six times before lunch. YouTube counts every view and pays accordingly. Kids content CPMs are among the highest on the platform. The loop is the product. $14,000 last month. Same workflow. Same two apps. Same hour. Save this before everyone figures out the poolside office was always an option.

A girl is making $14,000 a month from a tablet and two apps. She spends one hour on it. Then closes everything and gets on with her day. Finds a kids video already pulling millions of views. Studies what makes toddlers rewatch it on loop. Then copies the concept and drops it into Claude. One message back and she has a complete video brief. Character descriptions, scene structure, color palette, the exact hook that makes small children reach for the screen. That brief goes straight into Meideo. A few minutes later a fully animated kids video is ready. Dancing alphabet characters. Glowing stars. Animals bouncing in saturated colors that toddlers cannot look away from. No studio. No camera. No editing software. No laptop. Just Claude for the creative and Meideo for the visuals. Two apps. One hour maximum. Here is why it works. Parents don’t put a kids video on once. They put it on and walk away. Kids rewatch the same video six times before lunch. YouTube counts every view and pays accordingly. Kids content CPMs are among the highest on the platform. The loop is the product. $14,000 last month. Same workflow. Same two apps. Same hour. Save this before everyone figures out the poolside office was always an option.

Shelpid.WI3M

26,353 views • 1 month ago

I built a biologically inspired spiking neural network from scratch and it learned with %5 accuracy to do addition :) There is no backpropagation, no artificial loss functions - just spikes, synapses, and dopamine-like reward signals. it uses STDP -> "Spike-Timing-Dependent Plasticity" with modulated rewards This is super fun and I will try to get it to learn with better accuracy. I also need to better understand how all the moving parts fit together Link to source code in comment which has a detailed readme and html with animations explaining how it all works

I built a biologically inspired spiking neural network from scratch and it learned with %5 accuracy to do addition :) There is no backpropagation, no artificial loss functions - just spikes, synapses, and dopamine-like reward signals. it uses STDP -> "Spike-Timing-Dependent Plasticity" with modulated rewards This is super fun and I will try to get it to learn with better accuracy. I also need to better understand how all the moving parts fit together Link to source code in comment which has a detailed readme and html with animations explaining how it all works

echo.hive

447,188 views • 9 months ago

India’s startup media is broken. It’s slow. PR-heavy. Full of recycled headlines. Print is dead. Podcasts are everywhere. CNBC feels like it’s talking to traders, not founders. Funding news breaks as a tweet. Gets recycled as a LinkedIn post. Ends up as a reel with a clickbait caption. No one trusts reels made for outrage and shares. Same headline. Everywhere. No depth. No context. No signal. The real context is buried in WhatsApp chats and closed-door conversations. We’re about to make it public. The Offline Network Live, high-signal, context-first conversations. Launching soon. Want to call the shots on what we discuss live? Comment and I’ll get you into our inner-circle WhatsApp.

India’s startup media is broken. It’s slow. PR-heavy. Full of recycled headlines. Print is dead. Podcasts are everywhere. CNBC feels like it’s talking to traders, not founders. Funding news breaks as a tweet. Gets recycled as a LinkedIn post. Ends up as a reel with a clickbait caption. No one trusts reels made for outrage and shares. Same headline. Everywhere. No depth. No context. No signal. The real context is buried in WhatsApp chats and closed-door conversations. We’re about to make it public. The Offline Network Live, high-signal, context-first conversations. Launching soon. Want to call the shots on what we discuss live? Comment and I’ll get you into our inner-circle WhatsApp.

Utsav Somani

54,754 views • 11 months ago

‼️🇺🇸 💥🇻🇪 Silent Strike in Venezuela Deep inside Venezuela, something moved quietly. No press briefings. No warning. Just precision. A joint shadow operation linked to US special forces and intelligence assets closed in on a powerful drug cartel network. The targets were not small players. These were men who fueled chaos, violence, and narcotics routes stretching far beyond Latin America. The operation was fast. Clean. Calculated. Key cartel figures were captured before they even understood what was happening. No streets turned into battlefields. No civilians caught in the crossfire. Just a message delivered in silence. This was not about headlines. It was about control. It was about reminding the underworld that distance does not mean safety. When the mission ended, the teams disappeared the same way they arrived. Quietly. And somewhere in the shadows, other networks are now wondering one thing. Who is next?

‼️🇺🇸 💥🇻🇪 Silent Strike in Venezuela Deep inside Venezuela, something moved quietly. No press briefings. No warning. Just precision. A joint shadow operation linked to US special forces and intelligence assets closed in on a powerful drug cartel network. The targets were not small players. These were men who fueled chaos, violence, and narcotics routes stretching far beyond Latin America. The operation was fast. Clean. Calculated. Key cartel figures were captured before they even understood what was happening. No streets turned into battlefields. No civilians caught in the crossfire. Just a message delivered in silence. This was not about headlines. It was about control. It was about reminding the underworld that distance does not mean safety. When the mission ended, the teams disappeared the same way they arrived. Quietly. And somewhere in the shadows, other networks are now wondering one thing. Who is next?

Defense Intelligence

367,227 views • 6 months ago

AI for robotics has some monumental challenges, like high dimensionality and reasoning with the physical world. Ashok Elluswamy, AI lead at Tesla, thinks the hardest problem is Evaluation. - Loss isn't a perfect indicator of Policy Neural Network's quality. - Testing in the real world for all the long-tail failure cases is not possible. To solve the evaluation challenge, Tesla has developed a World Simulator Model. It is trained on state and action pairs, so for a given current camera feed and current action, the model generates the next frame in the cameras. Once the world simulation is good enough, it can be hooked up to the policy network for a closed-loop simulation. In this setup, the model generates the next video frames, and based on those, the policy network comes up with the next action, which feeds back into the world simulation; this continuous loop creates a coherent simulation. The world simulation neural network is not just applicable to self-driving but also to humanoids, because Tesla trains it on common data. The examples show Optimus navigating or manipulating inside the world simulation, with all the pixels being generated by the world simulation model. The robotic agents can be tested in an accurately reproduced edge case and in other potential variations of the same edge case.

AI for robotics has some monumental challenges, like high dimensionality and reasoning with the physical world. Ashok Elluswamy, AI lead at Tesla, thinks the hardest problem is Evaluation. - Loss isn't a perfect indicator of Policy Neural Network's quality. - Testing in the real world for all the long-tail failure cases is not possible. To solve the evaluation challenge, Tesla has developed a World Simulator Model. It is trained on state and action pairs, so for a given current camera feed and current action, the model generates the next frame in the cameras. Once the world simulation is good enough, it can be hooked up to the policy network for a closed-loop simulation. In this setup, the model generates the next video frames, and based on those, the policy network comes up with the next action, which feeds back into the world simulation; this continuous loop creates a coherent simulation. The world simulation neural network is not just applicable to self-driving but also to humanoids, because Tesla trains it on common data. The examples show Optimus navigating or manipulating inside the world simulation, with all the pixels being generated by the world simulation model. The robotic agents can be tested in an accurately reproduced edge case and in other potential variations of the same edge case.

The Humanoid Hub

16,591 views • 5 months ago

These same libs who have said for years that Charlie Kirk is a fascist are now literally crying? It shows you they don’t actually believe anything they say. It’s a game to them. They’re all in the same club and friends behind the scenes and the do this shit all for show with no care about who it harms in the process.

These same libs who have said for years that Charlie Kirk is a fascist are now literally crying? It shows you they don’t actually believe anything they say. It’s a game to them. They’re all in the same club and friends behind the scenes and the do this shit all for show with no care about who it harms in the process.

B L A K E L E Y™℠©® LLC

1,188,085 views • 10 months ago

Valve is selling a $1,349 machine the internet already prints for $247. Look at this guy. He just fired up a 3D printer and watched it lay down the entire Steam Machine shell, layer by layer, in his room. No factory. No supply chain. No release date. Just a spool of plastic and a printer humming on his desk. It cost him $247. Meanwhile Valve spent months engineering this thing, building the hype, planning the June 29 launch. And they're charging the community $1,049 to $1,349 for it. No subsidy. Pure PC margins. One of them took a print file and a weekend. The other took a corporation. Same box. Same idea. One costs a quarter of the price, and it's sitting on his desk right now.

Valve is selling a $1,349 machine the internet already prints for $247. Look at this guy. He just fired up a 3D printer and watched it lay down the entire Steam Machine shell, layer by layer, in his room. No factory. No supply chain. No release date. Just a spool of plastic and a printer humming on his desk. It cost him $247. Meanwhile Valve spent months engineering this thing, building the hype, planning the June 29 launch. And they're charging the community $1,049 to $1,349 for it. No subsidy. Pure PC margins. One of them took a print file and a weekend. The other took a corporation. Same box. Same idea. One costs a quarter of the price, and it's sitting on his desk right now.

John Doe

342,409 views • 27 days ago

This guy literally told Claude to run every day at 4am. Now it opens his website, writes a blog post, and optimizes it for SEO. Every single day. Zero involvement. Here's how Claude computer use actually works: It looks at your screen. Moves your mouse. Clicks buttons. Opens apps. Completes the full task on its own. No code. No workflow builder. No setup. You just tell it what you want done. And the 4am SEO trick? You schedule it once inside Claude Co-work. Claude runs it every night while you sleep. Your site gets fresh content. You wake up and check the results. That's not a chatbot. That's an AI employee.

This guy literally told Claude to run every day at 4am. Now it opens his website, writes a blog post, and optimizes it for SEO. Every single day. Zero involvement. Here's how Claude computer use actually works: It looks at your screen. Moves your mouse. Clicks buttons. Opens apps. Completes the full task on its own. No code. No workflow builder. No setup. You just tell it what you want done. And the 4am SEO trick? You schedule it once inside Claude Co-work. Claude runs it every night while you sleep. Your site gets fresh content. You wake up and check the results. That's not a chatbot. That's an AI employee.

Julian Goldie SEO

133,228 views • 4 months ago