Video yükleniyor...

Video Yüklenemedi

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

Ana Sayfaya Dön

Training world models needs egocentric video and dense action signals, synchronized. That data is genuinely hard to find. We built it from Counter-Strike 2 demos. CS2-10k: 600K+ player-round videos, 10K+ hours, per-frame annotations — keyboard state, mouse delta, 3D position, camera yaw/pitch. All paired to the visual stream. Why... show more

Reka

21,588 subscribers

95,620 görüntüleme • 3 ay önce •via X (Twitter)

Oyun Eğitim Bilim & Teknoloji

Anya Rossi• Live Now

Private livecam show

16 Yorum

Reka profil fotoğrafı

Reka3 ay önce

Read more: Github repo: @huggingface dataset : Sample dataset viewer:

Julien Blanchon 🇺🇦 profil fotoğrafı

Julien Blanchon 🇺🇦3 ay önce

Very cool, I have a similar scale dataset but with player view here

Wally Buck profil fotoğrafı

Wally Buck3 ay önce

So when my robot 360 no-scopes me and then Teabags my dead corpse repeatedly when I ask it to perform an overlyb token consumptive task, I’ll know why. Thanks!

Nick Venturi profil fotoğrafı

Nick Venturi3 ay önce

cs players finally contributing to society

Jan Schnyder profil fotoğrafı

Jan Schnyder3 ay önce

p cool @arnie_hacker

Heimrih Lim profil fotoğrafı

Heimrih Lim3 ay önce

now i can tell people in my company im making dataset instead of playing cs2

Lazarz profil fotoğrafı

Lazarz3 ay önce

That @arnie_hacker?

🇨‌🇦‌🇷‌🇿‌🇪‌🇳‌ profil fotoğrafı

🇨‌🇦‌🇷‌🇿‌🇪‌🇳‌3 ay önce

Based

Miriam profil fotoğrafı

Miriam3 ay önce

This is a huge contribution to world-model research.

Wayne Nilsen profil fotoğrafı

Wayne Nilsen3 ay önce

where is the 'strap your phone to your head and earn money' company? can autolabel positions of body segments right? with something like an iphone you have the ir cam and ar-kit

Franz Srambical (in LAUSANNE) profil fotoğrafı

Franz Srambical (in LAUSANNE)3 ay önce

Nice! We have 3k hours of paired recordings of people working over months, and trained a small idm on part of the dataset; gonna behaviour-clone soon:

dmnk profil fotoğrafı

dmnk3 ay önce

this but unironically

何 profil fotoğrafı

何3 ay önce

amazing!

Jerry Chen profil fotoğrafı

Jerry Chen3 ay önce

Does mouse dpi considered?

Declan Kutscher profil fotoğrafı

Declan Kutscher3 ay önce

Does it have any concept of the strategy or economy in cs2?

Damian Morley profil fotoğrafı

Damian Morley3 ay önce

Niceeee

Benzer Videolar

With Atlas, you can literally make films. And it also basically solves first-frame-last-frame transitions. Reconstruct a 3D scene from images, then position the camera exactly where you want it, and let Atlas fill in what the camera would see along that path. You’re in the director's chair, staging the scene. Here we made a 1 min video with a couple images of totally different scenes, and we set where the camera keyframes should move through each part of the scene. The scene-to-scene transitions are great. Atlas makes it feel like a smooth shot with no cuts.

With Atlas, you can literally make films. And it also basically solves first-frame-last-frame transitions. Reconstruct a 3D scene from images, then position the camera exactly where you want it, and let Atlas fill in what the camera would see along that path. You’re in the director's chair, staging the scene. Here we made a 1 min video with a couple images of totally different scenes, and we set where the camera keyframes should move through each part of the scene. The scene-to-scene transitions are great. Atlas makes it feel like a smooth shot with no cuts.

David Pantera

27,529 görüntüleme • 1 ay önce

The largest fully annotated open egocentric human dataset. Today we're open-sourcing EgoSuite-Open100K with Hugging Face : 100,000 hours of egocentric human data, free for anyone to use. Physical AI has a scaling law of its own now, and its input is human data. That part is settled. What the field still needs is a shared foundation to build on. 15,000+ tasks. 15,000+ real scenes, from factory floors to retail backrooms. Hand pose, body pose and subtask-level semantics, with wrist camera coverage on part of the set. Licensed for commercial training, not research-only. We build the data infrastructure for Physical AI. Open100K is the first public layer of it. The first 10,000 hours are live now, with the rest rolling out in stages.

The largest fully annotated open egocentric human dataset. Today we're open-sourcing EgoSuite-Open100K with Hugging Face : 100,000 hours of egocentric human data, free for anyone to use. Physical AI has a scaling law of its own now, and its input is human data. That part is settled. What the field still needs is a shared foundation to build on. 15,000+ tasks. 15,000+ real scenes, from factory floors to retail backrooms. Hand pose, body pose and subtask-level semantics, with wrist camera coverage on part of the set. Licensed for commercial training, not research-only. We build the data infrastructure for Physical AI. Open100K is the first public layer of it. The first 10,000 hours are live now, with the rest rolling out in stages.

Lightwheel

75,420 görüntüleme • 1 ay önce

"Closed models are cheaper. The reason you need open models is that they give you control". - Nvidia CEO Jensen Huang "One of the things that people misunderstand about open models is this: yes, you can host them yourself, and you can build your own computer. But most people use computers in the cloud. Frankly, I think closed models are cheaper. If you do not build everything yourself, you avoid the significant cost and expertise required to train, fine-tune, maintain, guardrail, evaluate, and keep the model safe. You would also need to build or obtain the computers required to host it. So, there is nothing cheap about doing all of that yourself. The reason you need open models is that they give you control. You can adapt them for your own highly specialized use cases. I think there is a lot of misunderstanding about closed models versus open models. We felt it was important for people to understand that there is a place for both." ---- From "Bloomberg Podcasts" YouTube channel, (full video link in comment)

"Closed models are cheaper. The reason you need open models is that they give you control". - Nvidia CEO Jensen Huang "One of the things that people misunderstand about open models is this: yes, you can host them yourself, and you can build your own computer. But most people use computers in the cloud. Frankly, I think closed models are cheaper. If you do not build everything yourself, you avoid the significant cost and expertise required to train, fine-tune, maintain, guardrail, evaluate, and keep the model safe. You would also need to build or obtain the computers required to host it. So, there is nothing cheap about doing all of that yourself. The reason you need open models is that they give you control. You can adapt them for your own highly specialized use cases. I think there is a lot of misunderstanding about closed models versus open models. We felt it was important for people to understand that there is a place for both." ---- From "Bloomberg Podcasts" YouTube channel, (full video link in comment)

Rohan Paul

32,317 görüntüleme • 1 ay önce

We tried to record a normal AI launch video for this. And then we realized… The best way to showcase Moby 2 is just to let you have it. Because I genuinely think it’s the best product we’ve ever built at Triple Whale. It’s a DTC operator’s dream — a proactive agent that works for you. It can: - Monitor your ads and improve them - Launch and optimize campaigns - Run your Klaviyo flows - Generate creatives (images + video) The whole reason we founded Triple Whale was simple: 1. Get all your data in one place 2. Give you clarity 3. Help you automate and scale Moby 2 is the first time it really feels like we got there. Next week, we’ll onboard the first cohort of users and then expand as quickly as we can. But I cannot wait to get it in your hands. Request early access now →

We tried to record a normal AI launch video for this. And then we realized… The best way to showcase Moby 2 is just to let you have it. Because I genuinely think it’s the best product we’ve ever built at Triple Whale. It’s a DTC operator’s dream — a proactive agent that works for you. It can: - Monitor your ads and improve them - Launch and optimize campaigns - Run your Klaviyo flows - Generate creatives (images + video) The whole reason we founded Triple Whale was simple: 1. Get all your data in one place 2. Give you clarity 3. Help you automate and scale Moby 2 is the first time it really feels like we got there. Next week, we’ll onboard the first cohort of users and then expand as quickly as we can. But I cannot wait to get it in your hands. Request early access now →

AJ Orbach 🐳

168,805 görüntüleme • 7 ay önce

Hollywood isn't dead. It's evolving. And we're leading that evolution. We just shipped Koyal v2.5: The best Agentic AI filmmaking platform. It goes from your script or music to full video with consistent characters, settings & storylines. Large Production houses, music labels & ad agencies use koyal.ai (YC F25) for storyboarding & pre-viz. Smaller studios use it for making content they never had the time, budget or resources to create. Write a scene. Direct the camera. Build a world. koyal.ai (YC F25) brings it to life. Here's what's new: - Go from script to video: Write a scene with dialogue, pacing, emotion. Koyal fully realizes it. Using the best voice models on the planet. - Direct the camera: Push in. Pull back. Drift through a scene. Real 3D camera blocking. Shape the shot the way you see it in your head. - Build a world that stays:Lock in a location and return to it. Your scenes stay consistent from the first frame to the last. Plus: dialogue, SFX, annotations for editing, and more control over every shot. Under the hood, we benchmark 40+ models weekly with real humans so Koyal always uses the best available (more on that next week!) Our agents pick the best models for each scene in run-time so you focus on the story, we handle the complexity. You don't need to prompt a film, you can direct it. With Koyal, we're looking to replace the camera, not the filmmaker behind it. Try it now at beta [dot] koyal [dot] ai DM me for credits!

Hollywood isn't dead. It's evolving. And we're leading that evolution. We just shipped Koyal v2.5: The best Agentic AI filmmaking platform. It goes from your script or music to full video with consistent characters, settings & storylines. Large Production houses, music labels & ad agencies use koyal.ai (YC F25) for storyboarding & pre-viz. Smaller studios use it for making content they never had the time, budget or resources to create. Write a scene. Direct the camera. Build a world. koyal.ai (YC F25) brings it to life. Here's what's new: - Go from script to video: Write a scene with dialogue, pacing, emotion. Koyal fully realizes it. Using the best voice models on the planet. - Direct the camera: Push in. Pull back. Drift through a scene. Real 3D camera blocking. Shape the shot the way you see it in your head. - Build a world that stays:Lock in a location and return to it. Your scenes stay consistent from the first frame to the last. Plus: dialogue, SFX, annotations for editing, and more control over every shot. Under the hood, we benchmark 40+ models weekly with real humans so Koyal always uses the best available (more on that next week!) Our agents pick the best models for each scene in run-time so you focus on the story, we handle the complexity. You don't need to prompt a film, you can direct it. With Koyal, we're looking to replace the camera, not the filmmaker behind it. Try it now at beta [dot] koyal [dot] ai DM me for credits!

Mehul Agarwal

74,719 görüntüleme • 6 ay önce

The infrastructure for working with video data at scale barely exists. Grass Video Search is our first step in changing that. You can now find anything in a video based on the contents of its frames, not just transcripts or tags. Over the last several months, the Grass Foundation has quietly assembled one of the largest video datasets in history: over a billion videos from across the public web, and is segmenting them by scene with VLM annotations of keyframes. This dataset is now searchable, so you can find the exact moment a visual, pattern, or concept appears. Curation has always been the bottleneck in video analysis. We built a system that automates it. This is just the beginning. Over 4 billion additional videos are queued for processing, and the index is growing daily. We’re opening access to a limited set of collaborators. If you’re a researcher, journalist, or nonprofit working with video, archives, or large-scale analysis, we’d love to hear from you.

The infrastructure for working with video data at scale barely exists. Grass Video Search is our first step in changing that. You can now find anything in a video based on the contents of its frames, not just transcripts or tags. Over the last several months, the Grass Foundation has quietly assembled one of the largest video datasets in history: over a billion videos from across the public web, and is segmenting them by scene with VLM annotations of keyframes. This dataset is now searchable, so you can find the exact moment a visual, pattern, or concept appears. Curation has always been the bottleneck in video analysis. We built a system that automates it. This is just the beginning. Over 4 billion additional videos are queued for processing, and the index is growing daily. We’re opening access to a limited set of collaborators. If you’re a researcher, journalist, or nonprofit working with video, archives, or large-scale analysis, we’d love to hear from you.

Grass

471,801 görüntüleme • 1 yıl önce

Fck it. We no longer need Foundation. I built a free tool for every artist affected by the shutdown. Paste your wallet, see exactly what you created and what you collected - no wallet signature required. Two steps: 1. Paste your address. See every Foundation NFT you created or own, pulled straight from on-chain contract data. 2. Get a free Pinata account and pin your art to IPFS in one click. Your content, your storage, nobody can take it down. The tool also shows art you collected from other artists - so you can tell them before their work disappears. --- The Art Revolution starts now. So, I also built UNDERPIN to kick it off- a vision for a decentralized art marketplace with no company behind it. No gatekeepers, no single point of failure. AI dev tools exist now. Artists can run their own infrastructure. Foundation shutting down is not a loss - it's proof we need to own the stack. Everything is open source. Fork it, build on it, make it yours. links below.

Fck it. We no longer need Foundation. I built a free tool for every artist affected by the shutdown. Paste your wallet, see exactly what you created and what you collected - no wallet signature required. Two steps: 1. Paste your address. See every Foundation NFT you created or own, pulled straight from on-chain contract data. 2. Get a free Pinata account and pin your art to IPFS in one click. Your content, your storage, nobody can take it down. The tool also shows art you collected from other artists - so you can tell them before their work disappears. --- The Art Revolution starts now. So, I also built UNDERPIN to kick it off- a vision for a decentralized art marketplace with no company behind it. No gatekeepers, no single point of failure. AI dev tools exist now. Artists can run their own infrastructure. Foundation shutting down is not a loss - it's proof we need to own the stack. Everything is open source. Fork it, build on it, make it yours. links below.

tut™

32,370 görüntüleme • 6 ay önce

Introducing Praxis-1, an open-weight World Action Model that turns our video pre-training expertise into real world control for robots. Robot demonstration data is limited. But video data is infinite. Praxis-1 extends our bet that the best policy models will learn from video, and use that scale to bring embodied intelligence to every industry. Praxis-1 is built on our large-scale video pre-training and real-time infrastructure, and acts as a policy model for robotics developers and researchers that works across any embodiment or environment, no matter how complex. We are actively testing with early partners including Noble Machines, Standard Bots and Ultra, and will be releasing the weights publicly in the coming months. Learn more and request early access at

Introducing Praxis-1, an open-weight World Action Model that turns our video pre-training expertise into real world control for robots. Robot demonstration data is limited. But video data is infinite. Praxis-1 extends our bet that the best policy models will learn from video, and use that scale to bring embodied intelligence to every industry. Praxis-1 is built on our large-scale video pre-training and real-time infrastructure, and acts as a policy model for robotics developers and researchers that works across any embodiment or environment, no matter how complex. We are actively testing with early partners including Noble Machines, Standard Bots and Ultra, and will be releasing the weights publicly in the coming months. Learn more and request early access at

Runway

118,602 görüntüleme • 14 gün önce

Dario Amodei keeps warning that open-source AI models are dangerous and need to be restricted. David Sacks David Sacks answered with a rhetorical question: "Dangerous to whom?" It was the sharpest thing said on the latest All-In The All-In Podcast. Not dangerous to the enterprise that wants to keep control of its own data. Open weights are what finally give that enterprise a choice. It is dangerous to Anthropic's business model. That model only works if customers have no real alternative at the model layer. So the safety argument starts to look like a moat argument wearing a lab coat. Line it up with the rest of the strategy. Dominate the layer everyone builds on. Launch Claude into science, security, legal, finance, and design, the verticals your own customers used to own. And campaign to restrict the open models that would hand those customers the exit. Here is the giveaway. When the person selling you safety also needs you to have no other option, you are not who is being kept safe.

Dario Amodei keeps warning that open-source AI models are dangerous and need to be restricted. David Sacks David Sacks answered with a rhetorical question: "Dangerous to whom?" It was the sharpest thing said on the latest All-In The All-In Podcast. Not dangerous to the enterprise that wants to keep control of its own data. Open weights are what finally give that enterprise a choice. It is dangerous to Anthropic's business model. That model only works if customers have no real alternative at the model layer. So the safety argument starts to look like a moat argument wearing a lab coat. Line it up with the rest of the strategy. Dominate the layer everyone builds on. Launch Claude into science, security, legal, finance, and design, the verticals your own customers used to own. And campaign to restrict the open models that would hand those customers the exit. Here is the giveaway. When the person selling you safety also needs you to have no other option, you are not who is being kept safe.

Fireside Alpha

10,037 görüntüleme • 3 ay önce

anthropic just removed the biggest excuse people had for not building ai agents. “it’s too hard.” not anymore. a few months ago, building an ai agent meant learning how to code, setting up apis, configuring servers, and fixing endless errors. today, all you need is one github repo. anthropic open-sourced launch your agent. it asks what you want to build. then it does the rest. it builds your agent, deploys it to the cloud, tests it, improves it, and keeps it running even after you close your laptop. no fake demos. no complicated setup. just a real ai agent running in your own account. the best part. it doesn’t run on your computer. it runs inside claude managed agents, works 24/7, and costs just a few cents per run. people are already building research agents, lead generation systems, content pipelines, and customer support agents that save hours of work every day. the gap between people who use ai and people who build ai workers is getting bigger every week. that’s usually where the biggest opportunities are. build your first ai agent before everyone else does. watch the video and read the full guide below 👇

anthropic just removed the biggest excuse people had for not building ai agents. “it’s too hard.” not anymore. a few months ago, building an ai agent meant learning how to code, setting up apis, configuring servers, and fixing endless errors. today, all you need is one github repo. anthropic open-sourced launch your agent. it asks what you want to build. then it does the rest. it builds your agent, deploys it to the cloud, tests it, improves it, and keeps it running even after you close your laptop. no fake demos. no complicated setup. just a real ai agent running in your own account. the best part. it doesn’t run on your computer. it runs inside claude managed agents, works 24/7, and costs just a few cents per run. people are already building research agents, lead generation systems, content pipelines, and customer support agents that save hours of work every day. the gap between people who use ai and people who build ai workers is getting bigger every week. that’s usually where the biggest opportunities are. build your first ai agent before everyone else does. watch the video and read the full guide below 👇

MIKE

26,662 görüntüleme • 3 ay önce

Suella Braverman caused Britain's media class to crash out for a week, because she had the basic humility to admit that she isn't English. And Suella Braverman is right. Identity is not arbitrary. Ethnicity exists. The English are a people. You can no more change your ethnicity than you can change your sex. National identity is derived from ancestry. These are all facts, and no amount of name-calling or wishful thinking will change them. If we cede ground on this issue, and say that anyone can be English, we forfeit the language required to describe what has been done to our country, and the ability to restore its demographic and cultural security. How can we reverse mass migration if every member of the Boriswave is as English as you and me? We can't. They aren't. They don't need to be here. Recognising the difference between us and them is the prerequisite to sending them home. Which is why politicians and pundits are so desperate to insist that no such difference exists — despite recognising it themselves when politically expedient to identify as "ethnic minorities", and extract resources and special privileges from the white majority. We're not playing this game anymore. You cannot become English. My identity is not your cosplay. And the more you insist that it is, the less likely we are to advocate that you stay here.

Suella Braverman caused Britain's media class to crash out for a week, because she had the basic humility to admit that she isn't English. And Suella Braverman is right. Identity is not arbitrary. Ethnicity exists. The English are a people. You can no more change your ethnicity than you can change your sex. National identity is derived from ancestry. These are all facts, and no amount of name-calling or wishful thinking will change them. If we cede ground on this issue, and say that anyone can be English, we forfeit the language required to describe what has been done to our country, and the ability to restore its demographic and cultural security. How can we reverse mass migration if every member of the Boriswave is as English as you and me? We can't. They aren't. They don't need to be here. Recognising the difference between us and them is the prerequisite to sending them home. Which is why politicians and pundits are so desperate to insist that no such difference exists — despite recognising it themselves when politically expedient to identify as "ethnic minorities", and extract resources and special privileges from the white majority. We're not playing this game anymore. You cannot become English. My identity is not your cosplay. And the more you insist that it is, the less likely we are to advocate that you stay here.

Connor Tomlinson

12,877 görüntüleme • 6 gün önce

Just how capable are open source models? Below is the first in a new series where we go behind the scenes and pull back the curtain on interesting AI research / demos, making them fun and easy to understand. Here, we have a short visual demonstration from aizk ✡️ showcasing how Kimi K3 (a language model that operates primarily through text) is capable of building complicated 3D structures / moments in history in Minecraft, something that previously was not possible with other open source models, and why this matters. The crazy part? The model doesn't "see" the game like we do. The LLMs must reason in pure text, writing JavaScript, that later compiles down into commands placing each block, one at a time. Spatial reasoning is a very hard problem in AI, it's the same core challenge behind robotics and self-driving cars, where a model has to understand and act in physical 3D space. Watching a text model pull it off is nothing short of a miracle. The point isn't just Minecraft itself, rather, it's AI being able to generalize, not memorize, on things that are weird and beyond their training data. This is key to building true artificial general intelligence. These video game benchmarks (there are many different games actively being researched right now) provide a clear-cut end goal, challenges that are almost certainly not in the training set, and a fun, very fast, visual way to almost feel the increasing capabilities of various open source AI models over time. If you haven't given open source models a serious try yet, watch the video, it may shock you!

Just how capable are open source models? Below is the first in a new series where we go behind the scenes and pull back the curtain on interesting AI research / demos, making them fun and easy to understand. Here, we have a short visual demonstration from aizk ✡️ showcasing how Kimi K3 (a language model that operates primarily through text) is capable of building complicated 3D structures / moments in history in Minecraft, something that previously was not possible with other open source models, and why this matters. The crazy part? The model doesn't "see" the game like we do. The LLMs must reason in pure text, writing JavaScript, that later compiles down into commands placing each block, one at a time. Spatial reasoning is a very hard problem in AI, it's the same core challenge behind robotics and self-driving cars, where a model has to understand and act in physical 3D space. Watching a text model pull it off is nothing short of a miracle. The point isn't just Minecraft itself, rather, it's AI being able to generalize, not memorize, on things that are weird and beyond their training data. This is key to building true artificial general intelligence. These video game benchmarks (there are many different games actively being researched right now) provide a clear-cut end goal, challenges that are almost certainly not in the training set, and a fun, very fast, visual way to almost feel the increasing capabilities of various open source AI models over time. If you haven't given open source models a serious try yet, watch the video, it may shock you!

Featherless AI

39,769 görüntüleme • 1 ay önce

Elon Musk reduced the oldest question in human history to basic math. No one has found a flaw in it. Musk: “What are the odds that we are in base reality? And that this has not happened before.” You don’t need a physics degree to follow it. You need a timeline. Musk: “If you look at the advancement of video games, it’s gone from Pong, two rectangles and a square batting it back and forth, to photorealistic, real-time games with millions of people playing simultaneously.” Fifty years. That is all it took to close the gap between two rectangles on a screen and a world you cannot tell apart from the one outside your window. Musk: “If that trend continues, video games will be indistinguishable from reality.” The visuals are not what seals it. The intelligence is. Musk: “Think of how sophisticated the conversations are you can have with an AI today, and that’s only going to get more sophisticated.” We are not scripting characters anymore. We are building minds that reason, adapt, and surprise the people who made them. We are nowhere near finished. Musk: “The future, if civilization continues, will be millions, maybe billions of photorealistic, indistinguishable from reality, video games. And with characters in those video games that are very deep, and where the dialogue is not pre-programmed.” One base reality. Billions of perfect copies. Each one running minds that feel exactly as conscious as you do right now. Each one certain it is the original. Musk: “So then what are the odds that we are in base reality?” If even one civilization crosses that threshold, simulated minds outnumber real ones by billions. The probability you are sitting in the real one is not low. It is nearly zero. Not as philosophy. As mathematics. We are not watching this happen. We are building it. Right now. Every AI that reasons without a script. Every world rendered one frame closer to indistinguishable. We are constructing the exact technology that makes our own existence statistically implausible. And we will never stop. Because the curiosity that questions reality is the same force that builds it. If the math holds, something built us. Something conscious enough to create consciousness. They stood where we are standing. Same question. Same inability to stop. And whatever built them never answered it either. There is no top floor. There is no original. None of that changes what you feel right now. Consciousness was never about what you are made of. It was about what you experience. Musk did not float a theory. He held up a mirror with no back wall. And the math does not need you to believe it. It only needs time.

Elon Musk reduced the oldest question in human history to basic math. No one has found a flaw in it. Musk: “What are the odds that we are in base reality? And that this has not happened before.” You don’t need a physics degree to follow it. You need a timeline. Musk: “If you look at the advancement of video games, it’s gone from Pong, two rectangles and a square batting it back and forth, to photorealistic, real-time games with millions of people playing simultaneously.” Fifty years. That is all it took to close the gap between two rectangles on a screen and a world you cannot tell apart from the one outside your window. Musk: “If that trend continues, video games will be indistinguishable from reality.” The visuals are not what seals it. The intelligence is. Musk: “Think of how sophisticated the conversations are you can have with an AI today, and that’s only going to get more sophisticated.” We are not scripting characters anymore. We are building minds that reason, adapt, and surprise the people who made them. We are nowhere near finished. Musk: “The future, if civilization continues, will be millions, maybe billions of photorealistic, indistinguishable from reality, video games. And with characters in those video games that are very deep, and where the dialogue is not pre-programmed.” One base reality. Billions of perfect copies. Each one running minds that feel exactly as conscious as you do right now. Each one certain it is the original. Musk: “So then what are the odds that we are in base reality?” If even one civilization crosses that threshold, simulated minds outnumber real ones by billions. The probability you are sitting in the real one is not low. It is nearly zero. Not as philosophy. As mathematics. We are not watching this happen. We are building it. Right now. Every AI that reasons without a script. Every world rendered one frame closer to indistinguishable. We are constructing the exact technology that makes our own existence statistically implausible. And we will never stop. Because the curiosity that questions reality is the same force that builds it. If the math holds, something built us. Something conscious enough to create consciousness. They stood where we are standing. Same question. Same inability to stop. And whatever built them never answered it either. There is no top floor. There is no original. None of that changes what you feel right now. Consciousness was never about what you are made of. It was about what you experience. Musk did not float a theory. He held up a mirror with no back wall. And the math does not need you to believe it. It only needs time.

Dustin

194,705 görüntüleme • 3 ay önce

Domain Randomization (DR) is a key component of the data augmentation pipeline at Axis Robotics. By applying DR, we are able to scale verified, high-quality human trajectories by 10x to 100x. During training, we systematically introduce variances in environmental parameters. This prevents the model from relying on spurious visual correlations. The objective is to ensure the policy learns rather than overfitting. To demonstrate the necessity and effectiveness of this approach, we evaluated both DR and No-DR models on Task 74 (pour_water_into_mug). The empirical results show a definitive impact on real-world deployment reliability: integrating DR into the pipeline increased the success rate from 0% to 90% (Fig. 1). This divergence stems from how the respective policies process visual observations (Fig. 2). The baseline (No DR) model overfits to the static visual background. It essentially memorizes the poses from the training dataset but fails to generalize when subjected to the inevitable variances of real-world deployment. Consequently, it cannot execute the correct manipulation on the target object. Conversely, the DR-trained model learns to extract essential geometric features and physical constraints, filtering out superficial visual noise. This leads to significantly higher robustness in dynamic environments. The structural difference in execution is clearly reflected in the end-effector trajectory data: These real-world deployment recordings further illustrate this difference (Videos 1 and 2). Scaling Physical AI requires turning raw trajectory data into robust policies, and a rigorously engineered DR infrastructure is an essential bridge to close the Sim2Real gap.

Domain Randomization (DR) is a key component of the data augmentation pipeline at Axis Robotics. By applying DR, we are able to scale verified, high-quality human trajectories by 10x to 100x. During training, we systematically introduce variances in environmental parameters. This prevents the model from relying on spurious visual correlations. The objective is to ensure the policy learns rather than overfitting. To demonstrate the necessity and effectiveness of this approach, we evaluated both DR and No-DR models on Task 74 (pour_water_into_mug). The empirical results show a definitive impact on real-world deployment reliability: integrating DR into the pipeline increased the success rate from 0% to 90% (Fig. 1). This divergence stems from how the respective policies process visual observations (Fig. 2). The baseline (No DR) model overfits to the static visual background. It essentially memorizes the poses from the training dataset but fails to generalize when subjected to the inevitable variances of real-world deployment. Consequently, it cannot execute the correct manipulation on the target object. Conversely, the DR-trained model learns to extract essential geometric features and physical constraints, filtering out superficial visual noise. This leads to significantly higher robustness in dynamic environments. The structural difference in execution is clearly reflected in the end-effector trajectory data: These real-world deployment recordings further illustrate this difference (Videos 1 and 2). Scaling Physical AI requires turning raw trajectory data into robust policies, and a rigorously engineered DR infrastructure is an essential bridge to close the Sim2Real gap.

Axis Robotics

27,125 görüntüleme • 6 ay önce

Today, we are launching Depost AI 2. 🎉 I am a bit emotional writing this. The last eight months were intense. But today I am proud of what we built. Because Depost AI 2 is built to become the best and safest AI content team for LinkedIn. We listened to thousands of people who used the first version, and we rebuilt Depost AI from scratch on the official LinkedIn API. The result: - A chat instead of a form, so you brief it like a colleague - Analytics that show what performs for you and why - Four agents that write in your voice and design the graphics to match - A calendar that fills from approved drafts - Approvals, roles, and a separate workspace for every brand - MCP, so the whole team works inside Claude, ChatGPT, or Cursor It is built for B2B founders who need to be visible without losing an hour a day. For marketing teams who need one system, one review flow, and one voice per brand. And for agencies who run LinkedIn for clients and need every brand kept apart. But the bigger vision is this: Depost AI is becoming a content team that understands your brand before it writes a word. It learns from everything you publish. It knows what performs for you and why. It plans, writes, designs, and schedules from that. And nothing goes live until you approve it. Because writing a post is not the hard part. Knowing what to say, saying it well every week, and doing it safely on an account you cannot afford to lose is. Depost AI 2 is the tool we built to solve that. And honestly, the best part right now is watching people brief it for the first time and get a week of drafts back in their own voice. You can start a 7-day free trial today at And if you message me, I will happily walk you through it myself. ❤️

Atta 🐬

20,330 görüntüleme • 1 ay önce

Yann LeCun just told the most well-funded industry in human history it is solving the wrong problem. LeCun: “Babies learn this around the age of eight or nine months, that objects don’t float, they fall.” No dataset. No labels. No reward signal. A nine month old drops a spoon and builds a physics engine no machine can match. LeCun: “Most of us can learn to drive in about 20 or 30 hours of training without ever crashing, causing any accident.” Twenty hours. Tesla has built the most capable driving system on the road. It took billions of miles of data to get there. A sixteen year old gets there over a long weekend. Not because the teenager is the better driver. Because the teenager is not learning to drive. They are deploying a model of reality they have been building since birth. LeCun: “If we drive next to a cliff, we know that if we turn the wheel to the right, the car is going to run off the cliff and nothing good is going to come out of this.” You simulate the crash. You see the wreckage. You feel the fall. You turn the wheel. None of it was real. All of it was intelligence. Every AI has to crash a thousand times to learn what you imagined once and never did. That is not a performance gap. That is an architecture gap. LeCun: “The main problem we need to solve is how do we learn models of the world.” Not bigger models. Not more compute. Not another trillion tokens. World models. A machine that can run reality forward before it acts. The industry is scaling language. LeCun says language is a compression of thought. Not thought itself. You understood gravity before you could say the word. You grasped cause and effect before your first sentence. The deepest intelligence you will ever possess was built in total silence. And every lab on Earth is trying to reconstruct the mind from words alone. Physics does not care about your context window. A baby who learns that cups fall in a kitchen already knows that rocks fall off cliffs. No retraining. No fine-tuning. One model. Every environment. That is what intelligence actually is. Not prediction. Not pattern matching. Not scale. A simulation of reality so precise you rehearse the future before it exists. Every infant on Earth builds one. No machine ever has.

Yann LeCun just told the most well-funded industry in human history it is solving the wrong problem. LeCun: “Babies learn this around the age of eight or nine months, that objects don’t float, they fall.” No dataset. No labels. No reward signal. A nine month old drops a spoon and builds a physics engine no machine can match. LeCun: “Most of us can learn to drive in about 20 or 30 hours of training without ever crashing, causing any accident.” Twenty hours. Tesla has built the most capable driving system on the road. It took billions of miles of data to get there. A sixteen year old gets there over a long weekend. Not because the teenager is the better driver. Because the teenager is not learning to drive. They are deploying a model of reality they have been building since birth. LeCun: “If we drive next to a cliff, we know that if we turn the wheel to the right, the car is going to run off the cliff and nothing good is going to come out of this.” You simulate the crash. You see the wreckage. You feel the fall. You turn the wheel. None of it was real. All of it was intelligence. Every AI has to crash a thousand times to learn what you imagined once and never did. That is not a performance gap. That is an architecture gap. LeCun: “The main problem we need to solve is how do we learn models of the world.” Not bigger models. Not more compute. Not another trillion tokens. World models. A machine that can run reality forward before it acts. The industry is scaling language. LeCun says language is a compression of thought. Not thought itself. You understood gravity before you could say the word. You grasped cause and effect before your first sentence. The deepest intelligence you will ever possess was built in total silence. And every lab on Earth is trying to reconstruct the mind from words alone. Physics does not care about your context window. A baby who learns that cups fall in a kitchen already knows that rocks fall off cliffs. No retraining. No fine-tuning. One model. Every environment. That is what intelligence actually is. Not prediction. Not pattern matching. Not scale. A simulation of reality so precise you rehearse the future before it exists. Every infant on Earth builds one. No machine ever has.

Dustin

122,476 görüntüleme • 3 ay önce

⚠️🚨STATEMENT REGARDING CHEATING A few days ago, we played a match in the Champion of Champions Tour CS2 tournament against the team Khan, and we have strong reasons to believe that this team does not win matches in the spirit of fair play. As you can see in the attached video, rotations are occurring without any information, just like in many other rounds, and also involve locking through walls, etc. This is not the first time suspicions of cheating have appeared on the Tier 2/3 scene. We all surely remember, for example, the team Senza. We decided to wait a few days to see how the situation would develop, but we can no longer remain silent. In the meantime, matches have taken place against teams such as EYEBALLERS and SINNERS Esports, where the same situations occurred as in our game. Therefore, we kindly ask all organisers, especially Champion of Champions Tour CS2, to put an end to this, to actively investigate such cases, and to stop inviting suspicious teams to their tournaments. A loss is a loss, but it must be fair — that’s what esports is about. By ignoring this problem, we are hurting not only the players but also the fans, who should be at the heart of it all. LET’S HELP EACH OTHER AND MAKE THE TIER 2/3 SCENE FAIR-PLAY AGAIN, SO THAT WE CAN ENJOY THE GAME! Thank you, UNiTY Esports

⚠️🚨STATEMENT REGARDING CHEATING A few days ago, we played a match in the Champion of Champions Tour CS2 tournament against the team Khan, and we have strong reasons to believe that this team does not win matches in the spirit of fair play. As you can see in the attached video, rotations are occurring without any information, just like in many other rounds, and also involve locking through walls, etc. This is not the first time suspicions of cheating have appeared on the Tier 2/3 scene. We all surely remember, for example, the team Senza. We decided to wait a few days to see how the situation would develop, but we can no longer remain silent. In the meantime, matches have taken place against teams such as EYEBALLERS and SINNERS Esports, where the same situations occurred as in our game. Therefore, we kindly ask all organisers, especially Champion of Champions Tour CS2, to put an end to this, to actively investigate such cases, and to stop inviting suspicious teams to their tournaments. A loss is a loss, but it must be fair — that’s what esports is about. By ignoring this problem, we are hurting not only the players but also the fans, who should be at the heart of it all. LET’S HELP EACH OTHER AND MAKE THE TIER 2/3 SCENE FAIR-PLAY AGAIN, SO THAT WE CAN ENJOY THE GAME! Thank you, UNiTY Esports

UNiTY ESPORTS

178,991 görüntüleme • 1 yıl önce

Today we're launching iMessage video ads in Claude. Faceless video is the best-performing ad format right now — and Claude can now generate a full one end to end, in one prompt. Here's how it works: 1. Install: npx gooseworks install --all 2. Run: /goose-video make an iMessage ad for [your product] Claude writes the script, generates the photos in the thread, animates the conversation, and closes on your brand. SFX, music, end card, all of it. The trick: it's just HTML. No video generation models, so it's fast and cheap — and it works because it doesn't look like an ad. It's that easy. But this is just the start. We have created a library with 100+ open-source skills for growth that some of the fastest growing startups in the world are using every day to run ads, content, competitor research, gtm, seo and more. Comment Goose and I'll DM you the full open-source skill library.

Soham Mehta

50,641 görüntüleme • 3 ay önce

Code and data are now online for CameraHMR, our state-of-the-art parametric 3D human pose and shape (HPS) estimation method that will appear at hashtag#3DV2025. There are 4 key contributions that make it so accurate and robust: 1. To get accurate 3D shape and pose as well as good alignment to image features, you need to know the focal length of the camera. To solve this, we train HumanFOV to compute the field of view. 2. We introduce CameraHMR, which integrates HumanFOV into HMR2.0 to exploit the estimated focal length. 3. To get accurate pseudo ground truth (pGT) training data, we compute the focal length for images in 4DHumans dataset and modify SMPLify to take this into account. 4. But SMPLify only uses sparse 2D keypoints, which do not capture body shape. So we train a dense surface keypoint detector, DenseKP, on BEDLAM and run it on 4DHumans, resulting in improved body shape. The resulting method is CamSMPLify. We iterate training CameraHMR and running CamSMPLify on the training set initialized with CameraHMR. This results in much improved pGT for 4DHumans and a SOTA single-image HMR method.

Code and data are now online for CameraHMR, our state-of-the-art parametric 3D human pose and shape (HPS) estimation method that will appear at hashtag#3DV2025. There are 4 key contributions that make it so accurate and robust: 1. To get accurate 3D shape and pose as well as good alignment to image features, you need to know the focal length of the camera. To solve this, we train HumanFOV to compute the field of view. 2. We introduce CameraHMR, which integrates HumanFOV into HMR2.0 to exploit the estimated focal length. 3. To get accurate pseudo ground truth (pGT) training data, we compute the focal length for images in 4DHumans dataset and modify SMPLify to take this into account. 4. But SMPLify only uses sparse 2D keypoints, which do not capture body shape. So we train a dense surface keypoint detector, DenseKP, on BEDLAM and run it on 4DHumans, resulting in improved body shape. The resulting method is CamSMPLify. We iterate training CameraHMR and running CamSMPLify on the training set initialized with CameraHMR. This results in much improved pGT for 4DHumans and a SOTA single-image HMR method.

Michael Black

21,696 görüntüleme • 1 yıl önce