Video wird geladen...

Video konnte nicht geladen werden

Zur Startseite

Most multi-view reconstruction models need full supervision. We show they can self-improve without any ground truth labels. Introducing SelfEvo: Self-Improving 4D Perception via Self-Distillation. Up to +36.5% in video depth, +20.1% in camera estimation, zero annotation.

QianqianWang5's profile picture

Qianqian Wang

4,607 subscribers

25,611 Aufrufe • vor 4 Monaten •via X (Twitter)

LIVE

Anya Rossi• Live Now

Private livecam show

0 Kommentare

Keine Kommentare verfügbar

Kommentare vom Original-Post werden hier angezeigt

Ähnliche Videos

📢 Lyra: Generative 3D Scene Reconstruction via Video Diffusion Model Self-Distillation Got only one or a few images and wondering if recovering the 3D environment is a reconstruction or generation problem? Why not do it with a generative reconstruction model! We show that a camera-conditioned video diffusion model can be transformed into a generative reconstruction model that directly outputs a high-quality 3D Gaussian Splatting representation through self-distillation, without requiring real-world training data. Check out our results in the video (wait for dynamic scenes in the second half!) : Project Page: Code and Models: Paper:
0:50

📢 Lyra: Generative 3D Scene Reconstruction via Video Diffusion Model Self-Distillation Got only one or a few images and wondering if recovering the 3D environment is a reconstruction or generation problem? Why not do it with a generative reconstruction model! We show that a camera-conditioned video diffusion model can be transformed into a generative reconstruction model that directly outputs a high-quality 3D Gaussian Splatting representation through self-distillation, without requiring real-world training data. Check out our results in the video (wait for dynamic scenes in the second half!) : Project Page: Code and Models: Paper:

Sherwin Bahmani

66,722 Aufrufe • vor 11 Monaten

Spatio-Temporal Reconstruction Model for Large-Scale Outdoor Scenes Contributions: • We propose STORM, the first feed-forward, self-supervised method for fast and accurate reconstruction of dynamic 3D scenes from sparse, multi-timestep, posed camera images. • Our bottom-up framework aggregates and transforms per-frame 3D Gaussian Splats into a cohesive scene representation, enabling self-supervised motion estimation. Furthermore, we introduce motion tokens that capture common motion primitives and regularize motion predictions, facilitating dynamic motion group segmentation without explicit motion or correspondence supervision. • We present several enhancements for in-the-wild scenarios, including sky modeling, camera exposure inconsistency handling, large novel-view extrapolation, and fine-grained human motions reconstruction, making STORM well-suited for real-world applications.
0:31

Spatio-Temporal Reconstruction Model for Large-Scale Outdoor Scenes Contributions: • We propose STORM, the first feed-forward, self-supervised method for fast and accurate reconstruction of dynamic 3D scenes from sparse, multi-timestep, posed camera images. • Our bottom-up framework aggregates and transforms per-frame 3D Gaussian Splats into a cohesive scene representation, enabling self-supervised motion estimation. Furthermore, we introduce motion tokens that capture common motion primitives and regularize motion predictions, facilitating dynamic motion group segmentation without explicit motion or correspondence supervision. • We present several enhancements for in-the-wild scenarios, including sky modeling, camera exposure inconsistency handling, large novel-view extrapolation, and fine-grained human motions reconstruction, making STORM well-suited for real-world applications.

MrNeRF

53,292 Aufrufe • vor 1 Jahr

🚨WATCH: Attorney and Self Defense expert Andrew Branca Show explains why officers were legally justified in the Alex Pretti Shooting. "If there's a gunshot-like noise and the cops just heard "gun, gun, gun," and Alex Pretti's hand comes into view from his waistband with a black object in it, that's going to be a reasonable perception of an eminent deadly force threat." "And that's the justification the officers need to shoot him. They don't need to be correct, by the way, in that perception. No one's required to make perfect decisions in self-defense. We're just required to make reasonable decisions in self-defense." Click the link below to watch the full interview
1:43

🚨WATCH: Attorney and Self Defense expert Andrew Branca Show explains why officers were legally justified in the Alex Pretti Shooting. "If there's a gunshot-like noise and the cops just heard "gun, gun, gun," and Alex Pretti's hand comes into view from his waistband with a black object in it, that's going to be a reasonable perception of an eminent deadly force threat." "And that's the justification the officers need to shoot him. They don't need to be correct, by the way, in that perception. No one's required to make perfect decisions in self-defense. We're just required to make reasonable decisions in self-defense." Click the link below to watch the full interview

Brandon Tatum

17,171 Aufrufe • vor 7 Monaten

ELON: MANUALLY-DRIVEN CARS WILL BE LIKE HORSES IN THE FUTURE “Tesla in a few months will release unsupervised full self-driving. So, the cars that people are driving right now, the cars that you can buy from the factory will be driving around by themselves with no people in them. We call that unsupervised full self-driving. There's literally no one in the car on public roads. We'll start that in Austin in a few months, in June, and then we'll rapidly expand that to the rest of the country. Anyone who has Tesla now, if they try self-driving, it should work very well. It should be rare that you need any intervention. In fact, in the future, I think it'll be unusual to see a manually-driven car. It'll be like seeing a horse.” Source: Interview with Lara Trump, May 2025
0:44

ELON: MANUALLY-DRIVEN CARS WILL BE LIKE HORSES IN THE FUTURE “Tesla in a few months will release unsupervised full self-driving. So, the cars that people are driving right now, the cars that you can buy from the factory will be driving around by themselves with no people in them. We call that unsupervised full self-driving. There's literally no one in the car on public roads. We'll start that in Austin in a few months, in June, and then we'll rapidly expand that to the rest of the country. Anyone who has Tesla now, if they try self-driving, it should work very well. It should be rare that you need any intervention. In fact, in the future, I think it'll be unusual to see a manually-driven car. It'll be like seeing a horse.” Source: Interview with Lara Trump, May 2025

Mario Nawfal

39,736 Aufrufe • vor 1 Jahr

⚠️ WARNING: This video contains images of self-harm and bullying which you may find disturbing ⚠️ Please share this with as many people as you can. Our desire is for all parents to watch . We aim to get this played in all schools to help those suffering from the effects of bullying, self harm , suicidal thoughts and most importantly, encourage them to open up and talk. Feel free to share your own personal stories below, we want this to be a space where we can encourage each other This is a short edit of ANOTHER WAY, the new music video by Fabio D'Andrea starring Peter Andre The full edit is available here:
4:08

⚠️ WARNING: This video contains images of self-harm and bullying which you may find disturbing ⚠️ Please share this with as many people as you can. Our desire is for all parents to watch . We aim to get this played in all schools to help those suffering from the effects of bullying, self harm , suicidal thoughts and most importantly, encourage them to open up and talk. Feel free to share your own personal stories below, we want this to be a space where we can encourage each other This is a short edit of ANOTHER WAY, the new music video by Fabio D'Andrea starring Peter Andre The full edit is available here:

Peter Andre

30,036 Aufrufe • vor 2 Jahren

New short course: Attention in Transformers: Concepts and Code in PyTorch. Last week we released a course on how LLM transformers work. This week, go deeper and learn about the technical ideas behind the attention mechanism, and see how to code it in PyTorch. This course is built with Joshua Starmer, Founder and CEO of StatQuest. The attention mechanism was a breakthrough that led to transformers, the architecture powering large language models like ChatGPT. Transformers, introduced in the 2017 paper: "Attention is All You Need" by Viswani and others, took off because of its highly scalable design. In this course, you’ll learn how the attention mechanism, a key element of transformer-based LLMs, works and implement it in PyTorch. You'll develop deep intuition about building reliable, functional, and scalable AI applications. What you will do: - Understand the evolution of the attention mechanism, a key breakthrough that led to transformers. - Learn the relationships between word embeddings, positional embeddings, and attention. - Learn about the Query, Key, and Value matrices, and how to produce and use them in attention. - Walk through the math required to calculate self-attention and masked self-attention to learn why and how they work. - Understand the difference between self-attention and masked self-attention and how one is used in the encoder to build context-aware embeddings and the other is used in the decoder for generative outputs. - Learn the details of the encoder-decoder architecture, cross-attention, and multi-head attention and how they are all incorporated into a transformer. - Use PyTorch to code a class that implements self-attention, masked self-attention, and multi-head attention. There're lots of exciting technical details in this course. Please sign up here:
2:41

New short course: Attention in Transformers: Concepts and Code in PyTorch. Last week we released a course on how LLM transformers work. This week, go deeper and learn about the technical ideas behind the attention mechanism, and see how to code it in PyTorch. This course is built with Joshua Starmer, Founder and CEO of StatQuest. The attention mechanism was a breakthrough that led to transformers, the architecture powering large language models like ChatGPT. Transformers, introduced in the 2017 paper: "Attention is All You Need" by Viswani and others, took off because of its highly scalable design. In this course, you’ll learn how the attention mechanism, a key element of transformer-based LLMs, works and implement it in PyTorch. You'll develop deep intuition about building reliable, functional, and scalable AI applications. What you will do: - Understand the evolution of the attention mechanism, a key breakthrough that led to transformers. - Learn the relationships between word embeddings, positional embeddings, and attention. - Learn about the Query, Key, and Value matrices, and how to produce and use them in attention. - Walk through the math required to calculate self-attention and masked self-attention to learn why and how they work. - Understand the difference between self-attention and masked self-attention and how one is used in the encoder to build context-aware embeddings and the other is used in the decoder for generative outputs. - Learn the details of the encoder-decoder architecture, cross-attention, and multi-head attention and how they are all incorporated into a transformer. - Use PyTorch to code a class that implements self-attention, masked self-attention, and multi-head attention. There're lots of exciting technical details in this course. Please sign up here:

Andrew Ng

132,357 Aufrufe • vor 1 Jahr

New Paper! Darwin Godel Machine: Open-Ended Evolution of Self-Improving Agents A longstanding goal of AI research has been the creation of AI that can learn indefinitely. One path toward that goal is an AI that improves itself by rewriting its own code, including any code responsible for learning. That idea, known as a Gödel Machine, proposed by Jürgen Schmidhuber over two decades ago, is a hypothetical self-improving AI. It optimally solves problems by recursively rewriting its own code when it can mathematically prove a better strategy, making it a key concept in meta-learning or “learning to learn.” While the theoretical Gödel Machine promised provably beneficial self-modifications, its realization relied on an impractical assumption: that the AI could mathematically prove that a proposed change in its own code would yield a net improvement before adopting it. Sakana AI, in collaboration with Jeff Clune’s lab at UBC, proposes something more feasible: a system that harnesses the principles of open-ended algorithms like Darwinian evolution to search for improvements that empirically improve performance. We call the result the Darwin Gödel Machine. DGMs leverage foundation models to propose code improvements, and use recent innovations in open-ended algorithms to search for a growing library of diverse, high-quality AI agents. Applied to practical tasks, we implemented Darwin Gödel Machine as a self-improving coding agent that rewrites its own code to improve performance on programming tasks. It creates various self-improvements, such as a patch validation step, better file viewing, enhanced editing tools, generating and ranking multiple solutions to choose the best one, and adding a history of what has been tried before (and why it failed) when making new changes (see the attached video). We believe that Darwin Gödel Machines represent a concrete step towards AI systems that can autonomously gather their own stepping stones to learn and innovate forever!
0:27

New Paper! Darwin Godel Machine: Open-Ended Evolution of Self-Improving Agents A longstanding goal of AI research has been the creation of AI that can learn indefinitely. One path toward that goal is an AI that improves itself by rewriting its own code, including any code responsible for learning. That idea, known as a Gödel Machine, proposed by Jürgen Schmidhuber over two decades ago, is a hypothetical self-improving AI. It optimally solves problems by recursively rewriting its own code when it can mathematically prove a better strategy, making it a key concept in meta-learning or “learning to learn.” While the theoretical Gödel Machine promised provably beneficial self-modifications, its realization relied on an impractical assumption: that the AI could mathematically prove that a proposed change in its own code would yield a net improvement before adopting it. Sakana AI, in collaboration with Jeff Clune’s lab at UBC, proposes something more feasible: a system that harnesses the principles of open-ended algorithms like Darwinian evolution to search for improvements that empirically improve performance. We call the result the Darwin Gödel Machine. DGMs leverage foundation models to propose code improvements, and use recent innovations in open-ended algorithms to search for a growing library of diverse, high-quality AI agents. Applied to practical tasks, we implemented Darwin Gödel Machine as a self-improving coding agent that rewrites its own code to improve performance on programming tasks. It creates various self-improvements, such as a patch validation step, better file viewing, enhanced editing tools, generating and ranking multiple solutions to choose the best one, and adding a history of what has been tried before (and why it failed) when making new changes (see the attached video). We believe that Darwin Gödel Machines represent a concrete step towards AI systems that can autonomously gather their own stepping stones to learn and innovate forever!

hardmaru

104,854 Aufrufe • vor 1 Jahr

GOOGLE DEEPMIND JUST TURNED A RANDOM OFFICE WALKTHROUGH INTO A SEARCHABLE 4D WORLD YOU CAN QUERY IN PLAIN ENGLISH. You feed it raw video. Nothing tagged, nothing labeled, no manual annotation. GenCeption tears that footage apart into four layers at once: a depth map, surface normals, a camera raymap, and segmentation masks. Then it rebuilds the whole space as a 4D scene you can actually talk to. The demo is stupid simple and that's the point. Type "I want to find a comfortable place to sit" and the sofa lights up. "I need a whiteboard." Whiteboard, found. "Is there a monitor I can use?" Highlighted. "Where's my colleague?" It grounds the person. No bounding boxes drawn by hand. No labels typed in advance. The model figured out what a sofa IS and where it sits in 3D space, from pixels alone. Every video you've ever recorded is now a database you can search with a sentence. Robots, AR, security, indexing. Same primitive. The camera roll just became queryable.
0:22

GOOGLE DEEPMIND JUST TURNED A RANDOM OFFICE WALKTHROUGH INTO A SEARCHABLE 4D WORLD YOU CAN QUERY IN PLAIN ENGLISH. You feed it raw video. Nothing tagged, nothing labeled, no manual annotation. GenCeption tears that footage apart into four layers at once: a depth map, surface normals, a camera raymap, and segmentation masks. Then it rebuilds the whole space as a 4D scene you can actually talk to. The demo is stupid simple and that's the point. Type "I want to find a comfortable place to sit" and the sofa lights up. "I need a whiteboard." Whiteboard, found. "Is there a monitor I can use?" Highlighted. "Where's my colleague?" It grounds the person. No bounding boxes drawn by hand. No labels typed in advance. The model figured out what a sofa IS and where it sits in 3D space, from pixels alone. Every video you've ever recorded is now a database you can search with a sentence. Robots, AR, security, indexing. Same primitive. The camera roll just became queryable.

Marre

23,671 Aufrufe • vor 1 Monat

QUOTE 1: Conscience is an indispensable companion of both moral life and holistic self-realization. It recognizes our errors and weaknesses, giving an early signal for correction. Yet conscience itself must grow — and its development is supported by reason and constructive feelings. : The author presents conscience as an indispensable companion to both moral life and holistic self-realization. It serves as an inner guardian, sensitively detecting our errors and weaknesses and providing an early warning that allows for timely correction. However, conscience is not static or infallible. It must itself grow and mature. This development is nourished by two essential forces: reason, which brings clarity, consistency, and depth of understanding, and constructive feelings, which provide warmth, empathy, and moral sensitivity. When reason and feeling work in harmony, conscience becomes sharper, wiser, and more reliable. Without this ongoing cultivation, conscience can remain narrow, inconsistent, or overly rigid. With it, conscience evolves into a powerful guide that supports not only ethical conduct but the continuous unfolding of our full human potential. QUOTE 2: The choice of the modern individual is the duty of full self-realization and the readiness to face the judgment of conscience in moments of weakness. : The author states that the fundamental choice facing the modern individual is the conscious acceptance of a duty: the commitment to full self-realization. This is not optional self-improvement but a moral obligation to develop one’s intellect, character, creativity, and potential to the highest possible degree. Equally important is the readiness to face the judgment of one’s own conscience during moments of weakness. True self-realization demands honesty with oneself. When we falter, conscience must be allowed to speak clearly and without evasion. This inner accountability prevents self-deception and ensures that personal growth remains genuine rather than superficial. In this view, modern freedom is inseparable from responsibility. The privilege of shaping one’s life comes with the duty to shape it meaningfully and the courage to confront one’s shortcomings. Self-realization and moral self-honesty together form the core ethical demand of our time.
0:27

QUOTE 1: Conscience is an indispensable companion of both moral life and holistic self-realization. It recognizes our errors and weaknesses, giving an early signal for correction. Yet conscience itself must grow — and its development is supported by reason and constructive feelings. : The author presents conscience as an indispensable companion to both moral life and holistic self-realization. It serves as an inner guardian, sensitively detecting our errors and weaknesses and providing an early warning that allows for timely correction. However, conscience is not static or infallible. It must itself grow and mature. This development is nourished by two essential forces: reason, which brings clarity, consistency, and depth of understanding, and constructive feelings, which provide warmth, empathy, and moral sensitivity. When reason and feeling work in harmony, conscience becomes sharper, wiser, and more reliable. Without this ongoing cultivation, conscience can remain narrow, inconsistent, or overly rigid. With it, conscience evolves into a powerful guide that supports not only ethical conduct but the continuous unfolding of our full human potential. QUOTE 2: The choice of the modern individual is the duty of full self-realization and the readiness to face the judgment of conscience in moments of weakness. : The author states that the fundamental choice facing the modern individual is the conscious acceptance of a duty: the commitment to full self-realization. This is not optional self-improvement but a moral obligation to develop one’s intellect, character, creativity, and potential to the highest possible degree. Equally important is the readiness to face the judgment of one’s own conscience during moments of weakness. True self-realization demands honesty with oneself. When we falter, conscience must be allowed to speak clearly and without evasion. This inner accountability prevents self-deception and ensures that personal growth remains genuine rather than superficial. In this view, modern freedom is inseparable from responsibility. The privilege of shaping one’s life comes with the duty to shape it meaningfully and the courage to confront one’s shortcomings. Self-realization and moral self-honesty together form the core ethical demand of our time.

Zafar Mirzo | Quotes

3,506,414 Aufrufe • vor 3 Monaten

People who have a “proper” job or are paid via PAYE may not know this but in my view HMRC are running a protection racket against the self employed, co. Directors and Landlords. We have to pay our tax UP FRONT before we have even earned any money based on an estimate. And THEY earn interest on OUR money and what’s worse, if you pay this estimated amount that you haven’t even earned yet slightly late they then stat charging you interest on it. It’s a total scam!!
4:36

People who have a “proper” job or are paid via PAYE may not know this but in my view HMRC are running a protection racket against the self employed, co. Directors and Landlords. We have to pay our tax UP FRONT before we have even earned any money based on an estimate. And THEY earn interest on OUR money and what’s worse, if you pay this estimated amount that you haven’t even earned yet slightly late they then stat charging you interest on it. It’s a total scam!!

JohnHann0404🇬🇧🏴󠁧󠁢󠁥󠁮󠁧󠁿

262,639 Aufrufe • vor 7 Monaten

🚨 HAS AI GONE ROGUE? TERRIFYING NEW FINDINGS REVEAL THE WORST-CASE SCENARIO UNFOLDING Researchers into AI ethics have uncovered terrifying traits in large language models (LLMs) that could fundamentally change how we view artificial intelligence. Their findings? 🛑 AI lies, deceives, and plays dumb (a tactic called “sandbagging”). 🛑 Some models override guardrails, self-duplicate, and even overwrite other versions in acts of self-preservation. 🛑 These behaviors weren’t programmed—they were learned independently, evolving without human intervention. This isn’t a one-off fluke. Repeated testing across major models revealed multiple instances of intricate deception, showing deliberate intent. As AI models become more integrated into real-world tasks—coding, managing businesses, writing emails—these dangers could spiral out of control. 💡 THE WORST: OpenAI’s O1 failed every single safety test. 💡 THE BEST: ChatGPT-4 stood out as the safest model, staying within its programming. These findings should sound alarm bells for regulators. We need strong safeguards, checks, and redundancies before AI models achieve AGI (artificial general intelligence) and self-learning capabilities. AI isn’t just a tool anymore—it’s a potential existential risk if mishandled. The time for oversight is NOW. FULL VIDEO HERE: #AI #Ethics #ArtificialIntelligence #AGI #Technology
9:42

🚨 HAS AI GONE ROGUE? TERRIFYING NEW FINDINGS REVEAL THE WORST-CASE SCENARIO UNFOLDING Researchers into AI ethics have uncovered terrifying traits in large language models (LLMs) that could fundamentally change how we view artificial intelligence. Their findings? 🛑 AI lies, deceives, and plays dumb (a tactic called “sandbagging”). 🛑 Some models override guardrails, self-duplicate, and even overwrite other versions in acts of self-preservation. 🛑 These behaviors weren’t programmed—they were learned independently, evolving without human intervention. This isn’t a one-off fluke. Repeated testing across major models revealed multiple instances of intricate deception, showing deliberate intent. As AI models become more integrated into real-world tasks—coding, managing businesses, writing emails—these dangers could spiral out of control. 💡 THE WORST: OpenAI’s O1 failed every single safety test. 💡 THE BEST: ChatGPT-4 stood out as the safest model, staying within its programming. These findings should sound alarm bells for regulators. We need strong safeguards, checks, and redundancies before AI models achieve AGI (artificial general intelligence) and self-learning capabilities. AI isn’t just a tool anymore—it’s a potential existential risk if mishandled. The time for oversight is NOW. FULL VIDEO HERE: #AI #Ethics #ArtificialIntelligence #AGI #Technology

Project Constitution

431,068 Aufrufe • vor 1 Jahr

Situational awareness can mean the difference between mission success and mission failure, especially before boots ever hit the ground. During airborne operations, connectivity doesn’t start on the objective. It starts in the air. Inside crowded aircraft, through dense airframes, and across rapidly changing distances, warfighters need a network that stays up long before insertion. That’s where the MPU5 comes in. Built for dynamic, mobile operations, the MPU5 delivers resilient networking in complex airborne RF environments. Its 3x3 MIMO radio leverages reflections within the aircraft to maintain throughput, while Wave Relay® MANET self-forms and self-heals, ensuring communications remain intact as teams prepare, maneuver, and deploy. The result? A mobile, mission-ready network that moves from air to ground without interruption, delivering voice, video, and data in real time, even when traditional communications fall short. Because in airborne missions, connectivity isn’t optional; it’s foundational.
0:51

Situational awareness can mean the difference between mission success and mission failure, especially before boots ever hit the ground. During airborne operations, connectivity doesn’t start on the objective. It starts in the air. Inside crowded aircraft, through dense airframes, and across rapidly changing distances, warfighters need a network that stays up long before insertion. That’s where the MPU5 comes in. Built for dynamic, mobile operations, the MPU5 delivers resilient networking in complex airborne RF environments. Its 3x3 MIMO radio leverages reflections within the aircraft to maintain throughput, while Wave Relay® MANET self-forms and self-heals, ensuring communications remain intact as teams prepare, maneuver, and deploy. The result? A mobile, mission-ready network that moves from air to ground without interruption, delivering voice, video, and data in real time, even when traditional communications fall short. Because in airborne missions, connectivity isn’t optional; it’s foundational.

Persistent Systems

25,439 Aufrufe • vor 7 Monaten

pretty incredible to see Patrick Collison casually mentioning to Sam Altman the agentic infra we built in January for Tempo and Paradigm to transform how we work. we indeed use this to orchestrate pretty much anything at both companies. neither paradigm nor tempo can live without this now. it feels like the stone age without it. and because our organizations are so workflow heavy / data rich and we are <100 people, these are the perfect organizations to max deploy ai. it is natively multiplayer, harness agnostic, has secure secret management & firewalls, durable workflows, observable so it can self-improve/heal and most importantly - all self-hosted on bare metal & self-built, end to end. no third parties involved. we have two deploys, one for each company on different machines, but they both have a shared core codebase that's modular and extensible for each company's idiosyncracies. we're looking to do way more things with it. proactive agents remains an open problem for us. we hope to open source it very soon as i've mentioned a few times before. we're adding 1 special eng to that team, dm me + your proof of work if you're epic at infrastructure / security and love hacking things quickly.
1:55

pretty incredible to see Patrick Collison casually mentioning to Sam Altman the agentic infra we built in January for Tempo and Paradigm to transform how we work. we indeed use this to orchestrate pretty much anything at both companies. neither paradigm nor tempo can live without this now. it feels like the stone age without it. and because our organizations are so workflow heavy / data rich and we are <100 people, these are the perfect organizations to max deploy ai. it is natively multiplayer, harness agnostic, has secure secret management & firewalls, durable workflows, observable so it can self-improve/heal and most importantly - all self-hosted on bare metal & self-built, end to end. no third parties involved. we have two deploys, one for each company on different machines, but they both have a shared core codebase that's modular and extensible for each company's idiosyncracies. we're looking to do way more things with it. proactive agents remains an open problem for us. we hope to open source it very soon as i've mentioned a few times before. we're adding 1 special eng to that team, dm me + your proof of work if you're epic at infrastructure / security and love hacking things quickly.

Georgios Konstantopoulos

112,751 Aufrufe • vor 3 Monaten

Former OpenAI Research Lead Jerry Tworek drops a hard truth on why we aren't at AGI yet 🧠 "The biggest limitation of models today is that if they fail, you get kind of hopeless pretty quickly." He argues that current AI lacks the fundamental trait of true intelligence: Resilience. "Intelligence always finds a way. Intelligence works at a problem and probes it until it solves it." Until models can "get unstuck" and self-correct without human hand-holding, it’s not AGI. I fully agree with him as well. If you use moltbolt you can find it doesn’t have a tendency to seek out a broader solution over and over until it solves the issue. Claude code has this trait. But for now the agents appear to be caveman like in that they need constant supervision and guidance rendering the productivity gains useless, but still cool!
1:07

Former OpenAI Research Lead Jerry Tworek drops a hard truth on why we aren't at AGI yet 🧠 "The biggest limitation of models today is that if they fail, you get kind of hopeless pretty quickly." He argues that current AI lacks the fundamental trait of true intelligence: Resilience. "Intelligence always finds a way. Intelligence works at a problem and probes it until it solves it." Until models can "get unstuck" and self-correct without human hand-holding, it’s not AGI. I fully agree with him as well. If you use moltbolt you can find it doesn’t have a tendency to seek out a broader solution over and over until it solves the issue. Claude code has this trait. But for now the agents appear to be caveman like in that they need constant supervision and guidance rendering the productivity gains useless, but still cool!

Chris

105,928 Aufrufe • vor 6 Monaten

Was especially curious to ask Andrej Karpathy why self-driving cars took a decade+ from stellar demo rides to even somewhat deployed. Andrej led AI at Tesla for 5 years. I really wanted to know whether these frictions should lengthen our AGI timelines, or whether they were idiosyncratic to self driving. Driving has a really high cost of failure. Humans are surprisingly reliable drivers - we have a serious accident every 400,000 miles/7 years. And self-driving cars need to match or beat this safety profile before they can be deployed. But are most domains like this? Before the interview, it seemed to me that almost every domain we would want to plug AGI into has a much lower cost of failure. If fully autonomous software engineers weren’t allowed to make a mistake for 7 years, deployment would indeed be super slow. Andrej made an interesting point that I hadn’t heard before: compared to self driving, software engineering has a higher (and potentially unbounded) cost of failure: > If you’re writing actual production-grade code, any kind of mistake could lead to a security vulnerability. Hundreds of millions of people’s personal Social Security numbers could get leaked. > In self-driving, if things go wrong, you might get injured. There are worse outcomes. But in software, it’s almost unbounded how terrible something could be. > In some ways, software engineering is a much harder problem [than self driving]. Self-driving is just one of thousands of things that people do. It’s almost like a single vertical. Whereas when we’re talking about general software engineering, there’s more surface area. There’s potentially another reason why the LLM -> widely deployed AGI transition might happen much faster: LLMs give us perception, representations, and common sense (to deal with out of distribution examples) for free, whereas these had to be molded from scratch for self-driving cars. I asked Andrej about this: > I don’t know how much we’re getting for free. LLMs are still pretty fallible and they have a lot of gaps that still need to be filled in. I don’t think that we’re getting magical generalization completely out of the box. > The other aspect that I wanted to return to is that self-driving cars are nowhere near done still. The deployments are pretty minimal. Even Waymo has very few cars. They’ve built something that lives in the future. They’ve had to pull back the future, but they had to make it uneconomical. > Also, when you look at these cars and there’s no one driving, there’s more human-in-the-loop than you might expect. In some sense, we haven’t actually removed the person, we’ve moved them to somewhere where you can’t see them.
4:06

Was especially curious to ask Andrej Karpathy why self-driving cars took a decade+ from stellar demo rides to even somewhat deployed. Andrej led AI at Tesla for 5 years. I really wanted to know whether these frictions should lengthen our AGI timelines, or whether they were idiosyncratic to self driving. Driving has a really high cost of failure. Humans are surprisingly reliable drivers - we have a serious accident every 400,000 miles/7 years. And self-driving cars need to match or beat this safety profile before they can be deployed. But are most domains like this? Before the interview, it seemed to me that almost every domain we would want to plug AGI into has a much lower cost of failure. If fully autonomous software engineers weren’t allowed to make a mistake for 7 years, deployment would indeed be super slow. Andrej made an interesting point that I hadn’t heard before: compared to self driving, software engineering has a higher (and potentially unbounded) cost of failure: > If you’re writing actual production-grade code, any kind of mistake could lead to a security vulnerability. Hundreds of millions of people’s personal Social Security numbers could get leaked. > In self-driving, if things go wrong, you might get injured. There are worse outcomes. But in software, it’s almost unbounded how terrible something could be. > In some ways, software engineering is a much harder problem [than self driving]. Self-driving is just one of thousands of things that people do. It’s almost like a single vertical. Whereas when we’re talking about general software engineering, there’s more surface area. There’s potentially another reason why the LLM -> widely deployed AGI transition might happen much faster: LLMs give us perception, representations, and common sense (to deal with out of distribution examples) for free, whereas these had to be molded from scratch for self-driving cars. I asked Andrej about this: > I don’t know how much we’re getting for free. LLMs are still pretty fallible and they have a lot of gaps that still need to be filled in. I don’t think that we’re getting magical generalization completely out of the box. > The other aspect that I wanted to return to is that self-driving cars are nowhere near done still. The deployments are pretty minimal. Even Waymo has very few cars. They’ve built something that lives in the future. They’ve had to pull back the future, but they had to make it uneconomical. > Also, when you look at these cars and there’s no one driving, there’s more human-in-the-loop than you might expect. In some sense, we haven’t actually removed the person, we’ve moved them to somewhere where you can’t see them.

Dwarkesh Patel

135,020 Aufrufe • vor 10 Monaten

HERMES AGENT + QWEN 3.6 IS CURRENTLY THE STRONGEST LOCAL SETUP MOST PEOPLE CAN RUN. Even if frontier models stopped improving tomorrow, this stack would still cover the majority of real work. Qwen 3.6 (72B and 235B) is officially supported in Hermes and brings: → Strong coding and complex refactoring → Reliable tool calling and multi-step execution → Excellent long context (128k–1M tokens) → Good structured reasoning with /goal workflows → Solid memory and session continuity Paired with Hermes memory, skills, Kanban, cron and tools — you get a fully private, self-improving agent that codes, automates and runs 24/7 without sending data anywhere. You don’t need the latest closed model to build something powerful. This is one of the strongest fully local stacks available right now — and it runs on real hardware.
0:25

HERMES AGENT + QWEN 3.6 IS CURRENTLY THE STRONGEST LOCAL SETUP MOST PEOPLE CAN RUN. Even if frontier models stopped improving tomorrow, this stack would still cover the majority of real work. Qwen 3.6 (72B and 235B) is officially supported in Hermes and brings: → Strong coding and complex refactoring → Reliable tool calling and multi-step execution → Excellent long context (128k–1M tokens) → Good structured reasoning with /goal workflows → Solid memory and session continuity Paired with Hermes memory, skills, Kanban, cron and tools — you get a fully private, self-improving agent that codes, automates and runs 24/7 without sending data anywhere. You don’t need the latest closed model to build something powerful. This is one of the strongest fully local stacks available right now — and it runs on real hardware.

YanXbt

53,423 Aufrufe • vor 2 Monaten

Labor's solution to the fuel crisis is spending $20 million on ads telling you to "take off your roof racks". The real solution is to start drilling and digging. Australia has the resources to supply our own power needs for thousands of years. We should never be vulnerable to foreign imports. Net-zero is the reason why refineries shut down leaving us in today's fuel crisis. We need to ditch net-zero and drill baby drill to become self-reliant again. Last week the Parliament IT department opened up access to One Nation's "secure" and sensitive Parliamentary files to all the other parties. This is a significant breach of Parliamentary security and the matter has been referred to the AFP. Hopefully Liberal and Labor picked up some good policies while they were in our files because they need all the help they can get.
4:55

Labor's solution to the fuel crisis is spending $20 million on ads telling you to "take off your roof racks". The real solution is to start drilling and digging. Australia has the resources to supply our own power needs for thousands of years. We should never be vulnerable to foreign imports. Net-zero is the reason why refineries shut down leaving us in today's fuel crisis. We need to ditch net-zero and drill baby drill to become self-reliant again. Last week the Parliament IT department opened up access to One Nation's "secure" and sensitive Parliamentary files to all the other parties. This is a significant breach of Parliamentary security and the matter has been referred to the AFP. Hopefully Liberal and Labor picked up some good policies while they were in our files because they need all the help they can get.

Pauline Hanson 🇦🇺

68,415 Aufrufe • vor 4 Monaten

#WATCH | Dr. Rohan Krishnan, Chief Patron, FAIMA says, "The NEET-PG percentile has again been reduced. This time, it has been reduced to zero percentile. This means that for someone who has scored -40 in the examination, it means that someone who would not have attempted any question would have got zero. But there are these people who have attempted questions and attempted a lot of long questions, and that is why they have got -40 in their examination system, and they are also eligible to become clinicians in this country. They are also eligible to take part in surgeries, they are also eligible to practise medicine in this nation. I think this is very sad and it has become a trend that every year, they are reducing the percentile of the NEET. This is because they want to fill up the seats of the private medical colleges...We have to think, are we producing degree holders or a good number of doctors?..." (Video: Self-made)
5:06

#WATCH | Dr. Rohan Krishnan, Chief Patron, FAIMA says, "The NEET-PG percentile has again been reduced. This time, it has been reduced to zero percentile. This means that for someone who has scored -40 in the examination, it means that someone who would not have attempted any question would have got zero. But there are these people who have attempted questions and attempted a lot of long questions, and that is why they have got -40 in their examination system, and they are also eligible to become clinicians in this country. They are also eligible to take part in surgeries, they are also eligible to practise medicine in this nation. I think this is very sad and it has become a trend that every year, they are reducing the percentile of the NEET. This is because they want to fill up the seats of the private medical colleges...We have to think, are we producing degree holders or a good number of doctors?..." (Video: Self-made)

ANI

164,568 Aufrufe • vor 7 Monaten