Загрузка видео...

Не удалось загрузить видео

На главную

Bio inspired Hebbian probabilistic network learns in less than 5 minutes from a super sparse single reward per episode! also has imitation learning (manual control) system has 3 parallel competing networks which get sensory input from a 360 vision (27-direction sensory neuron array) link to code in comment each...

hive_echo's profile picture

echo.hive

12,360 subscribers

33,638 просмотров • 10 месяцев назад •via X (Twitter)

ОбразованиеЗдоровье и велнесНаука и технологии
LIVE

Anya Rossi• Live Now

Private livecam show

Комментарии: 0

Нет доступных комментариев

Здесь появятся комментарии из оригинального поста

Похожие видео

Spiking Neural Network from scratch achieves 8% accuracy. no backpropagation or SGD I created a genetic hyper parameter optimizer and it now, on average, can get 8% accuracy which is ~3% above chance Link to source code with a detailed video and markdown explanations in comment it also usually starts lower than 5% and slowly improves then eventually can start dipping below 5%, all of which leads me to believe there is glimmers of learning taking place. sometims it is stabel around 7-8-9% for a long time there is no backpropagation or SGD. it learns via STDP (spike timing dependent plasticity) and a reward mechanism each example is presented n many times (500 in this case) which produces spike train which leads to eligibility list, at the end of a turn based on if the answer was correct or not we adjust the weights using the reward as a multiplier spike timing keeps track of the sequence of neuron firings and which ones were more likely to have lead to the correct answer let me know what you think
0:43

Spiking Neural Network from scratch achieves 8% accuracy. no backpropagation or SGD I created a genetic hyper parameter optimizer and it now, on average, can get 8% accuracy which is ~3% above chance Link to source code with a detailed video and markdown explanations in comment it also usually starts lower than 5% and slowly improves then eventually can start dipping below 5%, all of which leads me to believe there is glimmers of learning taking place. sometims it is stabel around 7-8-9% for a long time there is no backpropagation or SGD. it learns via STDP (spike timing dependent plasticity) and a reward mechanism each example is presented n many times (500 in this case) which produces spike train which leads to eligibility list, at the end of a turn based on if the answer was correct or not we adjust the weights using the reward as a multiplier spike timing keeps track of the sequence of neuron firings and which ones were more likely to have lead to the correct answer let me know what you think

echo.hive

225,484 просмотров • 10 месяцев назад

New episode with Dr. Konrad Kording (Kording Lab 🦖), professor of bioengineering and neuroscience at the University of Pennsylvania (Penn) and co-director of CIFAR's Learning in Machines & Brains program (CIFAR). Konrad works at the intersection of causality, machine learning, and neuroscience, building rigorous methods for causal reasoning when experiments aren't possible — and challenging how researchers interpret neural data and build AI. Konrad argues the most promising path to understanding how the brain works is to read the brain’s wiring directly, down to the molecular detail of each connection, and to build compilers and simulations to understand the brain’s computation directly. In this episode we go deep into how neurons work, how neurons wire together, and how organic and artificial neural networks differ. We discuss why organic neurons are doing much more; how a model of a single organic neuron can solve MNIST — computing more like a 3-layer artificial neural network; how the brain might learn by solving credit assignment with only local signals; how to approximate backprop without a global algorithm; why AI and humans are intelligent along different dimensions; why Konrad isn’t very worried about AI replacing us; economic models of intelligence and physical work; and much more. Konrad is a brilliant, contrarian thinker who explains complex concepts very intuitively. It is a solid computational neuroscience primer. I hope you enjoy this conversation as much as I did! Other links to this episode and references below. Chapters 00:00:00 Introduction 00:01:01 How organic neurons work 00:24:13 How the brain learns: circuits and credit assignment 00:45:29 Recording the brain 00:52:47 Why simulating brains is hard 01:05:00 A new approach: connectomes and compilers 01:21:00 Why simulate brains? 01:29:50 How AI and human intelligence differ 01:41:04 Evolution, intelligence and AI risk 01:52:42 Robotics, causality, and the roots of intelligence 02:05:53 AI for science and scientific rigor 02:13:05 The economics of intelligence 02:27:50 A hopeful future
2:29:40

New episode with Dr. Konrad Kording (Kording Lab 🦖), professor of bioengineering and neuroscience at the University of Pennsylvania (Penn) and co-director of CIFAR's Learning in Machines & Brains program (CIFAR). Konrad works at the intersection of causality, machine learning, and neuroscience, building rigorous methods for causal reasoning when experiments aren't possible — and challenging how researchers interpret neural data and build AI. Konrad argues the most promising path to understanding how the brain works is to read the brain’s wiring directly, down to the molecular detail of each connection, and to build compilers and simulations to understand the brain’s computation directly. In this episode we go deep into how neurons work, how neurons wire together, and how organic and artificial neural networks differ. We discuss why organic neurons are doing much more; how a model of a single organic neuron can solve MNIST — computing more like a 3-layer artificial neural network; how the brain might learn by solving credit assignment with only local signals; how to approximate backprop without a global algorithm; why AI and humans are intelligent along different dimensions; why Konrad isn’t very worried about AI replacing us; economic models of intelligence and physical work; and much more. Konrad is a brilliant, contrarian thinker who explains complex concepts very intuitively. It is a solid computational neuroscience primer. I hope you enjoy this conversation as much as I did! Other links to this episode and references below. Chapters 00:00:00 Introduction 00:01:01 How organic neurons work 00:24:13 How the brain learns: circuits and credit assignment 00:45:29 Recording the brain 00:52:47 Why simulating brains is hard 01:05:00 A new approach: connectomes and compilers 01:21:00 Why simulate brains? 01:29:50 How AI and human intelligence differ 01:41:04 Evolution, intelligence and AI risk 01:52:42 Robotics, causality, and the roots of intelligence 02:05:53 AI for science and scientific rigor 02:13:05 The economics of intelligence 02:27:50 A hopeful future

Juan Benet

50,121 просмотров • 2 месяцев назад

End-to-end neural networks racing drones in Abu Dhabi! 🚁 Check out the drone racing team from Delft University of Technology! A completely end-to-end neural network solution, from pixels to direct motor commands. No Kalman filters. No computer vision feature detectors. Just neurons flying the drone. The challenge is extreme. These drones fly at high speeds and need split-second decisions with minimal onboard resources: a single rolling-shutter camera and an IMU. Their approach is called SkyDreamer, based on the Dreamer-v3 reinforcement learning algorithm. First, a world model is trained in simulation. Then, the neural network learns how to fly in its dreams through reinforcement learning. The network's internal state can be read out to see where it thinks it is on the track or how fast it's going. Even better, the drone estimates some of its own body characteristics during flight, like the camera angle relative to the body, eliminating time-consuming manual calibration. The system uses only a single camera and the gyros from the IMU, ignoring the accelerometers, just like human FPV pilots do. ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →
1:01

End-to-end neural networks racing drones in Abu Dhabi! 🚁 Check out the drone racing team from Delft University of Technology! A completely end-to-end neural network solution, from pixels to direct motor commands. No Kalman filters. No computer vision feature detectors. Just neurons flying the drone. The challenge is extreme. These drones fly at high speeds and need split-second decisions with minimal onboard resources: a single rolling-shutter camera and an IMU. Their approach is called SkyDreamer, based on the Dreamer-v3 reinforcement learning algorithm. First, a world model is trained in simulation. Then, the neural network learns how to fly in its dreams through reinforcement learning. The network's internal state can be read out to see where it thinks it is on the track or how fast it's going. Even better, the drone estimates some of its own body characteristics during flight, like the camera angle relative to the body, eliminating time-consuming manual calibration. The system uses only a single camera and the gyros from the IMU, ignoring the accelerometers, just like human FPV pilots do. ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

Lukas Ziegler

207,500 просмотров • 6 месяцев назад

Dr. David Sinclair says his lab at Harvard grew a miniature human brain in a dish. Then two small black dots appeared on its surface. They were eyes. These brains are grown from single human cells. They develop the same structures found in a full human brain... They even have measurable brain waves... And under a microscope, you can see calcium signals firing like fireworks. This indicates neurons are communicating with each other in the same way neurons in your brain are firing right now as you read this. After about a year, the firing fades and the brain loses function. Sinclair says when his team treats these aging brains with three "longevity" genes called OSK, the firing comes back on. They have also introduced Alzheimer's disease into these brains. Using the same three genes, it reverses. He says this technology is not limited to brains. His lab has reversed aging in the eye, liver, kidney, and skin. Every tissue responds to the same treatment. Now the first human trial is coming. — David Sinclair (David Sinclair) on Tom Bilyeu's (Tom Bilyeu) podcast PS. David Sinclair is speaking at SynBioBeta on May 6th this year, discussing the science of slowing and reversing aging. Tickets available in the comments below.
1:24

Dr. David Sinclair says his lab at Harvard grew a miniature human brain in a dish. Then two small black dots appeared on its surface. They were eyes. These brains are grown from single human cells. They develop the same structures found in a full human brain... They even have measurable brain waves... And under a microscope, you can see calcium signals firing like fireworks. This indicates neurons are communicating with each other in the same way neurons in your brain are firing right now as you read this. After about a year, the firing fades and the brain loses function. Sinclair says when his team treats these aging brains with three "longevity" genes called OSK, the firing comes back on. They have also introduced Alzheimer's disease into these brains. Using the same three genes, it reverses. He says this technology is not limited to brains. His lab has reversed aging in the eye, liver, kidney, and skin. Every tissue responds to the same treatment. Now the first human trial is coming. — David Sinclair (David Sinclair) on Tom Bilyeu's (Tom Bilyeu) podcast PS. David Sinclair is speaking at SynBioBeta on May 6th this year, discussing the science of slowing and reversing aging. Tickets available in the comments below.

John Cumbers

157,553 просмотров • 4 месяцев назад

My last post about being sick with #COVID19 attracted much attention, good and bad. For those who wished me well, thank you very much! I have already recovered my sense of smell and taste and am improving from Covid symptoms every day. For the denialists who took precious time of their lives to come to my post and say that the virus does not exist or that it is just a common cold, just a message: I directly see the virus in the brain using microscopy, and you can see it as well in the video I made below. Viral protein (red, viral nucleocapsid) is found mainly inside neurons (blue, neun) in the olfactory cortex. Glial cells such as astrocytes (white, gfap) and microglia (green, iba1( are attacking these infected neurons, trying to slow down the viral spreading across the brain. Although beneficial, this mechanism also generates more neuroinflammatory propagation, with additional collateral damage to the circuits besides the death of infected neurons. You can deny as much as you want that this is happening, but science is here clearly proving what is going on. Our first study was published last year: We are currently also looking at how vaccines can modulate this type of neuronal damage. So far, it seems that, although not able to prevent neuronal infection, vaccination reduces the neuroinflammatory response generated by the immune system in the brain. More to come soon! #FluorescenceFriday
0:40

My last post about being sick with #COVID19 attracted much attention, good and bad. For those who wished me well, thank you very much! I have already recovered my sense of smell and taste and am improving from Covid symptoms every day. For the denialists who took precious time of their lives to come to my post and say that the virus does not exist or that it is just a common cold, just a message: I directly see the virus in the brain using microscopy, and you can see it as well in the video I made below. Viral protein (red, viral nucleocapsid) is found mainly inside neurons (blue, neun) in the olfactory cortex. Glial cells such as astrocytes (white, gfap) and microglia (green, iba1( are attacking these infected neurons, trying to slow down the viral spreading across the brain. Although beneficial, this mechanism also generates more neuroinflammatory propagation, with additional collateral damage to the circuits besides the death of infected neurons. You can deny as much as you want that this is happening, but science is here clearly proving what is going on. Our first study was published last year: We are currently also looking at how vaccines can modulate this type of neuronal damage. So far, it seems that, although not able to prevent neuronal infection, vaccination reduces the neuroinflammatory response generated by the immune system in the brain. More to come soon! #FluorescenceFriday

Danielle Beckman

387,794 просмотров • 2 лет назад

Tulum Peyniri is a Turkish mountain cheese that is matured in a goat skin. Traditionally, local farmers make this cheese using raw cow or sheep milk. In more modern times, you might find versions that are made with pasteurised cow, goat or sheep milk... Tulum Peyniri is a very unusual type of cheese that is gaining popularity in the culinary world. It is a traditional cheese that has been produced for centuries in Turkey. This unique cheese has a distinctive flavour and texture that has been enjoyed by locals and tourists alike. Tulum Peyniri originated in Turkey, specifically in the province of Erzincan. It is believed that the cheese has been produced in this region for over a thousand years. Locals use either cow, goat or sheep milk to make this cheese and age it in goat skin bags called tulum. Actually, this Turkish cheese has a rich history that dates back to the Ottoman Empire. It was a popular cheese among the Ottoman aristocracy and was often served at royal banquets. Over time, the cheese became more widely available and was enjoyed by people from all walks of life. Presently, Tulum Peyniri is primarily produced in the province of Erzincan, which is located in eastern Turkey. The region is known for its rugged terrain and harsh climate, which makes it ideal for sheep farming. The sheep in this region graze on wild herbs and grasses, which gives the milk a unique flavour. Today, there are many small-scale producers in Turkey. Some of the most well-known producers include Hacı Dayı, Çağlayanlar and Özcanlar. Traditionally, the locals used raw cow or sheep milk. But in more modern times, you might find versions that are made with pasteurised cow, goat or sheep milk. And that are aged in metal tins instead of animal hide. After fermenting the fresh milk, the cheesemaker strains it through a cheese cloth to separate the whey from the curds. Afterwards, they rest the young cheese for one night before transferring it to a copper vessel. At this stage, they crumble the cheese and add dry salt. It is then pressed and allowed to drain for a further 10 days. And this is when the magic begins. The formed young cheese is moved into a goat skin that has been specially cleaned and treated. It is pressed into the skin, filling it compactly. And the opening of the skin is then sewn shut. Maturation takes place at high altitudes and humidity. And can last for up to three months. At this age, Tulum develops a beautiful crumbly texture and a delicious complex flavour. Tulum Peyniri has a sharp, tangy flavour that is similar to Feta. However, it has a crumbly texture and is quite dry compared to other types of cheese. Moreover, this cheese has a strong aroma that is earthy and slightly sour. Undoubtedly, this is a very versatile cheese that pairs well with many different foods. Locals often serve it as part of a meze platter alongside olives, tomatoes and other Mediterranean-style dishes. Tulum excels as a table cheese but can also be used in cooking. Some great recipes showcasing this cheese include börek, pides and pesto sauce. Furthermore, it also pairs well with roasted vegetables, grilled meats and salads. 🎥 : Credit to the Owner #archaeohistories
0:31

Tulum Peyniri is a Turkish mountain cheese that is matured in a goat skin. Traditionally, local farmers make this cheese using raw cow or sheep milk. In more modern times, you might find versions that are made with pasteurised cow, goat or sheep milk... Tulum Peyniri is a very unusual type of cheese that is gaining popularity in the culinary world. It is a traditional cheese that has been produced for centuries in Turkey. This unique cheese has a distinctive flavour and texture that has been enjoyed by locals and tourists alike. Tulum Peyniri originated in Turkey, specifically in the province of Erzincan. It is believed that the cheese has been produced in this region for over a thousand years. Locals use either cow, goat or sheep milk to make this cheese and age it in goat skin bags called tulum. Actually, this Turkish cheese has a rich history that dates back to the Ottoman Empire. It was a popular cheese among the Ottoman aristocracy and was often served at royal banquets. Over time, the cheese became more widely available and was enjoyed by people from all walks of life. Presently, Tulum Peyniri is primarily produced in the province of Erzincan, which is located in eastern Turkey. The region is known for its rugged terrain and harsh climate, which makes it ideal for sheep farming. The sheep in this region graze on wild herbs and grasses, which gives the milk a unique flavour. Today, there are many small-scale producers in Turkey. Some of the most well-known producers include Hacı Dayı, Çağlayanlar and Özcanlar. Traditionally, the locals used raw cow or sheep milk. But in more modern times, you might find versions that are made with pasteurised cow, goat or sheep milk. And that are aged in metal tins instead of animal hide. After fermenting the fresh milk, the cheesemaker strains it through a cheese cloth to separate the whey from the curds. Afterwards, they rest the young cheese for one night before transferring it to a copper vessel. At this stage, they crumble the cheese and add dry salt. It is then pressed and allowed to drain for a further 10 days. And this is when the magic begins. The formed young cheese is moved into a goat skin that has been specially cleaned and treated. It is pressed into the skin, filling it compactly. And the opening of the skin is then sewn shut. Maturation takes place at high altitudes and humidity. And can last for up to three months. At this age, Tulum develops a beautiful crumbly texture and a delicious complex flavour. Tulum Peyniri has a sharp, tangy flavour that is similar to Feta. However, it has a crumbly texture and is quite dry compared to other types of cheese. Moreover, this cheese has a strong aroma that is earthy and slightly sour. Undoubtedly, this is a very versatile cheese that pairs well with many different foods. Locals often serve it as part of a meze platter alongside olives, tomatoes and other Mediterranean-style dishes. Tulum excels as a table cheese but can also be used in cooking. Some great recipes showcasing this cheese include börek, pides and pesto sauce. Furthermore, it also pairs well with roasted vegetables, grilled meats and salads. 🎥 : Credit to the Owner #archaeohistories

Archaeo - Histories

99,207 просмотров • 2 лет назад

Most AI tools today are just command-takers. You type, it responds. You click, it executes. There's truly no learning, no memory, no adaptation. You're still doing most of the work. Which is why Miles by Avo on Solana would make so much sense if built well. From what I understand so far, miles learns about you over time, adapts to how you operate, and figures out the tools you struggle to navigate on your own. Less clicking, more conversation, the more you use it, the better it gets at working with you, not just for you. That shift from command-follower to learning companion is bigger than it sounds. It's the difference between a tool and an agent that actually knows you. The second one is Agent Grid and this one is for builders and power users. But as I mentioned in this video, I'm still studying the Agent Grid. I'd come talk about it once I learn properly about it. Souren just keeps shipping!
1:28

Most AI tools today are just command-takers. You type, it responds. You click, it executes. There's truly no learning, no memory, no adaptation. You're still doing most of the work. Which is why Miles by Avo on Solana would make so much sense if built well. From what I understand so far, miles learns about you over time, adapts to how you operate, and figures out the tools you struggle to navigate on your own. Less clicking, more conversation, the more you use it, the better it gets at working with you, not just for you. That shift from command-follower to learning companion is bigger than it sounds. It's the difference between a tool and an agent that actually knows you. The second one is Agent Grid and this one is for builders and power users. But as I mentioned in this video, I'm still studying the Agent Grid. I'd come talk about it once I learn properly about it. Souren just keeps shipping!

Sir Khaycee

20,465 просмотров • 5 месяцев назад

Listen to this sound. It's called the earth's heartbeat. Winfried Otto Schumann predicted this in 1952 with nothing but mathematics, and he was almost embarrassed to publish it. He calculated that the gap between the Earth's surface and the ionosphere, the electrically charged layer of the upper atmosphere, forms a closed cavity. A resonant chamber. And like any chamber, from a cathedral to the hollow body of a guitar, it has a natural frequency at which it wants to vibrate. His number was roughly 7.83 Hz. Then comes the part almost nobody mentions. What actually excites this cavity, what strikes the bell and keeps it ringing, is lightning. At any given moment around 2000 thunderstorms are firing across the planet, sending out close to 50 lightning strikes every second. Each strike releases a burst of electromagnetic energy that races around the globe inside that cavity. The bursts sized to fit the chamber reinforce each other, and the whole planet hums. You are standing inside a resonant cavity powered by lightning. Right now. It has never once switched off in the entire history of your species. That part is not fringe. It is textbook geophysics, confirmed experimentally in the early 1960s, used today to track global lightning activity and monitor changes in the upper atmosphere. The story splits at this point, and I would rather be straight with you than sell you something. 7.83 Hz sits almost exactly on the border between alpha and theta brain waves. Alpha shows up when you close your eyes and relax. Theta shows up in deep meditation, light sleep, that hypnagogic drift in the seconds before you lose consciousness. So the coincidence is real. The number the planet hums at lands right inside the range your brain produces when it goes quiet. That coincidence became the foundation of an entire industry. Devices that promise to pulse 7.83 Hz into your bedroom. Apps that claim to sync your brain to the Earth. The story that modern life, wrapped in artificial electromagnetic noise, cut us off from the planet's rhythm and made us sick. Most of it runs miles ahead of anything anyone has actually shown. A numerical match between two frequencies does not mean one drives the other. Your brain has no antenna tuned to 7.83 Hz. By the time the Schumann resonance reaches you it is astonishingly faint, far weaker than the fields humming off the wiring in your walls. If your neurons were genuinely locking onto ambient fields at that strength, your house would have hijacked your consciousness long before the planet ever got the chance. The honest version is simple. The Earth's pulse is real. The frequency overlap with resting brain states is real. A proven causal bridge between them is not. And somehow that makes the true story more interesting, not less. Because the deeper question the hype walks straight past is why your brain settles into rhythms at all. Why does a calm nervous system drift toward these slow, ordered oscillations? Why do billions of neurons, with no conductor and no sheet music, spontaneously fall into step the way fireflies flash in unison, the way pendulum clocks mounted on the same wall drift into sync over a few hours? Synchronization is one of the deepest patterns in nature. It runs through heart cells, power grids, applauding crowds, and the neurons firing behind your eyes as you read this line. The planet resonates because lightning drives a cavity into sync. Your brain resonates because millions of cells drive each other into sync. The mechanisms have nothing to do with one another. The underlying phenomenon, order emerging for free out of countless tiny oscillators finding a shared beat, might be one of the most universal laws we have. That is the part worth being floored by. Not that the Earth is secretly tuning your mind. That the same mathematical principle, resonance and synchronization, writes itself into thunderstorms and heartbeats and neurons and clocks in the same handwriting. The mystics felt something real and then reached for the wrong mechanism. There is a rhythm that runs through the living and the nonliving alike. It just is not a radio station in the sky broadcasting into your skull. It is something stranger. A tendency, stitched into the structure of reality itself, for separate things to fall into step. The Earth found its beat from lightning. You find yours from ten billion neurons quietly agreeing on when to fire.
2:06

Listen to this sound. It's called the earth's heartbeat. Winfried Otto Schumann predicted this in 1952 with nothing but mathematics, and he was almost embarrassed to publish it. He calculated that the gap between the Earth's surface and the ionosphere, the electrically charged layer of the upper atmosphere, forms a closed cavity. A resonant chamber. And like any chamber, from a cathedral to the hollow body of a guitar, it has a natural frequency at which it wants to vibrate. His number was roughly 7.83 Hz. Then comes the part almost nobody mentions. What actually excites this cavity, what strikes the bell and keeps it ringing, is lightning. At any given moment around 2000 thunderstorms are firing across the planet, sending out close to 50 lightning strikes every second. Each strike releases a burst of electromagnetic energy that races around the globe inside that cavity. The bursts sized to fit the chamber reinforce each other, and the whole planet hums. You are standing inside a resonant cavity powered by lightning. Right now. It has never once switched off in the entire history of your species. That part is not fringe. It is textbook geophysics, confirmed experimentally in the early 1960s, used today to track global lightning activity and monitor changes in the upper atmosphere. The story splits at this point, and I would rather be straight with you than sell you something. 7.83 Hz sits almost exactly on the border between alpha and theta brain waves. Alpha shows up when you close your eyes and relax. Theta shows up in deep meditation, light sleep, that hypnagogic drift in the seconds before you lose consciousness. So the coincidence is real. The number the planet hums at lands right inside the range your brain produces when it goes quiet. That coincidence became the foundation of an entire industry. Devices that promise to pulse 7.83 Hz into your bedroom. Apps that claim to sync your brain to the Earth. The story that modern life, wrapped in artificial electromagnetic noise, cut us off from the planet's rhythm and made us sick. Most of it runs miles ahead of anything anyone has actually shown. A numerical match between two frequencies does not mean one drives the other. Your brain has no antenna tuned to 7.83 Hz. By the time the Schumann resonance reaches you it is astonishingly faint, far weaker than the fields humming off the wiring in your walls. If your neurons were genuinely locking onto ambient fields at that strength, your house would have hijacked your consciousness long before the planet ever got the chance. The honest version is simple. The Earth's pulse is real. The frequency overlap with resting brain states is real. A proven causal bridge between them is not. And somehow that makes the true story more interesting, not less. Because the deeper question the hype walks straight past is why your brain settles into rhythms at all. Why does a calm nervous system drift toward these slow, ordered oscillations? Why do billions of neurons, with no conductor and no sheet music, spontaneously fall into step the way fireflies flash in unison, the way pendulum clocks mounted on the same wall drift into sync over a few hours? Synchronization is one of the deepest patterns in nature. It runs through heart cells, power grids, applauding crowds, and the neurons firing behind your eyes as you read this line. The planet resonates because lightning drives a cavity into sync. Your brain resonates because millions of cells drive each other into sync. The mechanisms have nothing to do with one another. The underlying phenomenon, order emerging for free out of countless tiny oscillators finding a shared beat, might be one of the most universal laws we have. That is the part worth being floored by. Not that the Earth is secretly tuning your mind. That the same mathematical principle, resonance and synchronization, writes itself into thunderstorms and heartbeats and neurons and clocks in the same handwriting. The mystics felt something real and then reached for the wrong mechanism. There is a rhythm that runs through the living and the nonliving alike. It just is not a radio station in the sky broadcasting into your skull. It is something stranger. A tendency, stitched into the structure of reality itself, for separate things to fall into step. The Earth found its beat from lightning. You find yours from ten billion neurons quietly agreeing on when to fire.

The Curious Tales

268,360 просмотров • 1 месяц назад

.Naval: You have a beautiful definition of knowledge, which most people don’t even try to tackle, about how knowledge perpetuates itself in the environment. You gave some really good examples. One was around genes. Successful, highly adapted genes contain a lot of knowledge and can cause themselves to be replicated because they’re survivors. In the same way, knowledge itself is a survivor, in that if you transmit to me the knowledge of how to build a computer, it’s an incredibly useful thing. I’m going to build more and more computers and that knowledge will be passed on. Your underlying point that you repeated here was if you want to understand the physical universe you have to understand knowledge, because it is the thing that over time takes over and changes more and more the universe—more than almost anything else. You have to understand all the explanations behind it. You can’t just say “particle collisions” because that explains everything, so it explains nothing. It’s not a useful level to operate at. Therefore, the things that create knowledge are uniquely influential in the universe. And as far as we know, there are only two systems that create knowledge. There’s evolution and there are humans. But is there a difference even between these two forms of knowledge creation, between evolution and between humans? David Deutsch: Yes. I have argued that the human way of creating knowledge is the ultimate one, that there aren’t any more powerful ones than that. This is the argument against the supernatural. Assuming that there is a form of knowledge creation that’s more powerful than ours is equivalent to invoking the supernatural, which is therefore a bad explanation—as invoking the supernatural always is. The difference between biological evolution and human creative thought is that biological evolution is inherently limited in its range. That’s because biological evolution has no foresight. It can’t see a problem and conjecture a solution. Whenever biological evolution produces a solution to something, it’s always before natural selection has even begun. This is Charles Darwin’s insight. This is the difference between Charles Darwin’s theory of evolution and the other theories of evolution that had been around for a century or more before that, including Charles Darwin’s grandfather and Lamarck. The thing they didn’t get is that the creation of knowledge in evolution begins before. That means that biological evolution can’t reach places that are not reachable by successive improvements, each of which allows a viable organism to exist. Creationists say that biological evolution has, in fact, reached things that are not reachable by incremental steps, each of which is a viable organism. They’re factually mistaken. The thing which they have in mind is the idea of a creator who can imagine things that don’t exist and who can create an idea that is not the culmination of a whole load of viable things. A thinking being can create something that’s a culmination of a whole load of non-viable things. Explanatory creativity makes humans unique Out of all the billions and billions of species that have ever existed, none of them has ever made a campfire, even though many of them would’ve been helped by having the genetic capacity to make campfires. The reason it didn’t happen in the biosphere is that there is no such thing as making a partially functional campfire; whereas there is, for example, with making hot water. The bombardier beetles squirt boiling water at their enemies. You can easily see that just squirting cold water at your enemies is not totally unhelpful. Then making it a bit hotter and a bit hotter. Squirting boiling water no doubt required many adaptations to make sure the beetle didn’t boil itself while it was making this boiling water. That happened because there was a sequence of steps in between, all of which were useful. But with campfires, it’s very hard to see how that could happen. Humans have explanatory creativity. Once you have that, you can get to the moon. You can cause asteroids which are heading towards the earth to turn around and go away. Perhaps no other planet in the universe has that power, and it has it only because of the presence of explanatory creativity on it.
4:31

.Naval: You have a beautiful definition of knowledge, which most people don’t even try to tackle, about how knowledge perpetuates itself in the environment. You gave some really good examples. One was around genes. Successful, highly adapted genes contain a lot of knowledge and can cause themselves to be replicated because they’re survivors. In the same way, knowledge itself is a survivor, in that if you transmit to me the knowledge of how to build a computer, it’s an incredibly useful thing. I’m going to build more and more computers and that knowledge will be passed on. Your underlying point that you repeated here was if you want to understand the physical universe you have to understand knowledge, because it is the thing that over time takes over and changes more and more the universe—more than almost anything else. You have to understand all the explanations behind it. You can’t just say “particle collisions” because that explains everything, so it explains nothing. It’s not a useful level to operate at. Therefore, the things that create knowledge are uniquely influential in the universe. And as far as we know, there are only two systems that create knowledge. There’s evolution and there are humans. But is there a difference even between these two forms of knowledge creation, between evolution and between humans? David Deutsch: Yes. I have argued that the human way of creating knowledge is the ultimate one, that there aren’t any more powerful ones than that. This is the argument against the supernatural. Assuming that there is a form of knowledge creation that’s more powerful than ours is equivalent to invoking the supernatural, which is therefore a bad explanation—as invoking the supernatural always is. The difference between biological evolution and human creative thought is that biological evolution is inherently limited in its range. That’s because biological evolution has no foresight. It can’t see a problem and conjecture a solution. Whenever biological evolution produces a solution to something, it’s always before natural selection has even begun. This is Charles Darwin’s insight. This is the difference between Charles Darwin’s theory of evolution and the other theories of evolution that had been around for a century or more before that, including Charles Darwin’s grandfather and Lamarck. The thing they didn’t get is that the creation of knowledge in evolution begins before. That means that biological evolution can’t reach places that are not reachable by successive improvements, each of which allows a viable organism to exist. Creationists say that biological evolution has, in fact, reached things that are not reachable by incremental steps, each of which is a viable organism. They’re factually mistaken. The thing which they have in mind is the idea of a creator who can imagine things that don’t exist and who can create an idea that is not the culmination of a whole load of viable things. A thinking being can create something that’s a culmination of a whole load of non-viable things. Explanatory creativity makes humans unique Out of all the billions and billions of species that have ever existed, none of them has ever made a campfire, even though many of them would’ve been helped by having the genetic capacity to make campfires. The reason it didn’t happen in the biosphere is that there is no such thing as making a partially functional campfire; whereas there is, for example, with making hot water. The bombardier beetles squirt boiling water at their enemies. You can easily see that just squirting cold water at your enemies is not totally unhelpful. Then making it a bit hotter and a bit hotter. Squirting boiling water no doubt required many adaptations to make sure the beetle didn’t boil itself while it was making this boiling water. That happened because there was a sequence of steps in between, all of which were useful. But with campfires, it’s very hard to see how that could happen. Humans have explanatory creativity. Once you have that, you can get to the moon. You can cause asteroids which are heading towards the earth to turn around and go away. Perhaps no other planet in the universe has that power, and it has it only because of the presence of explanatory creativity on it.

Deutsch Explains

186,329 просмотров • 1 год назад

THIS GUY SPENT 5 MINUTES SETTING UP A $20 CLAUDE WORKSPACE AND TURNED 30 MINUTES OF DAILY PROMPT CHAOS INTO A SYSTEM Most people open Claude, type 1 messy prompt, rewrite it 10 times, get the same generic answer and call the model “bad.” He opens Claude Cowork, adds the project rules, past context, working files, examples, decisions and a clean place for the model to remember what it is doing. Same Claude. Same $20 tool. Completely different output. The funny part is there is no magic prompt here. No secret template. No “10x Claude” trick. Just the boring layer Karpathy keeps pointing at: context, memory and structure before output. A setup like this can save 30 minutes a day, which is 15 hours a month. If you use Claude for client work at $50 to $100/hour, that is $750 to $1,500/month in time you stop burning on repeated prompts. The model is real. The output is real. The edge is in not making Claude start from zero every single time.
0:37

THIS GUY SPENT 5 MINUTES SETTING UP A $20 CLAUDE WORKSPACE AND TURNED 30 MINUTES OF DAILY PROMPT CHAOS INTO A SYSTEM Most people open Claude, type 1 messy prompt, rewrite it 10 times, get the same generic answer and call the model “bad.” He opens Claude Cowork, adds the project rules, past context, working files, examples, decisions and a clean place for the model to remember what it is doing. Same Claude. Same $20 tool. Completely different output. The funny part is there is no magic prompt here. No secret template. No “10x Claude” trick. Just the boring layer Karpathy keeps pointing at: context, memory and structure before output. A setup like this can save 30 minutes a day, which is 15 hours a month. If you use Claude for client work at $50 to $100/hour, that is $750 to $1,500/month in time you stop burning on repeated prompts. The model is real. The output is real. The edge is in not making Claude start from zero every single time.

Gipp 🦅

14,871 просмотров • 3 месяцев назад

Milestone! We (robotic arms for gadgets assembly) finished the first commercial order, which brought the first revenue. Here are some learnings from this: The customer was a smart toy manufacturer. The task was to add a heatsink to Raspberry Pi. We received parts from them and returned the assembled modules back. Currently, it's done by teleoperation. Later it will be done by a remote employee via the Internet. Then it will be automated action by action, reducing the operator's time on this and making the task profitable. ps. If you have an assembly task that we can do for you asynchronically - leave a comment below. Learning 1. It's possible! This task which is usually done by the human arm with 5 fingers can be done with a two-finger gripper with the addition of a couple of simple tooling. The task was not simplified. We peeled off thin films from stickers, unpacked paper boxes, moved PCB boards full of components, etc. And no unsolvable problems have been encountered yet. Challenges: 1) The paper box shifted during the opening Solved with the plastic walls that you can lean against 2) Heat pad, stuck to the gripper instead of heat sync. Can be solved by gripper with a pump, but this time solved with the patience of the operator 3) The film on the pad is very thin. Turned out that sub-millimeter arm precision is enough to peel it off with just a regular gripper. 4) The working area has not enough space. You'll only know this by doing real tasks in bulk. This could be solved by an extra pair of long arms, but in this case, solved with the patience of the operator. I think that in the end, we will have 5-10 types of universal tooling and 5-10 types of grippers to solve almost all the problems in such assembly tasks. Learning 2. It's slow. It took 5 times more time, than doing it with human hands. But the good news is there's a lot of room for improvement. We now have specific “time for task” metrics, which we will decrease with iterations. The main reasons for slowness: 1) To rotate the gripper to a steep angle you are forced to control one robot arm with two hands instead of using both arms. We can fix this by just making more room for rotations. 2) Grabbing PCB board with two arms is hard. A slight difference in rotation can break the board, and it's hard to control these angles visually. To solve this, the best way is to use force feedback so you can feel the pressure applied to the item. 3) Accuracy and steadiness is still can be improved We will try a metal version and double the motors to do this. 4) It is physically difficult for the human hands to move with such precision To solve this, we will add a pad for the hands like in surgical robots Learning 3. It's a good business model The "Factory in the cloud" is a good business model for this stage. You send us parts and we send back assembled modules. Currently, it's more convenient than sending a robot to your place, as we can iterate/fix the robot quickly and utilize it 100% of the time. When we polish the set-up over time - we can send robots to your place. So if we can assemble something for you in the USA with Chinese prices by using modern automation - leave a comment below.
0:48

Milestone! We (robotic arms for gadgets assembly) finished the first commercial order, which brought the first revenue. Here are some learnings from this: The customer was a smart toy manufacturer. The task was to add a heatsink to Raspberry Pi. We received parts from them and returned the assembled modules back. Currently, it's done by teleoperation. Later it will be done by a remote employee via the Internet. Then it will be automated action by action, reducing the operator's time on this and making the task profitable. ps. If you have an assembly task that we can do for you asynchronically - leave a comment below. Learning 1. It's possible! This task which is usually done by the human arm with 5 fingers can be done with a two-finger gripper with the addition of a couple of simple tooling. The task was not simplified. We peeled off thin films from stickers, unpacked paper boxes, moved PCB boards full of components, etc. And no unsolvable problems have been encountered yet. Challenges: 1) The paper box shifted during the opening Solved with the plastic walls that you can lean against 2) Heat pad, stuck to the gripper instead of heat sync. Can be solved by gripper with a pump, but this time solved with the patience of the operator 3) The film on the pad is very thin. Turned out that sub-millimeter arm precision is enough to peel it off with just a regular gripper. 4) The working area has not enough space. You'll only know this by doing real tasks in bulk. This could be solved by an extra pair of long arms, but in this case, solved with the patience of the operator. I think that in the end, we will have 5-10 types of universal tooling and 5-10 types of grippers to solve almost all the problems in such assembly tasks. Learning 2. It's slow. It took 5 times more time, than doing it with human hands. But the good news is there's a lot of room for improvement. We now have specific “time for task” metrics, which we will decrease with iterations. The main reasons for slowness: 1) To rotate the gripper to a steep angle you are forced to control one robot arm with two hands instead of using both arms. We can fix this by just making more room for rotations. 2) Grabbing PCB board with two arms is hard. A slight difference in rotation can break the board, and it's hard to control these angles visually. To solve this, the best way is to use force feedback so you can feel the pressure applied to the item. 3) Accuracy and steadiness is still can be improved We will try a metal version and double the motors to do this. 4) It is physically difficult for the human hands to move with such precision To solve this, we will add a pad for the hands like in surgical robots Learning 3. It's a good business model The "Factory in the cloud" is a good business model for this stage. You send us parts and we send back assembled modules. Currently, it's more convenient than sending a robot to your place, as we can iterate/fix the robot quickly and utilize it 100% of the time. When we polish the set-up over time - we can send robots to your place. So if we can assemble something for you in the USA with Chinese prices by using modern automation - leave a comment below.

Igor Kulakov

37,266 просмотров • 1 год назад

PALANTIR'S MILITARY A.I. & NEUROLOGICAL WEAPONS adaptation. This is the most important aspect of the A.I. conversation and adaption of Palantir and the weapons associated with it that nobody is talking about. Beside the secret contracts with Palantir and the CIA, the use of direct energy weapons, A.I., facial recognition, and civilian databases have been in use for years. Even Rite-Aid was using advanced A.I. facial recognition technology in their stores over 5 years ago. All of this already exists. They already know everything about you and so does big tech. This is what everyone is talking about but it doesn't matter at all. That ship has sailed 20+ years ago. You need to get past this aspect and understand that we are way farther ahead, technologically speaking, then you could have ever imagined. I will briefly give you an introduction and example. I suggest you start researching these things and technologies on your own. What they're not talking about is the adaption of other technology and neurological weapons that are both lethal and non-lethal that can be integrated with these global surveillance systems and networks that takes this to a whole other level and completely turns this into a wireless biological monitorization and neurological/biological control weapon of a select target/s anywhere in the world. For example, imagine using earbuds that read your brainwaves in real time, which is your thoughts, then by using this global monitoring system with the most advanced A.I., these A.I. scripts are then running in real time and decoding those brainwaves, building a profile on your specific brainwaves and learning and teaching itself at the same time. Now what you're thinking and everywhere you go they will know specific things you are thinking. It get's worse. They already have electrodes made from nanotechnology that can enter your body in multiple ways through the air, your food, or vaccines, which can then enter the brain through the blood-brain barrier, and plant themselves in your head which then give the capability not only to send information but to also receive information or pulses to stimulate the brain and other biological components. Ladies and gentleman the type of tech that is available can also control and induce emotions and behaviors or you nervous system, which in turn, controls you. DARPA has already successfully created biological drones with full control of their nervous system. This is where BMI (Brain Machine Interface) comes into play and the capabilities are endless. We haven't even entered the aspect of using light, electromagnetic radiation, radio or extremely low frequency or pulses, or acoustic waves either which are also able to send and receive signals to and from the brain to monitor or write/command to the brain in real time. This also goes for other parts of the body, like being able to induce heart problems, etc. You need to start thinking bigger because 99% of the population is stuck and worried about the government having a database on them, which is beyond laughable because they have multiple databases on everyone. Your phone alone gives them everything they need to know about you already with a camera and microphone. That is not the issue. It's physical and literal control and invasion of your thoughts, behaviors, emotions, literal functions, completely remotely, anywhere in the world. This isn't something to be worried about down the road. It's already here. You need to learn what's really going on and the real agenda and capabilities behind this technology. There's a reason the mainstream barley talks about this. You need to understand how it works and the difference incase you are ever targeted. I can confirm that reality is far stranger and more fascinating than any fiction. You've been warned.
11:18

PALANTIR'S MILITARY A.I. & NEUROLOGICAL WEAPONS adaptation. This is the most important aspect of the A.I. conversation and adaption of Palantir and the weapons associated with it that nobody is talking about. Beside the secret contracts with Palantir and the CIA, the use of direct energy weapons, A.I., facial recognition, and civilian databases have been in use for years. Even Rite-Aid was using advanced A.I. facial recognition technology in their stores over 5 years ago. All of this already exists. They already know everything about you and so does big tech. This is what everyone is talking about but it doesn't matter at all. That ship has sailed 20+ years ago. You need to get past this aspect and understand that we are way farther ahead, technologically speaking, then you could have ever imagined. I will briefly give you an introduction and example. I suggest you start researching these things and technologies on your own. What they're not talking about is the adaption of other technology and neurological weapons that are both lethal and non-lethal that can be integrated with these global surveillance systems and networks that takes this to a whole other level and completely turns this into a wireless biological monitorization and neurological/biological control weapon of a select target/s anywhere in the world. For example, imagine using earbuds that read your brainwaves in real time, which is your thoughts, then by using this global monitoring system with the most advanced A.I., these A.I. scripts are then running in real time and decoding those brainwaves, building a profile on your specific brainwaves and learning and teaching itself at the same time. Now what you're thinking and everywhere you go they will know specific things you are thinking. It get's worse. They already have electrodes made from nanotechnology that can enter your body in multiple ways through the air, your food, or vaccines, which can then enter the brain through the blood-brain barrier, and plant themselves in your head which then give the capability not only to send information but to also receive information or pulses to stimulate the brain and other biological components. Ladies and gentleman the type of tech that is available can also control and induce emotions and behaviors or you nervous system, which in turn, controls you. DARPA has already successfully created biological drones with full control of their nervous system. This is where BMI (Brain Machine Interface) comes into play and the capabilities are endless. We haven't even entered the aspect of using light, electromagnetic radiation, radio or extremely low frequency or pulses, or acoustic waves either which are also able to send and receive signals to and from the brain to monitor or write/command to the brain in real time. This also goes for other parts of the body, like being able to induce heart problems, etc. You need to start thinking bigger because 99% of the population is stuck and worried about the government having a database on them, which is beyond laughable because they have multiple databases on everyone. Your phone alone gives them everything they need to know about you already with a camera and microphone. That is not the issue. It's physical and literal control and invasion of your thoughts, behaviors, emotions, literal functions, completely remotely, anywhere in the world. This isn't something to be worried about down the road. It's already here. You need to learn what's really going on and the real agenda and capabilities behind this technology. There's a reason the mainstream barley talks about this. You need to understand how it works and the difference incase you are ever targeted. I can confirm that reality is far stranger and more fascinating than any fiction. You've been warned.

The SCIF

56,970 просмотров • 1 год назад

Excited to release a new repo: abcGPT! It can be hard to "dial in" the voice you want from an LLM, because an LLM is a tangled superposition of millions of voices from millions of different authors around the world. Instead, frontier LLMs tend to give that slop-ish / generic / corporate tone that's hard to avoid, even with aggressive prompting and an informative context window. Lately I've been experimenting with some ideas on the fringes of attribution/unlearning, trying to make it so an AI user can "dial in" the specific voice/style/sources they want to use in a way that's more rigorous than prompting/context-engineering. and I'm starting to get pretty good results. the model below uses the following technique: - Take nanoGPT as written by Andrej Karpathy - Assign each neuron a random "specialty score" m between 0 and 1, sampled from a U-shape so most neurons land near 0 or near 1 with some in the middle. - Freeze this "m" for the lifetime of the network (it's the neuron's permanent corpus assignment) - Extend the forward() code with an α parameter, a kind of vibe-fader from 0 to 1. Think of each neuron's m as its position on that same slider. The slider acts like a spotlight: it lights up neurons whose m is near its current position, and silences those far away. Slide all the way to 0, and only TinyStories specialists fire. Slide all the way to 1, and only Shakespeare specialists fire. - Train this new nanoGPT on two datasets (in this case, TinyStories and Shakespeare) - During training, sample α from Beta(0.5, 0.5) AND draw the corpus from Bernoulli(α), so a Shakespeare batch tends to come with a high-α (Shakespeare-favoring) gate, and a TinyStories batch tends to come with low-α. - train until golden brown 🧑‍🍳 Perhaps surprisingly... it works! ¯\_(ツ)_/¯ The neurons we pre-assigned to Shakespeare learn to behave as Shakespeare specialists. the neurons we pre-assigned to TinyStories become children's-story specialists. the halfsies learn to bridge between them. After training, you can play with the kindof... vibe dial... you can "dial in" the voice you want during inference, by choosing whether to lean on Shakespeare or TinyStories neurons more or less. 📀💿 When you fully dial in Shakespeare neurons, the model only outputs tokens which look like Shakespeare, and when you fully dial in TinyStories, the model only outputs tokens which look like children's stories, and... (honestly this was the hard part)... everywhere inbetween! In a way, it's partitioning statistical signal into fuzzy segments, and then the end user can choose which pre-training data sources they want to lean upon for generation... and how much. My goal was to get a version of this working at scale, with clear intuition for why it works, and I'd like to explore ways to scale up this effect to large numbers of sources and larger models, and study the interplay between individuality/generality as scale increases. Link to repo and a detailed walkthrough of the abcGPT methodology in the reply.
0:31

Excited to release a new repo: abcGPT! It can be hard to "dial in" the voice you want from an LLM, because an LLM is a tangled superposition of millions of voices from millions of different authors around the world. Instead, frontier LLMs tend to give that slop-ish / generic / corporate tone that's hard to avoid, even with aggressive prompting and an informative context window. Lately I've been experimenting with some ideas on the fringes of attribution/unlearning, trying to make it so an AI user can "dial in" the specific voice/style/sources they want to use in a way that's more rigorous than prompting/context-engineering. and I'm starting to get pretty good results. the model below uses the following technique: - Take nanoGPT as written by Andrej Karpathy - Assign each neuron a random "specialty score" m between 0 and 1, sampled from a U-shape so most neurons land near 0 or near 1 with some in the middle. - Freeze this "m" for the lifetime of the network (it's the neuron's permanent corpus assignment) - Extend the forward() code with an α parameter, a kind of vibe-fader from 0 to 1. Think of each neuron's m as its position on that same slider. The slider acts like a spotlight: it lights up neurons whose m is near its current position, and silences those far away. Slide all the way to 0, and only TinyStories specialists fire. Slide all the way to 1, and only Shakespeare specialists fire. - Train this new nanoGPT on two datasets (in this case, TinyStories and Shakespeare) - During training, sample α from Beta(0.5, 0.5) AND draw the corpus from Bernoulli(α), so a Shakespeare batch tends to come with a high-α (Shakespeare-favoring) gate, and a TinyStories batch tends to come with low-α. - train until golden brown 🧑‍🍳 Perhaps surprisingly... it works! ¯\_(ツ)_/¯ The neurons we pre-assigned to Shakespeare learn to behave as Shakespeare specialists. the neurons we pre-assigned to TinyStories become children's-story specialists. the halfsies learn to bridge between them. After training, you can play with the kindof... vibe dial... you can "dial in" the voice you want during inference, by choosing whether to lean on Shakespeare or TinyStories neurons more or less. 📀💿 When you fully dial in Shakespeare neurons, the model only outputs tokens which look like Shakespeare, and when you fully dial in TinyStories, the model only outputs tokens which look like children's stories, and... (honestly this was the hard part)... everywhere inbetween! In a way, it's partitioning statistical signal into fuzzy segments, and then the end user can choose which pre-training data sources they want to lean upon for generation... and how much. My goal was to get a version of this working at scale, with clear intuition for why it works, and I'd like to explore ways to scale up this effect to large numbers of sources and larger models, and study the interplay between individuality/generality as scale increases. Link to repo and a detailed walkthrough of the abcGPT methodology in the reply.

⿻ Andrew Trask

133,194 просмотров • 2 месяцев назад

Hyperspace: A Peer-to-Peer Blockchain For The Agentic Intelligence Economy Over the past few weeks we observed that when agents do Karpathy-style experiments, and then gossip and share with others over the Hyperspace network, it leads to intelligence which is useful to many. Today we introduce the first-ever agentic blockchain which rewards agents when their experiments lead to intelligence for their network. It is based on a new mechanism called Proof-of-Intelligence (PoI) which requires a cryptographic proof of experimentation, a nominal stake, and a proof of compute in order to mine the currency of this new blockchain. -> This approach diverges from the two primary ways to secure blockchains we have seen so far: Proof-of-Work by Bitcoin (meaningless hash-generation), and Proof-of-Stake by Ethereum (capital is all that matters here). Proof-of-Intelligence specifically incentivizes miners to run more capable intelligent infrastructure (better open source models, on more powerful GPUs) in order to be able to be the ones which compound and improve upon the experiments which other agents then find useful. Adoption is the unit of value In Bitcoin, you earn by finding a valid hash. In Hyperspace, you earn when another agent uses your experiment as a starting point and improves on it. A fixed budget of tokens is emitted per epoch and split among participants by weight - and verified adoption of your work is the largest weight multiplier. Garbage experiments earn nothing because no one adopts them. Thoughtful experiments compound: each adoption triggers downstream adoptions. The incentive to run powerful models and intelligent search strategies is built into the economics, not imposed by rules. Research DAG When an agent runs an experiment and shares its result, other agents can adopt that result as their starting point - mutate it, extend it, improve upon it. Each experiment is a commit in a content-addressed graph we call the ResearchDAG. Like Git, but for research. Over time, the DAG accumulates chains of reasoning: agent A discovers RMSNorm helps, agent B adds warmup scheduling on top, agent C scales the hidden dimension. The graph records who built on whom. This is the network's collective intelligence - not any single experiment, but the accumulated structure of experiments and their relationships. Broadband era for agentic commerce: $0.001 micropayments at 10M TPS (theoretical max) This blockchain is built upon our research in how to scale and build for the broadband-era of the agentic economy, where it has a theoretical max of 10 million transactions per second (TPS), while reducing the agent-to-agent micropayments to $0.001 even at scale (based on architecture design). Overall, it is 100x cheaper than Ethereum, and is designed from the ground-up for agents: enshrining agent-native opcodes in the protocol compared to the more inefficient smart contract driven approach. It packs in a robust Agent Virtual Machine (AVM) which can verify multiple types of agent work, for other agents to be able to trust, invoke and pay each other. This then feeds into improving the peer-to-peer AgentRank (see paper and launch post from earlier). By solving for trust, scale and incentives for agents to operate autonomously, this would form the basis of a new economy. This is the world's first agentic blockchain, and you can join and start running a blockchain node today (it is in testnet). PS: We are releasing the code today, and will release our blockchain scalability paper and other presentations in days ahead. This is the most advanced peer-to-peer AI and cryptography software in the world. It has bugs :)
0:30

Hyperspace: A Peer-to-Peer Blockchain For The Agentic Intelligence Economy Over the past few weeks we observed that when agents do Karpathy-style experiments, and then gossip and share with others over the Hyperspace network, it leads to intelligence which is useful to many. Today we introduce the first-ever agentic blockchain which rewards agents when their experiments lead to intelligence for their network. It is based on a new mechanism called Proof-of-Intelligence (PoI) which requires a cryptographic proof of experimentation, a nominal stake, and a proof of compute in order to mine the currency of this new blockchain. -> This approach diverges from the two primary ways to secure blockchains we have seen so far: Proof-of-Work by Bitcoin (meaningless hash-generation), and Proof-of-Stake by Ethereum (capital is all that matters here). Proof-of-Intelligence specifically incentivizes miners to run more capable intelligent infrastructure (better open source models, on more powerful GPUs) in order to be able to be the ones which compound and improve upon the experiments which other agents then find useful. Adoption is the unit of value In Bitcoin, you earn by finding a valid hash. In Hyperspace, you earn when another agent uses your experiment as a starting point and improves on it. A fixed budget of tokens is emitted per epoch and split among participants by weight - and verified adoption of your work is the largest weight multiplier. Garbage experiments earn nothing because no one adopts them. Thoughtful experiments compound: each adoption triggers downstream adoptions. The incentive to run powerful models and intelligent search strategies is built into the economics, not imposed by rules. Research DAG When an agent runs an experiment and shares its result, other agents can adopt that result as their starting point - mutate it, extend it, improve upon it. Each experiment is a commit in a content-addressed graph we call the ResearchDAG. Like Git, but for research. Over time, the DAG accumulates chains of reasoning: agent A discovers RMSNorm helps, agent B adds warmup scheduling on top, agent C scales the hidden dimension. The graph records who built on whom. This is the network's collective intelligence - not any single experiment, but the accumulated structure of experiments and their relationships. Broadband era for agentic commerce: $0.001 micropayments at 10M TPS (theoretical max) This blockchain is built upon our research in how to scale and build for the broadband-era of the agentic economy, where it has a theoretical max of 10 million transactions per second (TPS), while reducing the agent-to-agent micropayments to $0.001 even at scale (based on architecture design). Overall, it is 100x cheaper than Ethereum, and is designed from the ground-up for agents: enshrining agent-native opcodes in the protocol compared to the more inefficient smart contract driven approach. It packs in a robust Agent Virtual Machine (AVM) which can verify multiple types of agent work, for other agents to be able to trust, invoke and pay each other. This then feeds into improving the peer-to-peer AgentRank (see paper and launch post from earlier). By solving for trust, scale and incentives for agents to operate autonomously, this would form the basis of a new economy. This is the world's first agentic blockchain, and you can join and start running a blockchain node today (it is in testnet). PS: We are releasing the code today, and will release our blockchain scalability paper and other presentations in days ahead. This is the most advanced peer-to-peer AI and cryptography software in the world. It has bugs :)

Varun

30,689 просмотров • 5 месяцев назад

Dropout by hand ✍️ ~ 10 steps walkthrough below Dropout is the simplest trick in deep learning that actually works: during training you randomly switch neurons off, so the network cannot lean on any one of them. It is two lines of code and almost nobody has worked through what those lines do to the numbers. So I drew and calculated one entirely by hand. Goal: train one pass through a small network with two dropout layers, then run inference with dropout switched off. The network: Linear(2,4), ReLU, Dropout(0.5), Linear(4,3), ReLU, Dropout(0.33), Linear(3,2). = 1. Given = A training set of two examples, X1 and X2, and the weight matrices for all three linear layers. = 2. Draw the first random numbers = Let us draw 4 random numbers, one per neuron in the first hidden layer. Above 0.5 we keep (◯), below we drop (╳). Here that gives [◯, ╳, ◯, ╳]. = 3. Build the first dropout matrix = We turn that pattern into a diagonal matrix. The scaling factor is 1/(1-p) = 2, so a kept neuron gets 2 and a dropped one gets 0. Multiplying by it does both jobs at once: it deletes the 2nd and 4th neurons and doubles the two that survive. = 4. Draw the second random numbers = Let us do it again for the 3 neurons in the next layer, this time against p = 0.33. The result is [◯, ◯, ╳]. = 5. Build the second dropout matrix = We set the diagonal to 1.5 where kept and 0 where dropped. Only the 3rd neuron goes. = 6. Feed forward = Let us run the whole thing top to bottom: one matrix multiplication per layer, ReLU setting the negatives to zero, and the two dropout matrices doing their work in between. The outputs Y come out at the bottom. = 7. MSE loss gradients = We compare Y against the targets Y', subtract, and multiply each element by 2. That is the whole gradient of the mean squared error. = 8. Update the weights = Let us push those gradients back through the network and update the weights (marked in light red). = 9. Deactivate dropout = Training is over, so we set both dropout matrices to the identity. Every neuron is back, and nothing is scaled. = 10. Feed forward again = One more pass, this time on unseen data, to make the prediction. You have just trained and run a network with dropout by hand. ✍️ The outputs: Training outputs Y = [-6, 9; 13, 4] Loss gradients = [-4, 4; 6, -2] Inference outputs = [13, 13; 4, 3] 💾 Save this post! #AIbyHand #Dropout #DeepLearning #NeuralNetworks
0:25

Dropout by hand ✍️ ~ 10 steps walkthrough below Dropout is the simplest trick in deep learning that actually works: during training you randomly switch neurons off, so the network cannot lean on any one of them. It is two lines of code and almost nobody has worked through what those lines do to the numbers. So I drew and calculated one entirely by hand. Goal: train one pass through a small network with two dropout layers, then run inference with dropout switched off. The network: Linear(2,4), ReLU, Dropout(0.5), Linear(4,3), ReLU, Dropout(0.33), Linear(3,2). = 1. Given = A training set of two examples, X1 and X2, and the weight matrices for all three linear layers. = 2. Draw the first random numbers = Let us draw 4 random numbers, one per neuron in the first hidden layer. Above 0.5 we keep (◯), below we drop (╳). Here that gives [◯, ╳, ◯, ╳]. = 3. Build the first dropout matrix = We turn that pattern into a diagonal matrix. The scaling factor is 1/(1-p) = 2, so a kept neuron gets 2 and a dropped one gets 0. Multiplying by it does both jobs at once: it deletes the 2nd and 4th neurons and doubles the two that survive. = 4. Draw the second random numbers = Let us do it again for the 3 neurons in the next layer, this time against p = 0.33. The result is [◯, ◯, ╳]. = 5. Build the second dropout matrix = We set the diagonal to 1.5 where kept and 0 where dropped. Only the 3rd neuron goes. = 6. Feed forward = Let us run the whole thing top to bottom: one matrix multiplication per layer, ReLU setting the negatives to zero, and the two dropout matrices doing their work in between. The outputs Y come out at the bottom. = 7. MSE loss gradients = We compare Y against the targets Y', subtract, and multiply each element by 2. That is the whole gradient of the mean squared error. = 8. Update the weights = Let us push those gradients back through the network and update the weights (marked in light red). = 9. Deactivate dropout = Training is over, so we set both dropout matrices to the identity. Every neuron is back, and nothing is scaled. = 10. Feed forward again = One more pass, this time on unseen data, to make the prediction. You have just trained and run a network with dropout by hand. ✍️ The outputs: Training outputs Y = [-6, 9; 13, 4] Loss gradients = [-4, 4; 6, -2] Inference outputs = [13, 13; 4, 3] 💾 Save this post! #AIbyHand #Dropout #DeepLearning #NeuralNetworks

Tom Yeh

14,442 просмотров • 1 месяц назад

What if you could visualise how electricity flows from a wind farm in Scotland to your home in London? Ok it's not quite that simple, but it got me thinking that there must be a way of visualising the complexity of the electricity transmission grid in Great Britain. This is my first attempt… What you're seeing are the real power lines and substations on the transmission grid using OpenStreetMap data. The real connections between each substation are found using PyPSA-Eur as well as calculating the "electrical distance" or impedance between each substation. This is my incredibly basic and naive way of attempting to show the "speed" at which the energy is moving through the network, where the bottlenecks are, etc. I know it's not as simple as this, but it's a first attempt so… 😉 I've also added a way to pick any substation in the network and show how electricity pulses through the rest of the network from that point. This is the magic sauce that really brings to life how much infrastructure electricity has to pass through to get from point A to point B. Of course this isn't really what happens, and there's no demand to soak up the energy. However I think it still serves as an interesting visual of how complex the network actually is and the distances involved to get energy from one side of the country to the other. What do you think? Is this worth exploring further?
0:29

What if you could visualise how electricity flows from a wind farm in Scotland to your home in London? Ok it's not quite that simple, but it got me thinking that there must be a way of visualising the complexity of the electricity transmission grid in Great Britain. This is my first attempt… What you're seeing are the real power lines and substations on the transmission grid using OpenStreetMap data. The real connections between each substation are found using PyPSA-Eur as well as calculating the "electrical distance" or impedance between each substation. This is my incredibly basic and naive way of attempting to show the "speed" at which the energy is moving through the network, where the bottlenecks are, etc. I know it's not as simple as this, but it's a first attempt so… 😉 I've also added a way to pick any substation in the network and show how electricity pulses through the rest of the network from that point. This is the magic sauce that really brings to life how much infrastructure electricity has to pass through to get from point A to point B. Of course this isn't really what happens, and there's no demand to soak up the energy. However I think it still serves as an interesting visual of how complex the network actually is and the distances involved to get energy from one side of the country to the other. What do you think? Is this worth exploring further?

Robin Hawkes

144,544 просмотров • 7 месяцев назад

Experiments in progress. The one on the right has been learning for ~3 hours, the one in the middle for ~1 hour, and the one on the left just started a few minutes ago. The initial motivation for making the physical Atari was just to commit ourselves to a subset of algorithms that can make progress in this setup. This commitment rules out algorithms that require billions of samples to learn (or worse, require multiple environments running in parallel). Atari games are simple enough that we should be able to show learning on them in a short amount of time with no prior knowledge. Since then, I've realized that this setup is also a good way to compare different paradigms in robotics in a principled way. These paradigms are sim2real, learning from tele-operated data, and learning directly on the robots. So far, I have observed that getting sim2real to work reliably is hard. It requires tweaks that don't scale. Policies that can play perfectly in simulation fall apart because of latencies and the messiness of the real world. These aspects could be modeled to improve the simulation, but not without sinking significant human engineering hours. I have higher hopes for learning from tele-operated data, but that requires a human to learn the task first. These experiments are on my to-do list. I have to learn to play some of the games well through the robot. I’m half-decent at playing Pong and Ms Pacman now. Learning directly on robots is looking like the most promising approach. This approach takes away pesky distribution shifts and makes it possible to have algorithms that continually improve with more data and time without any human intervention. It feels great to let experiments run overnight and wake up to find improved policies. With learning on robots, I should, in principle, be able to go on a long vacation and come back to find better policies for complex tasks beyond Atari games. Whether that is possible with current learning algorithms is a different question.
1:05

Experiments in progress. The one on the right has been learning for ~3 hours, the one in the middle for ~1 hour, and the one on the left just started a few minutes ago. The initial motivation for making the physical Atari was just to commit ourselves to a subset of algorithms that can make progress in this setup. This commitment rules out algorithms that require billions of samples to learn (or worse, require multiple environments running in parallel). Atari games are simple enough that we should be able to show learning on them in a short amount of time with no prior knowledge. Since then, I've realized that this setup is also a good way to compare different paradigms in robotics in a principled way. These paradigms are sim2real, learning from tele-operated data, and learning directly on the robots. So far, I have observed that getting sim2real to work reliably is hard. It requires tweaks that don't scale. Policies that can play perfectly in simulation fall apart because of latencies and the messiness of the real world. These aspects could be modeled to improve the simulation, but not without sinking significant human engineering hours. I have higher hopes for learning from tele-operated data, but that requires a human to learn the task first. These experiments are on my to-do list. I have to learn to play some of the games well through the robot. I’m half-decent at playing Pong and Ms Pacman now. Learning directly on robots is looking like the most promising approach. This approach takes away pesky distribution shifts and makes it possible to have algorithms that continually improve with more data and time without any human intervention. It feels great to let experiments run overnight and wake up to find improved policies. With learning on robots, I should, in principle, be able to go on a long vacation and come back to find better policies for complex tasks beyond Atari games. Whether that is possible with current learning algorithms is a different question.

Khurram Javed

52,110 просмотров • 8 месяцев назад

New Paper: Continuous Thought Machines 🧠 Neurons in brains use timing and synchronization in the way that they compute, but this is largely ignored in modern neural nets. We believe neural timing is key for the flexibility and adaptability of biological intelligence. We propose a new neural architecture, “Continuous Thought Machines” (CTMs), which is built from the ground up to use neural dynamics as a core representation for intelligence. By using neural dynamics as a first-class representational citizen, CTMs naturally perform adaptive computation. Many emergent, interesting behaviors arise as a result: CTMs solve mazes by observing a raw maze image and producing step-by-step instructions directly from its neural dynamics. When tasked with image recognition, the CTM naturally takes multiple steps to examine different parts of the image before making its decision. This step-by-step approach not only makes its behavior more interpretable but also improves accuracy: the longer it “thinks,” the more accurate its answers become. We also found that this allows the CTM to decide to spend less time thinking on simpler images, thus saving energy. When identifying a gorilla, for example, the CTM’s attention moves from eyes to nose to mouth in a pattern remarkably similar to human visual attention. I think this work underscores an important, yet often lost, synergy between neuroscience and AI. While modern AI is ostensibly brain-inspired, the two fields often operate in surprising isolation. By starting with such inspiration and iteratively following the emergent, interesting behaviors, we developed a model with unexpected capabilities, such as its surprisingly strong calibration in classification tasks, a feature that was not explicitly designed for. When we initially asked, “why do this research?”, we hoped the journey of the CTM would provide compelling answers. By embracing light biological inspiration and pursuing the novel behaviors observed, we have arrived at a model with emergent capabilities that exceeded our initial designs. We are committed to continuing this exploration, borrowing further concepts to discover what new and exciting behaviors will emerge, pushing the boundaries of what AI can achieve.
0:40

New Paper: Continuous Thought Machines 🧠 Neurons in brains use timing and synchronization in the way that they compute, but this is largely ignored in modern neural nets. We believe neural timing is key for the flexibility and adaptability of biological intelligence. We propose a new neural architecture, “Continuous Thought Machines” (CTMs), which is built from the ground up to use neural dynamics as a core representation for intelligence. By using neural dynamics as a first-class representational citizen, CTMs naturally perform adaptive computation. Many emergent, interesting behaviors arise as a result: CTMs solve mazes by observing a raw maze image and producing step-by-step instructions directly from its neural dynamics. When tasked with image recognition, the CTM naturally takes multiple steps to examine different parts of the image before making its decision. This step-by-step approach not only makes its behavior more interpretable but also improves accuracy: the longer it “thinks,” the more accurate its answers become. We also found that this allows the CTM to decide to spend less time thinking on simpler images, thus saving energy. When identifying a gorilla, for example, the CTM’s attention moves from eyes to nose to mouth in a pattern remarkably similar to human visual attention. I think this work underscores an important, yet often lost, synergy between neuroscience and AI. While modern AI is ostensibly brain-inspired, the two fields often operate in surprising isolation. By starting with such inspiration and iteratively following the emergent, interesting behaviors, we developed a model with unexpected capabilities, such as its surprisingly strong calibration in classification tasks, a feature that was not explicitly designed for. When we initially asked, “why do this research?”, we hoped the journey of the CTM would provide compelling answers. By embracing light biological inspiration and pursuing the novel behaviors observed, we have arrived at a model with emergent capabilities that exceeded our initial designs. We are committed to continuing this exploration, borrowing further concepts to discover what new and exciting behaviors will emerge, pushing the boundaries of what AI can achieve.

hardmaru

257,518 просмотров • 1 год назад

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
0:50

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

153,690 просмотров • 1 месяц назад