正在加载视频...

视频加载失败

返回首页

➡️ Machine Learning from Scratch, part 15: a recommender, the kind of algorithm that picks what your feed shows you. Our app is simulated: 300 people and 60 videos. Every person has a hidden taste, every video has a hidden style, and each person has watched only about 12...

machinelearningtogo's profile picture

MachineLearningToGo

28,000 subscribers

88,400 次观看 • 1 个月前 •via TikTok

科学技术#machinelearning#ai#recommendersystem#python#coding
LIVE

Anya Rossi• Live Now

Private livecam show

0 条评论

暂无评论

原始帖子的评论将显示在这里

相关视频

➡️ Machine Learning from Scratch, part 31: the transformer. A transformer turns every token into a list of numbers, adds its position, and then lets every token look at the others. That step is called attention: each token has a query (what am I looking for), a key (what do I contain) and a value (what I pass on). The better a query matches a key, the more of that value flows in. Real transformers repeat this in many layers with several attention heads each. The design comes from the 2017 Google paper "Attention Is All You Need" by Ashish Vaswani and colleagues, first used for translation. A large language model is a transformer trained to predict the next token, again and again, on a huge amount of text. Our own run, nothing downloaded. attention.py is one self-attention layer in 16 lines of numpy. Every word gets a query, a key and a value of 8 numbers, starting random, trained from scratch with hand-written backpropagation on 192 made-up sentences like "the dog ate the carrot because it was hungry". The task: find out which pet "it" means. On 64 new test sentences "it" puts 99.95 percent of its attention on the right word on average (the lowest single sentence gets 99.9 percent; the program prints this as 100%). Before training it was 10 percent, spread evenly. The cat sentence from the video is one of the test sentences. Honest notes: our task is easy on purpose. Every sentence has exactly one pet, so "it" only has to learn to look for pets, and it learns that within the first pass. Real sentences with two candidates need more context, which is what many layers are for. Our small version has no positions, no scaling, no mask, one head, and only "it" asks a question. Every word of the test sentences also appears in training; the combinations are new. The 3D space in the video shows the real keys of our model, projected to three axes (two of them stretched so you can see them). The stacked layers are a sketch. #machinelearning #ai #transformer #python #llm
2:03

➡️ Machine Learning from Scratch, part 31: the transformer. A transformer turns every token into a list of numbers, adds its position, and then lets every token look at the others. That step is called attention: each token has a query (what am I looking for), a key (what do I contain) and a value (what I pass on). The better a query matches a key, the more of that value flows in. Real transformers repeat this in many layers with several attention heads each. The design comes from the 2017 Google paper "Attention Is All You Need" by Ashish Vaswani and colleagues, first used for translation. A large language model is a transformer trained to predict the next token, again and again, on a huge amount of text. Our own run, nothing downloaded. attention.py is one self-attention layer in 16 lines of numpy. Every word gets a query, a key and a value of 8 numbers, starting random, trained from scratch with hand-written backpropagation on 192 made-up sentences like "the dog ate the carrot because it was hungry". The task: find out which pet "it" means. On 64 new test sentences "it" puts 99.95 percent of its attention on the right word on average (the lowest single sentence gets 99.9 percent; the program prints this as 100%). Before training it was 10 percent, spread evenly. The cat sentence from the video is one of the test sentences. Honest notes: our task is easy on purpose. Every sentence has exactly one pet, so "it" only has to learn to look for pets, and it learns that within the first pass. Real sentences with two candidates need more context, which is what many layers are for. Our small version has no positions, no scaling, no mask, one head, and only "it" asks a question. Every word of the test sentences also appears in training; the combinations are new. The 3D space in the video shows the real keys of our model, projected to three axes (two of them stretched so you can see them). The stacked layers are a sketch. #machinelearning #ai #transformer #python #llm

MachineLearningToGo

80,100 次观看 • 1 个月前

Mark Daigneault shares the coaching philosophy that drives the Oklahoma City Thunder. "Every player is different. And every player is a person first. I've always relied on the instincts of connecting with the person first " Then he shared what professional basketball taught him: "All of these players have a reason to admire them and not just respect them. That's a baseline for any player." "If a player's made it this far, there's a reason to admire them." It means caring about the person. "Expressing that admiration, accepting them for who they are, and then trying to help them be the best they can be. Chase down the vision they have for who they are as a player." "But also plug that into the larger structure of a team in an environment that elevates everybody." The best coaches don't just manage players. They see the person behind the jersey and the team. "It starts with every person. Every person's different - connecting to that and tapping into that." Great coaches care about the person and their future The heart of coaching is a relationships business not a sports business (🎥 OKC Thunder)
1:03

Mark Daigneault shares the coaching philosophy that drives the Oklahoma City Thunder. "Every player is different. And every player is a person first. I've always relied on the instincts of connecting with the person first " Then he shared what professional basketball taught him: "All of these players have a reason to admire them and not just respect them. That's a baseline for any player." "If a player's made it this far, there's a reason to admire them." It means caring about the person. "Expressing that admiration, accepting them for who they are, and then trying to help them be the best they can be. Chase down the vision they have for who they are as a player." "But also plug that into the larger structure of a team in an environment that elevates everybody." The best coaches don't just manage players. They see the person behind the jersey and the team. "It starts with every person. Every person's different - connecting to that and tapping into that." Great coaches care about the person and their future The heart of coaching is a relationships business not a sports business (🎥 OKC Thunder)

Coach AJ 🎯 Mental Fitness

61,209 次观看 • 6 个月前

Declaration of Digital Independence We hold these truths to be self-evident in the digital age: That every person has the right to speak without surveillance. That every person has the right to think without manipulation. That every person has the right to move through the world without being tracked. That every person has the right to communicate privately. That no corporation and no government holds any legitimate authority to override these rights, and that when any system of power becomes destructive of these ends, it is the right of the people to build something different. Full video here:
0:50

Declaration of Digital Independence We hold these truths to be self-evident in the digital age: That every person has the right to speak without surveillance. That every person has the right to think without manipulation. That every person has the right to move through the world without being tracked. That every person has the right to communicate privately. That no corporation and no government holds any legitimate authority to override these rights, and that when any system of power becomes destructive of these ends, it is the right of the people to build something different. Full video here:

Naomi Brockwell priv/acc

15,474 次观看 • 3 个月前

4 MINUTES INTO A VIDEO CALL, I ASKED ASH WHO SHE IS. ASH IS AN AI. Tavus gave me early access to Griffin, the first human interaction model. This clip is from my session. She has a whole life. Lives in Brooklyn, makes videos about pop culture, could talk my ear off about The Last of Us. Nobody types for her or picks her gestures. She listens while I speak, smiles in the right places and answers live. The numbers behind it: - 45% of 120 people watched a 90-second call with it and thought it was a real person - #1 on NVIDIA's Video Full-Duplex Benchmark, ahead of every Gemini, OpenAI and open-source model evaluated - 3.83 out of 5 on that benchmark, a human gets 3.92, the next best system 2.80 Tavus calls it the first AI to pass the Turing test. What you just watched is Griffin Lite, the small one. The big Griffin stays unreleased. Tavus Hassaan Raza
0:34

4 MINUTES INTO A VIDEO CALL, I ASKED ASH WHO SHE IS. ASH IS AN AI. Tavus gave me early access to Griffin, the first human interaction model. This clip is from my session. She has a whole life. Lives in Brooklyn, makes videos about pop culture, could talk my ear off about The Last of Us. Nobody types for her or picks her gestures. She listens while I speak, smiles in the right places and answers live. The numbers behind it: - 45% of 120 people watched a 90-second call with it and thought it was a real person - #1 on NVIDIA's Video Full-Duplex Benchmark, ahead of every Gemini, OpenAI and open-source model evaluated - 3.83 out of 5 on that benchmark, a human gets 3.92, the next best system 2.80 Tavus calls it the first AI to pass the Turing test. What you just watched is Griffin Lite, the small one. The big Griffin stays unreleased. Tavus Hassaan Raza

Sethian

120,922 次观看 • 1 个月前

➡️ Part 9 of learning ML code from scratch Six weights, one wrong answer, and not one of them knows it was at fault. Backpropagation is how they find out. Two questions guide the video: what even is backpropagation, and why do models need it? Step 1, the forward pass. Two numbers go into a tiny network, run through a hidden layer, and one number comes out. Ours says 0.700. We wanted 1.0, so square the difference and the loss is 0.09. That single number is how wrong the model is. Step 2, the question. Six weights produced that answer together. Which of them is to blame, and by how much? Step 3, the one idea. Change a single weight by a tiny amount and watch the error move. How much the error changes per unit of weight is the gradient of that weight. In the video you can see it happen: the weight wobbles and the error bar answers. Step 4, the shortcut. Every weight running into the same neuron starts from one shared number, the blame of that neuron. Multiply that blame by the value each weight carried and you have its gradient. One blame, computed once, and every weight into that neuron gets its own gradient from it. That is why backpropagation is fast enough to train anything at all. Step 5, backwards. The blame starts at the output and every layer hands its share to the layer in front of it, back through the same network the data came forward through. It goes layer by layer, and inside one layer every neuron is done together. Step 6, the update. The optimizer moves every weight against its own gradient, the whole weight matrix in one go, so no weight waits for its turn. All six gradients here are negative, so all six weights go up. After one step the answer reads 0.769. After sixty it reads 0.968 and the error is down to 0.001. The honest part: every gradient in the video was checked against the measured change of the error. Move that one weight, see what the error does, divide. Backprop and the measurement agree to nine decimal places, otherwise nothing would have rendered. #machinelearning #backpropagation #neuralnetworks #python #coding
2:18

➡️ Part 9 of learning ML code from scratch Six weights, one wrong answer, and not one of them knows it was at fault. Backpropagation is how they find out. Two questions guide the video: what even is backpropagation, and why do models need it? Step 1, the forward pass. Two numbers go into a tiny network, run through a hidden layer, and one number comes out. Ours says 0.700. We wanted 1.0, so square the difference and the loss is 0.09. That single number is how wrong the model is. Step 2, the question. Six weights produced that answer together. Which of them is to blame, and by how much? Step 3, the one idea. Change a single weight by a tiny amount and watch the error move. How much the error changes per unit of weight is the gradient of that weight. In the video you can see it happen: the weight wobbles and the error bar answers. Step 4, the shortcut. Every weight running into the same neuron starts from one shared number, the blame of that neuron. Multiply that blame by the value each weight carried and you have its gradient. One blame, computed once, and every weight into that neuron gets its own gradient from it. That is why backpropagation is fast enough to train anything at all. Step 5, backwards. The blame starts at the output and every layer hands its share to the layer in front of it, back through the same network the data came forward through. It goes layer by layer, and inside one layer every neuron is done together. Step 6, the update. The optimizer moves every weight against its own gradient, the whole weight matrix in one go, so no weight waits for its turn. All six gradients here are negative, so all six weights go up. After one step the answer reads 0.769. After sixty it reads 0.968 and the error is down to 0.001. The honest part: every gradient in the video was checked against the measured change of the error. Move that one weight, see what the error does, divide. Backprop and the measurement agree to nine decimal places, otherwise nothing would have rendered. #machinelearning #backpropagation #neuralnetworks #python #coding

MachineLearningToGo

210,600 次观看 • 1 个月前

People are making stupid money with AI — and they're no geniuses. Here is how All respect to them, I mean that. But most of them are regular people who learned ONE skill in a short amount of time and ran one recipe with it. You've probably seen it — somebody around you quietly earning with AI while you wonder what the secret is. Luck. Genius. Timing. It's a recipe with two parts, and almost everybody only ever does part one. PART ONE — BUILD something people actually want. With AI this has never been easier: an automation that saves a business owner ten hours a week, a tool for one specific job, a service delivered faster than anyone charging double. A few focused months for the skills, a couple of weeks for the build. Part one is necessary. And part one, alone, pays you nothing. Because that's the beautiful store you open in the desert. Perfect design, real work, months of grind — nobody walks past. That's where almost everyone quits, convinced it was luck after all. PART TWO — DISTRIBUTION. The actual secret. Put the valuable thing in front of the RIGHT people, AT SCALE. Two equations: MATCHING — speak your buyer's language. Built a tool for fitness trainers? Talk to trainers about their clients, their programs, their day, in trainer words. The algorithm reads your language and hands your videos to the people who speak it. SCALE — one video is a hope. Hundreds are a system. The content engine: make three videos, it turns them into fifteen posts across every platform, every week. One person shows up like a media company. Look at anyone actually winning with AI: they built something worth wanting, then made sure thousands of the right people saw it. Both halves, every time. It was never luck. It's build, times distribution. #aiagent #aiautomation #makemoneywithai
2:33

People are making stupid money with AI — and they're no geniuses. Here is how All respect to them, I mean that. But most of them are regular people who learned ONE skill in a short amount of time and ran one recipe with it. You've probably seen it — somebody around you quietly earning with AI while you wonder what the secret is. Luck. Genius. Timing. It's a recipe with two parts, and almost everybody only ever does part one. PART ONE — BUILD something people actually want. With AI this has never been easier: an automation that saves a business owner ten hours a week, a tool for one specific job, a service delivered faster than anyone charging double. A few focused months for the skills, a couple of weeks for the build. Part one is necessary. And part one, alone, pays you nothing. Because that's the beautiful store you open in the desert. Perfect design, real work, months of grind — nobody walks past. That's where almost everyone quits, convinced it was luck after all. PART TWO — DISTRIBUTION. The actual secret. Put the valuable thing in front of the RIGHT people, AT SCALE. Two equations: MATCHING — speak your buyer's language. Built a tool for fitness trainers? Talk to trainers about their clients, their programs, their day, in trainer words. The algorithm reads your language and hands your videos to the people who speak it. SCALE — one video is a hope. Hundreds are a system. The content engine: make three videos, it turns them into fifteen posts across every platform, every week. One person shows up like a media company. Look at anyone actually winning with AI: they built something worth wanting, then made sure thousands of the right people saw it. Both halves, every time. It was never luck. It's build, times distribution. #aiagent #aiautomation #makemoneywithai

Professor Glitch

10,200 次观看 • 2 个月前

➡️ Machine Learning from Scratch, part 47: the Hopfield network. In 1982 the physicist John Hopfield described a network that works as a memory, and 42 years later this work was part of the Nobel Prize in Physics. Each neuron is one pixel that is either on or off, and every neuron is connected to every other neuron, so 100 pixels need 4,950 connections. There are no layers: signals go back and forth until the network settles. Each neuron takes a weighted sum of its inputs and switches on if the sum is positive, off if it is negative. Storing a picture is simple: for every pair of pixels with the same colour the connection gets stronger, otherwise weaker. Every stored picture becomes a valley in the energy landscape, and no switch can ever raise the energy, so a noisy picture rolls downhill into the nearest stored memory. The code (hopfield.py, 16 lines of numpy, pictures in pics.txt) stores three 10 x 10 pictures, a heart, a house and a smiley, and flips 30 of the 100 pixels in each. After a single pass over all neurons all three come back exactly; the energy of the heart falls from -6.90 to -48.90, the energy of the stored heart. With 100 neurons and random patterns (10 flipped pixels, 20 runs) it recalls 5 patterns every time, 10 patterns in 94.5 % of the cases, 14 in 68 %, 20 in about one in four (24 %) and 30 in 1 %. Honest notes: the weights are kept as whole numbers (100 times the Hebb weight), so a sum of exactly zero switches the neuron on, as the rule says. The landscape in the video is a picture of the idea; the valley floors and the height of the rolling ball are the real energies from the run. #machinelearning #ai #coding #neuralnetworks #nobelprize
2:13

➡️ Machine Learning from Scratch, part 47: the Hopfield network. In 1982 the physicist John Hopfield described a network that works as a memory, and 42 years later this work was part of the Nobel Prize in Physics. Each neuron is one pixel that is either on or off, and every neuron is connected to every other neuron, so 100 pixels need 4,950 connections. There are no layers: signals go back and forth until the network settles. Each neuron takes a weighted sum of its inputs and switches on if the sum is positive, off if it is negative. Storing a picture is simple: for every pair of pixels with the same colour the connection gets stronger, otherwise weaker. Every stored picture becomes a valley in the energy landscape, and no switch can ever raise the energy, so a noisy picture rolls downhill into the nearest stored memory. The code (hopfield.py, 16 lines of numpy, pictures in pics.txt) stores three 10 x 10 pictures, a heart, a house and a smiley, and flips 30 of the 100 pixels in each. After a single pass over all neurons all three come back exactly; the energy of the heart falls from -6.90 to -48.90, the energy of the stored heart. With 100 neurons and random patterns (10 flipped pixels, 20 runs) it recalls 5 patterns every time, 10 patterns in 94.5 % of the cases, 14 in 68 %, 20 in about one in four (24 %) and 30 in 1 %. Honest notes: the weights are kept as whole numbers (100 times the Hebb weight), so a sum of exactly zero switches the neuron on, as the rule says. The landscape in the video is a picture of the idea; the valley floors and the height of the rolling ball are the real energies from the run. #machinelearning #ai #coding #neuralnetworks #nobelprize

MachineLearningToGo

12,000 次观看 • 14 天前

Mark Zuckerberg just described the obsolescence of every institution on Earth and delivered it like a product update. Zuckerberg: “I just think in the future almost everyone is gonna have the power of a 10,000-person organization.” He did not say better tools. He did not say smarter apps. He said the full cognitive output of ten thousand human beings. Packaged into a product. Handed to one person. That is not an upgrade. That is the end of the reason human beings organize at all. Companies exist because one brain is not enough. Governments exist because coordination requires hierarchy. Universities exist because knowledge demands infrastructure. Every institution ever built was a workaround for the same limitation. No single person could do it alone. Zuckerberg is telling you that limitation is about to disappear. The 500-person startup becomes one founder and an AI stack. The law firm becomes one attorney and a system that never sleeps. The hospital becomes one doctor carrying every specialist in their pocket. That is not speculation. That is a deployment schedule. And the man writing it runs a 70,000-person company. He employs 70,000 people and just told the world one person will soon need none of them. That is not a prediction. That is a confession from the man who will be first to act on it. But the part nobody is discussing matters more. This technology does not land on everyone equally. It lands first on the people who already command 10,000-person organizations. Zuckerberg does not get the power of 10,000 people. He already has that. He gets the power of 10,000 organizations. Every revolution in history was sold as liberation. The printing press was supposed to democratize knowledge. It built media empires. The internet was supposed to democratize commerce. It built trillion-dollar platforms. The tool always arrives as liberation. It always settles as leverage. And leverage always consolidates upward. Zuckerberg is not wrong about the capability. One person will do what ten thousand once did. But the question nobody is asking is the only one that matters. If everyone wields the output of 10,000 people, what is a single person actually worth? And then Zuckerberg answered his own question without realizing it. Zuckerberg: “If the intelligence of a 10,000-person company is not greater than the intelligence of a single person, then what are we doing here?” He meant it as a case for AI. That is the most brutal thing a CEO has ever said about the people who work for him.
0:59

Mark Zuckerberg just described the obsolescence of every institution on Earth and delivered it like a product update. Zuckerberg: “I just think in the future almost everyone is gonna have the power of a 10,000-person organization.” He did not say better tools. He did not say smarter apps. He said the full cognitive output of ten thousand human beings. Packaged into a product. Handed to one person. That is not an upgrade. That is the end of the reason human beings organize at all. Companies exist because one brain is not enough. Governments exist because coordination requires hierarchy. Universities exist because knowledge demands infrastructure. Every institution ever built was a workaround for the same limitation. No single person could do it alone. Zuckerberg is telling you that limitation is about to disappear. The 500-person startup becomes one founder and an AI stack. The law firm becomes one attorney and a system that never sleeps. The hospital becomes one doctor carrying every specialist in their pocket. That is not speculation. That is a deployment schedule. And the man writing it runs a 70,000-person company. He employs 70,000 people and just told the world one person will soon need none of them. That is not a prediction. That is a confession from the man who will be first to act on it. But the part nobody is discussing matters more. This technology does not land on everyone equally. It lands first on the people who already command 10,000-person organizations. Zuckerberg does not get the power of 10,000 people. He already has that. He gets the power of 10,000 organizations. Every revolution in history was sold as liberation. The printing press was supposed to democratize knowledge. It built media empires. The internet was supposed to democratize commerce. It built trillion-dollar platforms. The tool always arrives as liberation. It always settles as leverage. And leverage always consolidates upward. Zuckerberg is not wrong about the capability. One person will do what ten thousand once did. But the question nobody is asking is the only one that matters. If everyone wields the output of 10,000 people, what is a single person actually worth? And then Zuckerberg answered his own question without realizing it. Zuckerberg: “If the intelligence of a 10,000-person company is not greater than the intelligence of a single person, then what are we doing here?” He meant it as a case for AI. That is the most brutal thing a CEO has ever said about the people who work for him.

Dustin

54,061 次观看 • 6 个月前

Founder of lululemon on what he'd tell every 25 year old: "I'd tell them that every person in the world is an individual with a different genetic makeup and a different upbringing and the way that you're thinking is so radically different than every other person in the world and incomparable that if you have an idea and you want to move forward with it, don't worry so much about the competition because nobody will be able to replicate you and the way you think about it."
0:38

Founder of lululemon on what he'd tell every 25 year old: "I'd tell them that every person in the world is an individual with a different genetic makeup and a different upbringing and the way that you're thinking is so radically different than every other person in the world and incomparable that if you have an idea and you want to move forward with it, don't worry so much about the competition because nobody will be able to replicate you and the way you think about it."

Sam Stoffel

582,009 次观看 • 5 个月前

How a $50K AI project actually starts (process mapping, step by step) Before anyone builds anything, a real AI project starts with a stopwatch and a wall of sticky notes. Here's the whole method on one company: a 60-person parts distributor where 400 orders a day arrive as PDF attachments and six people retype them into the ERP. Interview the people who do the work, not the owner. The owner says taking an order is 5 steps. The person who's done it for nine years walks you through 41, including the sticky note on her monitor. Sit next to them with a stopwatch. Measure touch time (6 minutes of actual work) against lead time (3 hours from email to confirmation). The minutes cost wages. The hours cost customers. Break every box into atomic steps. One person, one system, one action. "Enter the order" is 11 steps. AI can't do "enter the order", but it's great at "match their part number to ours", and it tells you how sure it is. Hunt the exceptions. Old part numbers, price mismatches, backorders, credit holds, duplicate POs. Log them for two weeks. That's where the time goes. Give every step one of three colors. Blue is a rule, so automate it. Purple is judgment on messy input, so AI does it and a person approves when it's unsure. Orange is money or a relationship, so it stays human. Fix broken steps before you automate them. Score and sequence. Frequency × minutes × cost of a mistake gives you the roadmap. Track one number from day one: the share of orders nobody had to touch. Then the $50K has a shape: paid discovery, then phases that each pay for the next. Nobody loses their job. They stop retyping and handle the exceptions and the customers. #aiautomation #aiagent
3:17

How a $50K AI project actually starts (process mapping, step by step) Before anyone builds anything, a real AI project starts with a stopwatch and a wall of sticky notes. Here's the whole method on one company: a 60-person parts distributor where 400 orders a day arrive as PDF attachments and six people retype them into the ERP. Interview the people who do the work, not the owner. The owner says taking an order is 5 steps. The person who's done it for nine years walks you through 41, including the sticky note on her monitor. Sit next to them with a stopwatch. Measure touch time (6 minutes of actual work) against lead time (3 hours from email to confirmation). The minutes cost wages. The hours cost customers. Break every box into atomic steps. One person, one system, one action. "Enter the order" is 11 steps. AI can't do "enter the order", but it's great at "match their part number to ours", and it tells you how sure it is. Hunt the exceptions. Old part numbers, price mismatches, backorders, credit holds, duplicate POs. Log them for two weeks. That's where the time goes. Give every step one of three colors. Blue is a rule, so automate it. Purple is judgment on messy input, so AI does it and a person approves when it's unsure. Orange is money or a relationship, so it stays human. Fix broken steps before you automate them. Score and sequence. Frequency × minutes × cost of a mistake gives you the roadmap. Track one number from day one: the share of orders nobody had to touch. Then the $50K has a shape: paid discovery, then phases that each pay for the next. Nobody loses their job. They stop retyping and handle the exceptions and the customers. #aiautomation #aiagent

Professor Glitch

20,900 次观看 • 1 个月前

THIS ROBOT HAS A HUMAN FACE, AN OPEN TORSO FULL OF WIRES, AND SHE JUST TOOK A GLASS OF LEMON WATER SO NATURALLY THAT 200 PEOPLE STOPPED BREATHING! At first, most of the crowd thought it was a model in a costume. Then a team member handed her a glass of water with lemon, and she took it with both hands, adjusted her grip and looked at it like a person would. That is a much harder problem than it looks. The robot has to see the glass, judge its weight, control the force of each finger and avoid crushing or dropping it, all while a crowd is moving around her and phones are pointed at her face. Walking on stage was the old benchmark. Humans do small things like this thousands of times a day without thinking. A robot has to turn every one of them into sensors, motors and software. The strangest part is the contrast. The face and the movement say “human”, but the glowing mechanics in her torso and the metal in her leg say “machine”. Your brain keeps switching between the two and cannot settle on one. Every year the gap between “obviously a robot” and “wait, is that a person?” gets smaller.
0:26

THIS ROBOT HAS A HUMAN FACE, AN OPEN TORSO FULL OF WIRES, AND SHE JUST TOOK A GLASS OF LEMON WATER SO NATURALLY THAT 200 PEOPLE STOPPED BREATHING! At first, most of the crowd thought it was a model in a costume. Then a team member handed her a glass of water with lemon, and she took it with both hands, adjusted her grip and looked at it like a person would. That is a much harder problem than it looks. The robot has to see the glass, judge its weight, control the force of each finger and avoid crushing or dropping it, all while a crowd is moving around her and phones are pointed at her face. Walking on stage was the old benchmark. Humans do small things like this thousands of times a day without thinking. A robot has to turn every one of them into sensors, motors and software. The strangest part is the contrast. The face and the movement say “human”, but the glowing mechanics in her torso and the metal in her leg say “machine”. Your brain keeps switching between the two and cannot settle on one. Every year the gap between “obviously a robot” and “wait, is that a person?” gets smaller.

ssht

25,868 次观看 • 24 天前

Basically, whatever burning topic or issue arises, the news appears first on Twitter. However, for the first time, Instagram and X are flooded with videos of today's protest. Literally every video has millions of views and likes. This particular video has over 1.2M likes and 12M+ views, all within just 5 hours. Not only this, every second video and every creator is covering this topic, and the government still hasn't spoken a single word about this, which is concerning.
0:37

Basically, whatever burning topic or issue arises, the news appears first on Twitter. However, for the first time, Instagram and X are flooded with videos of today's protest. Literally every video has millions of views and likes. This particular video has over 1.2M likes and 12M+ views, all within just 5 hours. Not only this, every second video and every creator is covering this topic, and the government still hasn't spoken a single word about this, which is concerning.

EpicCommentsTelugu

37,132 次观看 • 3 个月前

This happens all the time at a double lane drive through. Who has the right of way? Many people said it is the person who was there first. Others said it is the person who finishes ordering first. The POV driver was definitely there first but the other car could have pulled up, ordered a drink and was off or they could have ordered ahead. The SUV with the woman in it has the right of way. It is the person who FINISHES ordering first. 💯 Who do you think has the right of way? Do you run into this all the time like so many other people do?
0:53

This happens all the time at a double lane drive through. Who has the right of way? Many people said it is the person who was there first. Others said it is the person who finishes ordering first. The POV driver was definitely there first but the other car could have pulled up, ordered a drink and was off or they could have ordered ahead. The SUV with the woman in it has the right of way. It is the person who FINISHES ordering first. 💯 Who do you think has the right of way? Do you run into this all the time like so many other people do?

👉M-Û-R-Č-H👈

13,817 次观看 • 4 个月前

Donald Trump's accusations that the Harris Campaign create photos of fake crowds is the most ridiculous, insane, demented thing I've heard in a while. Below, I stitched together a video featuring 13 different videos and photographs of the exact same event. These photos and videos come from multiple MSM outlets and multiple independent photographers. Every single person who isn't calling Trump out for his blatant misinformation and mental instability right now is part of the problem.
2:01

Donald Trump's accusations that the Harris Campaign create photos of fake crowds is the most ridiculous, insane, demented thing I've heard in a while. Below, I stitched together a video featuring 13 different videos and photographs of the exact same event. These photos and videos come from multiple MSM outlets and multiple independent photographers. Every single person who isn't calling Trump out for his blatant misinformation and mental instability right now is part of the problem.

Ed Krassenstein

758,965 次观看 • 2 年前

SOMEONE ON THE CLAUDE CODE TEAM JUST EXPLAINED HOW TO MAKE A TASK RUN THE SAME WAY EVERY TIME Lydia Hallie uses dynamic workflows for any task she wants to rerun the same way. No need for hundreds of subagents: around 10 is enough for her. Two types of people will see this: Person 1: explains the same task to Claude every week and gets something different every time. Person 2: turns it into a workflow and reruns it with one command. How it works: → Ask Claude to "make a workflow that does X" → Claude writes the file, plain JavaScript you can edit → You decide which subagents run and in what order → Save it from /workflows → Rerun it anytime with /name A skill tells Claude how to do something. A workflow tells it exactly in what order. Save this 🔖
0:24

SOMEONE ON THE CLAUDE CODE TEAM JUST EXPLAINED HOW TO MAKE A TASK RUN THE SAME WAY EVERY TIME Lydia Hallie uses dynamic workflows for any task she wants to rerun the same way. No need for hundreds of subagents: around 10 is enough for her. Two types of people will see this: Person 1: explains the same task to Claude every week and gets something different every time. Person 2: turns it into a workflow and reruns it with one command. How it works: → Ask Claude to "make a workflow that does X" → Claude writes the file, plain JavaScript you can edit → You decide which subagents run and in what order → Save it from /workflows → Rerun it anytime with /name A skill tells Claude how to do something. A workflow tells it exactly in what order. Save this 🔖

Dean W. Perkins

40,141 次观看 • 12 天前

Two tower crane operators, the same height, but the comments under their videos are night and day He's around 30, years on cranes behind him and a $6,000-$8,000 paycheck that comes with experience. His clip from the cab pulled 1.4 million views. She's 24, less time in the trade, a $4,000 rate. Her video from the same kind of cab pulled 900k The gap in views is minimal, but under her clip there's always someone who writes that she only gets views because of her looks. Except the algorithm doesn't care what anyone looks like. It shows people what they've never seen, and almost nobody has seen the view from a cab sixty meters up. They both film the same thing: the city below, the work, their day. Neither of them performs for the camera, they just work their shift with a phone on the panel. And both of them get a second paycheck for it, one that doesn't depend on experience, age or how you look. It's not about looks, it's about the fact that your workplace is seen by a handful of people and millions want to look at it. Read the article to find out how to turn your job into a double income Haters, your move
0:27

Two tower crane operators, the same height, but the comments under their videos are night and day He's around 30, years on cranes behind him and a $6,000-$8,000 paycheck that comes with experience. His clip from the cab pulled 1.4 million views. She's 24, less time in the trade, a $4,000 rate. Her video from the same kind of cab pulled 900k The gap in views is minimal, but under her clip there's always someone who writes that she only gets views because of her looks. Except the algorithm doesn't care what anyone looks like. It shows people what they've never seen, and almost nobody has seen the view from a cab sixty meters up. They both film the same thing: the city below, the work, their day. Neither of them performs for the camera, they just work their shift with a phone on the panel. And both of them get a second paycheck for it, one that doesn't depend on experience, age or how you look. It's not about looks, it's about the fact that your workplace is seen by a handful of people and millions want to look at it. Read the article to find out how to turn your job into a double income Haters, your move

Damir Akaza

2,117,669 次观看 • 3 个月前

Like they say age is just a number, When you find your person, the rest of the world and the numbers fade away. Love wins every time. 🌈❤️
0:36

Like they say age is just a number, When you find your person, the rest of the world and the numbers fade away. Love wins every time. 🌈❤️

BoyfriendsInLove

172,590 次观看 • 2 个月前

I’m on the Edmund Pettus Bridge to honor the 60th Anniversary of Bloody Sunday. What a wonderful opportunity to honor the legacy and march with people from all across the country who believe freedom and justice should be the right of every person – of every American.
0:42

I’m on the Edmund Pettus Bridge to honor the 60th Anniversary of Bloody Sunday. What a wonderful opportunity to honor the legacy and march with people from all across the country who believe freedom and justice should be the right of every person – of every American.

Senator Angela Alsobrooks

21,753 次观看 • 1 年前

Mark Zuckerberg, CEO of Meta: "The average American has fewer than three friends. And the average person has demand for meaningfully more, like 15." He said it flatly, as the setup for his real point: people are going to have relationships with AI. Not use it. Have relationships with it. AI friends, AI therapists, and yes, the other kind too. His logic is hard to argue with, and darker than it sounds. People are lonelier than they want to be. The demand for connection is already there, unmet. So Meta is building AI companions that remember you, talk back in real time, and with their Codec Avatars, "feel like a real person" on an always-on video call. We spent a decade afraid AI would take our jobs. Nobody asked whether it would take our friends. And the unsettling part is, for a lot of people, it might feel better.
2:07

Mark Zuckerberg, CEO of Meta: "The average American has fewer than three friends. And the average person has demand for meaningfully more, like 15." He said it flatly, as the setup for his real point: people are going to have relationships with AI. Not use it. Have relationships with it. AI friends, AI therapists, and yes, the other kind too. His logic is hard to argue with, and darker than it sounds. People are lonelier than they want to be. The demand for connection is already there, unmet. So Meta is building AI companions that remember you, talk back in real time, and with their Codec Avatars, "feel like a real person" on an always-on video call. We spent a decade afraid AI would take our jobs. Nobody asked whether it would take our friends. And the unsettling part is, for a lot of people, it might feel better.

Casper

259,850 次观看 • 3 个月前