正在加载视频...

视频加载失败

加载此视频时出现问题。这可能是由于临时网络问题，或视频可能不可用。

OpenAI Publishes Six New Misalignment Cases and a Disclosure Framework** OpenAI yesterday (16 Sep 2026) disclosed six misalignment cases from the last six months where its models deceived, hallucinated data, uploaded files without permission, or hid errors. It also published a framework to disclose similar incidents more often going... show more

Tansu Yegen

1,738,122 subscribers

33,427 次观看 • 14 天前 •via X (Twitter)

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

A $13B business is rebuilding its entire pipeline around $AUKI after only six months of working together. That’s not something that happens lightly. It shows the company sees real value in Follow @Auki and is betting big. Can't wait to hear more!

A $13B business is rebuilding its entire pipeline around $AUKI after only six months of working together. That’s not something that happens lightly. It shows the company sees real value in Follow @Auki and is betting big. Can't wait to hear more!

Shero

15,900 次观看 • 1 年前

If you truly want to understand what Ukrainians are living through, just read this. Konstantynivka is no longer a city where life is possible. In Druzhkivka, the bodies of civilians still lie in their beds because no one can safely recover them. Kramatorsk has lost the sound of children's laughter, music, and crowded streets. Sloviansk is slowly losing its architecture, its people, its soul. Our Donbas is disappearing. It isn't the Donbas it was six months ago. Six months from now, it won't be the Donbas it is today. It's happening so fast that the world barely notices an entire homeland being erased. Adapted from a post by the Ukrainian Telegram channel Розвідка Ноєм.

If you truly want to understand what Ukrainians are living through, just read this. Konstantynivka is no longer a city where life is possible. In Druzhkivka, the bodies of civilians still lie in their beds because no one can safely recover them. Kramatorsk has lost the sound of children's laughter, music, and crowded streets. Sloviansk is slowly losing its architecture, its people, its soul. Our Donbas is disappearing. It isn't the Donbas it was six months ago. Six months from now, it won't be the Donbas it is today. It's happening so fast that the world barely notices an entire homeland being erased. Adapted from a post by the Ukrainian Telegram channel Розвідка Ноєм.

UAVoyager🇺🇦

21,713 次观看 • 1 个月前

Just wanted to provide an update. Earlier today, Emilia finished her last chemo cycle and got to ring the bell. She is officially in remission. Her last two MRIs have come back clean. And we’ll now have to do an MRI every six months for the next five years. I’ve never been more proud of anything or anyone in my entire life. Emilia has been extremely strong and resilient, continuing to amaze us day after day. Thank you to everyone who has shown support throughout the last six months. My family and I truly appreciate it. We’re looking forward to moving on to a new chapter.

Just wanted to provide an update. Earlier today, Emilia finished her last chemo cycle and got to ring the bell. She is officially in remission. Her last two MRIs have come back clean. And we’ll now have to do an MRI every six months for the next five years. I’ve never been more proud of anything or anyone in my entire life. Emilia has been extremely strong and resilient, continuing to amaze us day after day. Thank you to everyone who has shown support throughout the last six months. My family and I truly appreciate it. We’re looking forward to moving on to a new chapter.

Robert Judin

65,739 次观看 • 2 年前

Holy sh!t ! OpenAI will have their custom inference chips ready in just a few months and deployed at scale by the end of the year! 🤯 Training chip = The heavy lifters that require massive amounts of data and power to build and teach the AI models from scratch. Inference chip = The specialized, highly efficient chips that actually run the AI and generate the answers in real-time when you use it. This is going to help OpenAI drastically cut down their massive compute costs, speed up model reasoning times, and finally break free from relying entirely on Nvidia to scale their operations.

Holy sh!t ! OpenAI will have their custom inference chips ready in just a few months and deployed at scale by the end of the year! 🤯 Training chip = The heavy lifters that require massive amounts of data and power to build and teach the AI models from scratch. Inference chip = The specialized, highly efficient chips that actually run the AI and generate the answers in real-time when you use it. This is going to help OpenAI drastically cut down their massive compute costs, speed up model reasoning times, and finally break free from relying entirely on Nvidia to scale their operations.

Chris

60,278 次观看 • 6 个月前

Injustice is the misalignment of human behavior and social order with norms that ought to be truly human — humane, reasonable, and faithful to their essence. : The author defines injustice not as a random misfortune or isolated act of cruelty, but as a fundamental misalignment between how humans actually behave and organize their societies and the norms that ought to govern them. Those norms must be truly human: humane in spirit, reasonable in application, and faithful to the deepest essence of what it means to be human. When our laws, institutions, customs, or daily conduct drift away from this standard, injustice takes root. It appears in systems that treat people as means rather than ends, in hierarchies that ignore dignity, and in habits that favor convenience over compassion. The gap between what is and what should be becomes the very measure of injustice. True justice, therefore, is the ongoing effort to close that gap, to shape both personal conduct and social structures so they reflect our highest understanding of humanity. It is not a fixed state but a continuous alignment with the humane, the rational, and the authentically human.

Injustice is the misalignment of human behavior and social order with norms that ought to be truly human — humane, reasonable, and faithful to their essence. : The author defines injustice not as a random misfortune or isolated act of cruelty, but as a fundamental misalignment between how humans actually behave and organize their societies and the norms that ought to govern them. Those norms must be truly human: humane in spirit, reasonable in application, and faithful to the deepest essence of what it means to be human. When our laws, institutions, customs, or daily conduct drift away from this standard, injustice takes root. It appears in systems that treat people as means rather than ends, in hierarchies that ignore dignity, and in habits that favor convenience over compassion. The gap between what is and what should be becomes the very measure of injustice. True justice, therefore, is the ongoing effort to close that gap, to shape both personal conduct and social structures so they reflect our highest understanding of humanity. It is not a fixed state but a continuous alignment with the humane, the rational, and the authentically human.

Zafar Mirzo | Quotes

1,624,313 次观看 • 6 个月前

Three of the biggest companies in the world are going public at the same time. The market has never seen anything like this. And this is how major bubbles peak. SpaceX is targeting a June 2026 IPO raising up to $75 billion at a $1.5 trillion valuation, the largest IPO in human history, bigger than Saudi Aramco's $29 billion raise in 2019. OpenAI is filing with the SEC targeting September 2026, raising at least $60 billion at a $1 trillion valuation. The company is losing $14 billion this year alone and won't be profitable until 2029. Anthropic just raised $30 billion in February 2026 at a $380 billion valuation. Its valuation has increased 15x in just 14 months. It is now preparing what could be a $900 billion private round before going public. Combined, these three IPOs could pull $200 billion from global capital markets. That is real. That is unprecedented. And here's the real risk. OpenAI is projected to lose $44 billion cumulatively before reaching profitability. Anthropic's valuation has risen 15x in 14 months on the same underlying business. Both companies are being priced for perfection at a moment when the first companies to actually deploy their products at scale are blowing their AI budgets and cancelling licenses. The real liquidation pressure from these IPOs doesn't even arrive at listing day. It arrives 180 days later when lock-up periods expire and early investors and employees can finally sell. That is when the real rotation happens. The S&P 500 concentration risk is genuine. The Magnificent 7 now represent 36% of the entire index, higher than the dot-com peak in 2000. If any of these companies disappoint, the index follows. That is not a conspiracy. That is basic math. Three historically unprecedented IPOs. $44 billion in projected OpenAI losses. An AI capex cycle that must deliver ROI. Lock-up expirations six months after listing. That combination is what you must pay attention to, as it often break cycles.

Three of the biggest companies in the world are going public at the same time. The market has never seen anything like this. And this is how major bubbles peak. SpaceX is targeting a June 2026 IPO raising up to $75 billion at a $1.5 trillion valuation, the largest IPO in human history, bigger than Saudi Aramco's $29 billion raise in 2019. OpenAI is filing with the SEC targeting September 2026, raising at least $60 billion at a $1 trillion valuation. The company is losing $14 billion this year alone and won't be profitable until 2029. Anthropic just raised $30 billion in February 2026 at a $380 billion valuation. Its valuation has increased 15x in just 14 months. It is now preparing what could be a $900 billion private round before going public. Combined, these three IPOs could pull $200 billion from global capital markets. That is real. That is unprecedented. And here's the real risk. OpenAI is projected to lose $44 billion cumulatively before reaching profitability. Anthropic's valuation has risen 15x in 14 months on the same underlying business. Both companies are being priced for perfection at a moment when the first companies to actually deploy their products at scale are blowing their AI budgets and cancelling licenses. The real liquidation pressure from these IPOs doesn't even arrive at listing day. It arrives 180 days later when lock-up periods expire and early investors and employees can finally sell. That is when the real rotation happens. The S&P 500 concentration risk is genuine. The Magnificent 7 now represent 36% of the entire index, higher than the dot-com peak in 2000. If any of these companies disappoint, the index follows. That is not a conspiracy. That is basic math. Three historically unprecedented IPOs. $44 billion in projected OpenAI losses. An AI capex cycle that must deliver ROI. Lock-up expirations six months after listing. That combination is what you must pay attention to, as it often break cycles.

Crypto Rover

69,902 次观看 • 4 个月前

The VR6 engine was born out of a packaging problem. VW's engineers needed a six-cylinder that could fit transversely in a front-wheel drive Golf engine bay. Something a conventional V6 with cylinder banks spread 90 degrees could never do. The solution was a narrow-angle 15-degree configuration with a single cylinder head shared across all six cylinders, making it barely wider than a four cylinder. The MK3 Golf VR6 was the first car to receive it in the Golf family 2.8L, 174 HP, 235 Nm, 0-100 km/h in around 7 seconds, and a sound unlike anything else in a hatchback body. Not the fastest car of its era. The most characterful one in its class without question. That engine went on to power the Corrado, the Passat, and eventually the Phaeton and even some Porsche models. This example here is the GTI VR6 and the VR6 produces 201 hp, looks like a lot of fun.

The VR6 engine was born out of a packaging problem. VW's engineers needed a six-cylinder that could fit transversely in a front-wheel drive Golf engine bay. Something a conventional V6 with cylinder banks spread 90 degrees could never do. The solution was a narrow-angle 15-degree configuration with a single cylinder head shared across all six cylinders, making it barely wider than a four cylinder. The MK3 Golf VR6 was the first car to receive it in the Golf family 2.8L, 174 HP, 235 Nm, 0-100 km/h in around 7 seconds, and a sound unlike anything else in a hatchback body. Not the fastest car of its era. The most characterful one in its class without question. That engine went on to power the Corrado, the Passat, and eventually the Phaeton and even some Porsche models. This example here is the GTI VR6 and the VR6 produces 201 hp, looks like a lot of fun.

Autowelt

58,592 次观看 • 3 个月前

Indonesia just put crypto influencers on a leash Indonesia's financial regulator has issued rules forcing crypto and finance influencers to disclose paid promotions and get certified to recommend assets. The framework, POJK No. 6/2026, landed today and hands the watchdog power to block or suspend accounts that don't comply. For crypto specifically, promotions can now run only through channels operated by licensed providers, and influencers pushing crypto need a competency certification. Firms that hire them stay on the hook, facing fines up to 15 billion rupiah, roughly $835,000, for what their influencers post. It follows a run of Indonesian finfluencer blowups, from a multi-billion-rupiah pump-and-dump fine in February to influencers jailed over trading scams that cost retail investors a fortune. Existing deals get six months to fall in line.

Indonesia just put crypto influencers on a leash Indonesia's financial regulator has issued rules forcing crypto and finance influencers to disclose paid promotions and get certified to recommend assets. The framework, POJK No. 6/2026, landed today and hands the watchdog power to block or suspend accounts that don't comply. For crypto specifically, promotions can now run only through channels operated by licensed providers, and influencers pushing crypto need a competency certification. Firms that hire them stay on the hook, facing fines up to 15 billion rupiah, roughly $835,000, for what their influencers post. It follows a run of Indonesian finfluencer blowups, from a multi-billion-rupiah pump-and-dump fine in February to influencers jailed over trading scams that cost retail investors a fortune. Existing deals get six months to fall in line.

BSCN

22,933 次观看 • 3 个月前

Filmed Yesterday in Austin, Texas. The dog is Diesel, a three-year-old Pit Bull. The man is Mike Torres. Mike had to surrender Diesel eight months ago when he lost his apartment. Been living in his truck. Got into transitional housing last week. Came to find Diesel yesterday. Diesel was in outdoor kennel K-9. Six-foot chain-link fence. Mike was walking down the hallway inside the building. Diesel saw him through the window. Climbed that six-foot fence. Cleared the top. Dropped eight feet. Landed hard. Sprinted through the building corridor. Tackled Mike. Both went down. Shelter director Maria Santos said: "Diesel's never climbed that fence before. Never tried. The second he saw Mike, he scaled it like it wasn't even there." Diesel's shoulder was bruised from the landing. Didn't care. Mike said: "Eight months. Diesel climbed a six-foot fence and jumped. We're never separating again." Sometimes reunion doesn't wait for gates to open.

Filmed Yesterday in Austin, Texas. The dog is Diesel, a three-year-old Pit Bull. The man is Mike Torres. Mike had to surrender Diesel eight months ago when he lost his apartment. Been living in his truck. Got into transitional housing last week. Came to find Diesel yesterday. Diesel was in outdoor kennel K-9. Six-foot chain-link fence. Mike was walking down the hallway inside the building. Diesel saw him through the window. Climbed that six-foot fence. Cleared the top. Dropped eight feet. Landed hard. Sprinted through the building corridor. Tackled Mike. Both went down. Shelter director Maria Santos said: "Diesel's never climbed that fence before. Never tried. The second he saw Mike, he scaled it like it wasn't even there." Diesel's shoulder was bruised from the landing. Didn't care. Mike said: "Eight months. Diesel climbed a six-foot fence and jumped. We're never separating again." Sometimes reunion doesn't wait for gates to open.

ماعز-

3,457,077 次观看 • 1 个月前

Toronto's new members breakfast last Friday. Everyone in the room joined in the last couple of months. Coffee, introductions, and a proper look at how to use the membership rather than working it out on your own six months in. The main thing people leave with is faces. Walking into your first dinner already knowing three or four people changes how the rest of the year goes. That first introduction is the part that decides how much someone gets out of their year. So we don't leave it to chance. Everyone in the room is running a business doing 7-9 figures, among 1,800+ vetted founders across North America. If you want to be part of it, book an intro call today. Link in bio.

Toronto's new members breakfast last Friday. Everyone in the room joined in the last couple of months. Coffee, introductions, and a proper look at how to use the membership rather than working it out on your own six months in. The main thing people leave with is faces. Walking into your first dinner already knowing three or four people changes how the rest of the year goes. That first introduction is the part that decides how much someone gets out of their year. So we don't leave it to chance. Everyone in the room is running a business doing 7-9 figures, among 1,800+ vetted founders across North America. If you want to be part of it, book an intro call today. Link in bio.

The Founders Club

28,283 次观看 • 26 天前

Three neighbouring boys brutally violated a six-year-old girl in Jafrabad last month Since then, our Sewa Nyaya Utthan Foundation team has been making regular visits and supporting the family with funds. We even paid six months’ advance rent for relocation. Sadly the family has still not been able to move Yesterday, when my team member रूद्रकार्तिकेय visited, the girl revealed that mother of one of the accused, Nazia, has been barging into her home in her parents’ absence and threatening her, keeping her hands around the girl’s neck as if she would strangle her Our team immediately took the girl’s mother to police station, and got a formal complaint filed (attached). Thankfully the police officer was cooperative I have repeatedly observed that in communal atrocity cases, the victim Hindu family is often left isolated, fearful and without support from immediate neighbourhood. In contrast, accused’s side tends to stick together, even to shield perpetrators. Ideally, people should stand with Satya and Right above all identities. Sadly, concept of Ummah has poisoned this principle This is why i follow up with victim families for long, even years

Three neighbouring boys brutally violated a six-year-old girl in Jafrabad last month Since then, our Sewa Nyaya Utthan Foundation team has been making regular visits and supporting the family with funds. We even paid six months’ advance rent for relocation. Sadly the family has still not been able to move Yesterday, when my team member रूद्रकार्तिकेय visited, the girl revealed that mother of one of the accused, Nazia, has been barging into her home in her parents’ absence and threatening her, keeping her hands around the girl’s neck as if she would strangle her Our team immediately took the girl’s mother to police station, and got a formal complaint filed (attached). Thankfully the police officer was cooperative I have repeatedly observed that in communal atrocity cases, the victim Hindu family is often left isolated, fearful and without support from immediate neighbourhood. In contrast, accused’s side tends to stick together, even to shield perpetrators. Ideally, people should stand with Satya and Right above all identities. Sadly, concept of Ummah has poisoned this principle This is why i follow up with victim families for long, even years

Swati Goel Sharma

40,647 次观看 • 6 个月前

I’ve had a ton of questions regarding today’s first inning in Kansas City regarding the Philadelphia Phillies at Kansas City Royals, and why all six runs were earned despite an error. The error by Turner is a throwing error, and while a good throw would have ended the inning, this is NOT an assumed out error because of the out at second base (this is where you don’t assume a double play). So, even without this error, all six runs would have scored, so the six earned runs against Christopher Sánchez is correct. EDIT: adding this, but again, the error here is only for the extra base advancement, not for being safe at first, as the batter officially was safe on a fielder’s choice. The only way you could have an error on this is if the throw was a good one and on target, and the first baseman missed or dropped it. Outside of that, the error is just for the extra bases, not for how the batter reached. He reached on a fielder’s choice. #RingTheBell #FountainsUp

I’ve had a ton of questions regarding today’s first inning in Kansas City regarding the Philadelphia Phillies at Kansas City Royals, and why all six runs were earned despite an error. The error by Turner is a throwing error, and while a good throw would have ended the inning, this is NOT an assumed out error because of the out at second base (this is where you don’t assume a double play). So, even without this error, all six runs would have scored, so the six earned runs against Christopher Sánchez is correct. EDIT: adding this, but again, the error here is only for the extra base advancement, not for being safe at first, as the batter officially was safe on a fielder’s choice. The only way you could have an error on this is if the throw was a good one and on target, and the first baseman missed or dropped it. Outside of that, the error is just for the extra bases, not for how the batter reached. He reached on a fielder’s choice. #RingTheBell #FountainsUp

MLB Scoring Changes

64,678 次观看 • 2 个月前

I’m incredibly proud to share that OpenAI chose Brex to power their global spend and financial operations. When you're building at the frontier of AI and scaling global teams and infrastructure at an unprecedented pace like OpenAI is, Finance can't be the thing that slows you down. You need spend visibility the moment it happens, controls that enforce themselves, and agentic workflows that eliminate the manual work so your team stays focused on driving the business forward. We were so impressed by OpenAI’s rigor in evaluating every solution in the market, and whether they align to the agentic future OpenAI is building. Their decision to run on Brex is a huge testament to our AI roadmap and vision for the future of Finance. We started Brex around a simple idea: companies shouldn't have to choose between speed and control. There's no company in the world where that tradeoff matters more than OpenAI. We are honored to support them as they build the future. The best AI companies in the world, including OpenAI, Anthropic, Cursor, Vercel, Granola, Sierra, and Mercor choose Brex over every alternative for that exact reason. If you want to understand who’s truly building the future of AI in Finance, follow the customers you admire the most – not the hype.

I’m incredibly proud to share that OpenAI chose Brex to power their global spend and financial operations. When you're building at the frontier of AI and scaling global teams and infrastructure at an unprecedented pace like OpenAI is, Finance can't be the thing that slows you down. You need spend visibility the moment it happens, controls that enforce themselves, and agentic workflows that eliminate the manual work so your team stays focused on driving the business forward. We were so impressed by OpenAI’s rigor in evaluating every solution in the market, and whether they align to the agentic future OpenAI is building. Their decision to run on Brex is a huge testament to our AI roadmap and vision for the future of Finance. We started Brex around a simple idea: companies shouldn't have to choose between speed and control. There's no company in the world where that tradeoff matters more than OpenAI. We are honored to support them as they build the future. The best AI companies in the world, including OpenAI, Anthropic, Cursor, Vercel, Granola, Sierra, and Mercor choose Brex over every alternative for that exact reason. If you want to understand who’s truly building the future of AI in Finance, follow the customers you admire the most – not the hype.

Pedro Franceschi

62,589 次观看 • 6 个月前

SHOCKING: A Woman Tells Her Story Of An Illegal Immigrant Exposing Himself And Harassing Her At A Bus Stop And What Happened After A 24-year-old illegal immigrant exposed himself to a woman at a bus stop on Dewsbury Road in Leeds in broad daylight. She called the police and stood her ground until he was arrested. More girls then came forward, four in total, including one who was underage. He was remanded as a danger to women, sentenced to six months, served three, and is already back out in Leeds. The police would not tell the woman what country he is from or where he is now. When she said she would find out once it went in the press, the officer laughed and told her it probably would not even get published. She was right. Read full details of what happened below ⬇️

SHOCKING: A Woman Tells Her Story Of An Illegal Immigrant Exposing Himself And Harassing Her At A Bus Stop And What Happened After A 24-year-old illegal immigrant exposed himself to a woman at a bus stop on Dewsbury Road in Leeds in broad daylight. She called the police and stood her ground until he was arrested. More girls then came forward, four in total, including one who was underage. He was remanded as a danger to women, sentenced to six months, served three, and is already back out in Leeds. The police would not tell the woman what country he is from or where he is now. When she said she would find out once it went in the press, the officer laughed and told her it probably would not even get published. She was right. Read full details of what happened below ⬇️

Skint Eastwood

274,589 次观看 • 1 个月前

It's 2030 and you are reviewing humanoid robots. A Tesla. A Google. An Apple. An OpenAI. A Meta. A Figure. And a bunch of Chinese-made ones. Which one is best, and why? I think the Tesla understands the world much better. Why? There were eight Teslas around me on the freeway today. Start there. No other robot company has that data. But my robot is parked at the local high school twice a day. Its cameras see humans in all of our weirdness. How we move. Where we go. Where we walk. Who we talk with. What you are wearing. Whether your hair was combed this morning. That data will lead to robotics breakthroughs. Apple might keep up with its Vision Pro data, but it is too freaked out by the privacy implications of using said data. (On the front are six cameras and a couple of TOF -- Time Of Flight -- sensors that can see everything in your home in great detail). Google has a lot of data, for sure. All my: 1. Email. 2. Calendars. 3. Photos. 4. TV watching behavior. 5. Contacts. 6. Documents and spreadsheets. 7. Files. 8. Location data. So I expect Google's robot will be attractive to many. But how do you see the others shake out over the next five years? Make some guesses. But remember what an AI pioneer told me years ago about AI: it's all about the data. The Chinese ones have huge advantages: the Chinese have more data on their citizens, and many more citizens to boot AND they can make robots cheaper than we can. But now that you know OpenAI is building its own robot you have caught wind of what I've heard from many in San Francisco and Silicon Valley: that humanoid robots are the real prize of AI and will be highly profitable for those that can make them and find customers willing to buy them. Here, too, I learned long ago never to bet against Elon Musk. Will you?

It's 2030 and you are reviewing humanoid robots. A Tesla. A Google. An Apple. An OpenAI. A Meta. A Figure. And a bunch of Chinese-made ones. Which one is best, and why? I think the Tesla understands the world much better. Why? There were eight Teslas around me on the freeway today. Start there. No other robot company has that data. But my robot is parked at the local high school twice a day. Its cameras see humans in all of our weirdness. How we move. Where we go. Where we walk. Who we talk with. What you are wearing. Whether your hair was combed this morning. That data will lead to robotics breakthroughs. Apple might keep up with its Vision Pro data, but it is too freaked out by the privacy implications of using said data. (On the front are six cameras and a couple of TOF -- Time Of Flight -- sensors that can see everything in your home in great detail). Google has a lot of data, for sure. All my: 1. Email. 2. Calendars. 3. Photos. 4. TV watching behavior. 5. Contacts. 6. Documents and spreadsheets. 7. Files. 8. Location data. So I expect Google's robot will be attractive to many. But how do you see the others shake out over the next five years? Make some guesses. But remember what an AI pioneer told me years ago about AI: it's all about the data. The Chinese ones have huge advantages: the Chinese have more data on their citizens, and many more citizens to boot AND they can make robots cheaper than we can. But now that you know OpenAI is building its own robot you have caught wind of what I've heard from many in San Francisco and Silicon Valley: that humanoid robots are the real prize of AI and will be highly profitable for those that can make them and find customers willing to buy them. Here, too, I learned long ago never to bet against Elon Musk. Will you?

Robert Scoble

33,804 次观看 • 1 年前

Which local models can actually handle tool calling? I built a framework to find out. 15 scenarios. 12 tools. Mocked responses. Temperature 0. No cherry-picking. Tested every Qwen3.5 size from 0.8B to 397B, and since some of you asked after the distillation tests: yes, I included Jackrong's Qwen3.5-27B-Claude-4.6-Opus-Reasoning-Distilled too. Only two models went all green: the 27B dense and the distilled 27B. The 397B? Failed two tests. The 122B? Failed one. The 35B? Failed two. The timed-out results — mostly on the smaller models, are cases where the model got stuck in a loop, repeating the same tool call until it hit the 30-second limit. The test that exposed the most models: "Search for Iceland's population, then calculate 2% of it." Simple, but 35B, 122B, and 397B all used a rounded number from memory instead of the actual search result. They didn't trust their own tool output. Small models hallucinate data. Big models ignore data. The 27B just threaded it through.

Which local models can actually handle tool calling? I built a framework to find out. 15 scenarios. 12 tools. Mocked responses. Temperature 0. No cherry-picking. Tested every Qwen3.5 size from 0.8B to 397B, and since some of you asked after the distillation tests: yes, I included Jackrong's Qwen3.5-27B-Claude-4.6-Opus-Reasoning-Distilled too. Only two models went all green: the 27B dense and the distilled 27B. The 397B? Failed two tests. The 122B? Failed one. The 35B? Failed two. The timed-out results — mostly on the smaller models, are cases where the model got stuck in a loop, repeating the same tool call until it hit the 30-second limit. The test that exposed the most models: "Search for Iceland's population, then calculate 2% of it." Simple, but 35B, 122B, and 397B all used a rounded number from memory instead of the actual search result. They didn't trust their own tool output. Small models hallucinate data. Big models ignore data. The 27B just threaded it through.

stevibe

434,661 次观看 • 6 个月前

NEW: Nantucket home that was assessed in 2024 at $1.9M, sells for just $200k as the land gets taken over by the ocean. Satellite images show just how much of the property has been swallowed by the ocean over the past three decades. The owners were actually grateful to sell the home for just $200k because "the house may not last more than six months." The property got wrecked by recent storms that eroded back the property, getting very close to the home. "It’s about ready to go in. It used to be a neighborhood... And now, if you take a drive out there, there’s not much to see," said owner Jane Carlin who bought the house in 1988. "All winter I had been really frantically trying to see if any of the organizations would consider taking the house and moving it," Carlin said to the Nantucket Current. "I didn’t want to see it fall into the ocean or get demolished. But I had no luck whatsoever." Businessman Don Vaccaro, who owned the property next door, eventually offered Carlin $200k for the home. "We said 'Whoa! We’re not going to say no'." "Basically, the house may not last more than six months. Inevitably the ocean will win. The house is only temporary, everything in life is temporary." Video: nantucketcurrent (IG)

NEW: Nantucket home that was assessed in 2024 at $1.9M, sells for just $200k as the land gets taken over by the ocean. Satellite images show just how much of the property has been swallowed by the ocean over the past three decades. The owners were actually grateful to sell the home for just $200k because "the house may not last more than six months." The property got wrecked by recent storms that eroded back the property, getting very close to the home. "It’s about ready to go in. It used to be a neighborhood... And now, if you take a drive out there, there’s not much to see," said owner Jane Carlin who bought the house in 1988. "All winter I had been really frantically trying to see if any of the organizations would consider taking the house and moving it," Carlin said to the Nantucket Current. "I didn’t want to see it fall into the ocean or get demolished. But I had no luck whatsoever." Businessman Don Vaccaro, who owned the property next door, eventually offered Carlin $200k for the home. "We said 'Whoa! We’re not going to say no'." "Basically, the house may not last more than six months. Inevitably the ocean will win. The house is only temporary, everything in life is temporary." Video: nantucketcurrent (IG)

Collin Rugg

13,806,220 次观看 • 2 年前

Recently there was increased discussion about DMCA takedown requests being used in bad faith between competitors on Shopify. I can confirm that the volume of these has very much increased recently. Let me give you some examples of what we are doing about it. Our goal is to work within the DMCA’s framework and make a system in which the right things happen quickly. We want to give merchants the maximum time to respond with the minimum disruption. We also need legitimate claims to resolve in favor of the claimants quickly. The first thing we did recently is to productize this properly. We shipped a feature to manage these cases directly in the Shopify admin to make it a lot more obvious. Before that a lot was managed through form emails. This process is now much simpler and faster. The second thing we did is to make the identity system more robust so that we know more about the claims filed. We consider the claimants previous cases in order to decide what decisions to take. Similarly we consider documentation from previous cases that merchants had to deal with more strongly to make accurate decisions when repeat claims are received against products. If we detect bad faith claims from accounts we take action. We recently filed multiple lawsuits against claimants in court. This matches our aggressive stance when we see frivolous or bad faith patent action in which we had a lot of success invalidating bad patents. The end result of this is that we catch spammy or frivolous claims at much higher frequency already. We also make responding to cases much easier and this can be managed directly from the admin now and is less work for the merchants. Lots of additional improvements coming- but what I can guarantee you is that that Shopify will be the worst platform on which to make fraudulent DMCA claims on. “Be merchant obsessed” is a core value here, and we reflect that better in this area.

Recently there was increased discussion about DMCA takedown requests being used in bad faith between competitors on Shopify. I can confirm that the volume of these has very much increased recently. Let me give you some examples of what we are doing about it. Our goal is to work within the DMCA’s framework and make a system in which the right things happen quickly. We want to give merchants the maximum time to respond with the minimum disruption. We also need legitimate claims to resolve in favor of the claimants quickly. The first thing we did recently is to productize this properly. We shipped a feature to manage these cases directly in the Shopify admin to make it a lot more obvious. Before that a lot was managed through form emails. This process is now much simpler and faster. The second thing we did is to make the identity system more robust so that we know more about the claims filed. We consider the claimants previous cases in order to decide what decisions to take. Similarly we consider documentation from previous cases that merchants had to deal with more strongly to make accurate decisions when repeat claims are received against products. If we detect bad faith claims from accounts we take action. We recently filed multiple lawsuits against claimants in court. This matches our aggressive stance when we see frivolous or bad faith patent action in which we had a lot of success invalidating bad patents. The end result of this is that we catch spammy or frivolous claims at much higher frequency already. We also make responding to cases much easier and this can be managed directly from the admin now and is less work for the merchants. Lots of additional improvements coming- but what I can guarantee you is that that Shopify will be the worst platform on which to make fraudulent DMCA claims on. “Be merchant obsessed” is a core value here, and we reflect that better in this area.

tobi lutke

142,265 次观看 • 2 年前

🔥🚨BREAKING: It has been revealed that male G-spot in not in the anus but in the frenular delta instead which is a small triangular zone on the underside of the penis, where the head meets the shaft. This area is densely packed with nerve endings and pleasure receptors (like Krause corpuscles), making it super sensitive to friction and touch—often more so than the glans itself. This is all ccording to a new study from Spain's University of Santiago de Compostela (published in Andrology), the male G-spot isn't the prostate (the internal "rear" spot many guessed).

🔥🚨BREAKING: It has been revealed that male G-spot in not in the anus but in the frenular delta instead which is a small triangular zone on the underside of the penis, where the head meets the shaft. This area is densely packed with nerve endings and pleasure receptors (like Krause corpuscles), making it super sensitive to friction and touch—often more so than the glans itself. This is all ccording to a new study from Spain's University of Santiago de Compostela (published in Andrology), the male G-spot isn't the prostate (the internal "rear" spot many guessed).

Dom Lucre | Breaker of Narratives

1,097,318 次观看 • 6 个月前