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。

Class conditioning doesn’t just add new capabilities to flow models. It can make the learning problem easier. When there are multiple plausible paths, the class label tells the model which one to take, reducing errors from averaging over conflicting trajectories.

Alec Helbling

10,807 subscribers

21,016 次观看 • 1 个月前 •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

The more AI can do, the more we need to ask what it should and shouldn’t do. OpenAI researcher Jason Wolfe joins host Andrew Mayne to explore the Model Spec, the public framework that defines how models are intended to behave. They break down how it works in practice, from the chain of command that resolves conflicting instructions to the way it evolves over time through real-world use, feedback, and new model capabilities.

The more AI can do, the more we need to ask what it should and shouldn’t do. OpenAI researcher Jason Wolfe joins host Andrew Mayne to explore the Model Spec, the public framework that defines how models are intended to behave. They break down how it works in practice, from the chain of command that resolves conflicting instructions to the way it evolves over time through real-world use, feedback, and new model capabilities.

OpenAI

225,402 次观看 • 3 个月前

It never was about reducing the debt. Trump and Elon Musk are firing working class Americans from their jobs to give tax cuts to billionaires — which is exactly what my song "New Religion" is about. These motherfuckers are plundering and pillaging America to line their pockets.

It never was about reducing the debt. Trump and Elon Musk are firing working class Americans from their jobs to give tax cuts to billionaires — which is exactly what my song "New Religion" is about. These motherfuckers are plundering and pillaging America to line their pockets.

Bill Madden

498,340 次观看 • 1 年前

Let’s strike the expression “ready to learn” from our collective vocabulary when it comes to young children. Learning begins in utero. Kids are born learning. There’s nothing to get ready for. Suggesting there is perpetuates the notion that learning begins in school. It doesn’t

Let’s strike the expression “ready to learn” from our collective vocabulary when it comes to young children. Learning begins in utero. Kids are born learning. There’s nothing to get ready for. Suggesting there is perpetuates the notion that learning begins in school. It doesn’t

Dan Wuori

96,418 次观看 • 3 年前

Anthropic engineer: "Fable is one of those models you’ll just remember - like Sonnet 3.5, Opus 4, Opus 4.5. We just removed 80% of Claude Code’s system prompt. A new class of models wants a smaller prompt." in 20-minute session, an Anthropic core engineer reveals the capabilities of Fable. Watch it now, then read how to use this frontier model in the article below.

Anthropic engineer: "Fable is one of those models you’ll just remember - like Sonnet 3.5, Opus 4, Opus 4.5. We just removed 80% of Claude Code’s system prompt. A new class of models wants a smaller prompt." in 20-minute session, an Anthropic core engineer reveals the capabilities of Fable. Watch it now, then read how to use this frontier model in the article below.

Movez

549,021 次观看 • 17 天前

Engram cofounder Jack Morris just raised $98M to build a new type of AI. He says models don't need to get smarter over time. Instead, they just need to know you better and better over time. Jack describes what he's building: "Our product is a new type of AI. We have a pretty different vision from a lot of the frontier labs, which are working on one model per lab, and trying to make that model smarter every month." "There's another way to think about it, which is that the model doesn't need to get smarter every month. It needs to know you better." "So we're working on a whole different stack, which is a way to train models that train themselves to know your world better and adjust to the things that you say." "So: new ways of training, new ways of running the models."

Engram cofounder Jack Morris just raised $98M to build a new type of AI. He says models don't need to get smarter over time. Instead, they just need to know you better and better over time. Jack describes what he's building: "Our product is a new type of AI. We have a pretty different vision from a lot of the frontier labs, which are working on one model per lab, and trying to make that model smarter every month." "There's another way to think about it, which is that the model doesn't need to get smarter every month. It needs to know you better." "So we're working on a whole different stack, which is a way to train models that train themselves to know your world better and adjust to the things that you say." "So: new ways of training, new ways of running the models."

TBPN

67,455 次观看 • 19 天前

We shouldn’t send a single penny to ANY other country while we are $39 TRILLION in debt and working class families are struggling to make it from one paycheck to the next.

We shouldn’t send a single penny to ANY other country while we are $39 TRILLION in debt and working class families are struggling to make it from one paycheck to the next.

Derrick Evans

22,347 次观看 • 1 个月前

"...and when (the optic) did die, they didn't have any backup irons on the pistol so there was no way for them to aim the pistol." -TheGunGuy There are ways to use the optic housing as an aiming reference. All you have to do is just take 1 class from someone who knows.

"...and when (the optic) did die, they didn't have any backup irons on the pistol so there was no way for them to aim the pistol." -TheGunGuy There are ways to use the optic housing as an aiming reference. All you have to do is just take 1 class from someone who knows.

B Mitori

46,120 次观看 • 3 个月前

Jeremy's ideas are always original My favorite from this one is "peak guy" and why the poaster is our new priestly class "I think the billionaire class has become subservient to the poasting class. It doesn't make sense that we spend so much time caring about what billionaires think about theology or health or topics unrelated to accruing a billion dollars. The reason we do is because these are the people that have ascended to the highest realm of piety in our value system. We've looked to them to provide us these answers and it has not been satisfying. Peak guy is I think we've had enough of the billionaire. The logical question is what is the next class, our next set of priests? I think it's the poaster. You can always look which class is subservient to the next to see who's next. I was at this thing with a bunch of billionaire investors and they were all fighting over who could sit next to tylercowen because he's the most interesting person there."

Jeremy's ideas are always original My favorite from this one is "peak guy" and why the poaster is our new priestly class "I think the billionaire class has become subservient to the poasting class. It doesn't make sense that we spend so much time caring about what billionaires think about theology or health or topics unrelated to accruing a billion dollars. The reason we do is because these are the people that have ascended to the highest realm of piety in our value system. We've looked to them to provide us these answers and it has not been satisfying. Peak guy is I think we've had enough of the billionaire. The logical question is what is the next class, our next set of priests? I think it's the poaster. You can always look which class is subservient to the next to see who's next. I was at this thing with a bunch of billionaire investors and they were all fighting over who could sit next to tylercowen because he's the most interesting person there."

Patrick OShaughnessy

225,569 次观看 • 11 天前

🚨 Introducing the 2025 Culture Class 🚨 This class is stepping heavy — featuring a bold new crew of creators and disruptors who move culture, not just follow it. From viral moments to real-world impact, they’ve been outside setting the tone all year. So of course we had to get them all ⁠under one roof and watch the magic unfold. Watch what happens in the house and as they head to LA to take over the #BETExperience and Culture’s biggest night the 2025 #BETAwards.⁠ Class is officially in session 📗

🚨 Introducing the 2025 Culture Class 🚨 This class is stepping heavy — featuring a bold new crew of creators and disruptors who move culture, not just follow it. From viral moments to real-world impact, they’ve been outside setting the tone all year. So of course we had to get them all ⁠under one roof and watch the magic unfold. Watch what happens in the house and as they head to LA to take over the #BETExperience and Culture’s biggest night the 2025 #BETAwards.⁠ Class is officially in session 📗

BET

14,845 次观看 • 1 年前

The new SpaceX-Cursor model is going to be insane. It is being trained from scratch, unlike previous Composer models, which were built off Kimi. This new model is Elon’s attempt to get back in the AI race, and he might rejoin as the frontrunner. Composer 2.5 is already a world-class model, only losing to models like GPT-5.5, Opus 4.8, and Fable 5. Grok Composer 3, with SpaceX’s massive compute and xAI’s team, could become the best AI model on the planet. Elon is back.

The new SpaceX-Cursor model is going to be insane. It is being trained from scratch, unlike previous Composer models, which were built off Kimi. This new model is Elon’s attempt to get back in the AI race, and he might rejoin as the frontrunner. Composer 2.5 is already a world-class model, only losing to models like GPT-5.5, Opus 4.8, and Fable 5. Grok Composer 3, with SpaceX’s massive compute and xAI’s team, could become the best AI model on the planet. Elon is back.

Ali Romman

119,837 次观看 • 1 个月前

Tired of guessing which AI model is best? 🤔 We've got you covered! 👀 Here's a sneak peek of our new model comparison feature in action! Run the same prompt across multiple models and see which one works best for YOU. Which models are you excited to test? #Inferium #AI #Inference #Compare

Tired of guessing which AI model is best? 🤔 We've got you covered! 👀 Here's a sneak peek of our new model comparison feature in action! Run the same prompt across multiple models and see which one works best for YOU. Which models are you excited to test? #Inferium #AI #Inference #Compare

Inferium

12,329 次观看 • 1 年前

🚨JUST IN: Stephen Miller just EXPOSED the DEMOCRATS’ PLOT… to turn America into “SOMALIA” 🚨 “It’s the Somalification of America — easier to rule over a divided, chaotic empire than a functioning, high-trust, middle-class society.” “That’s their model for America: to make the whole country into a version of Somalia. Everything they do tracks back to that goal.”

🚨JUST IN: Stephen Miller just EXPOSED the DEMOCRATS’ PLOT… to turn America into “SOMALIA” 🚨 “It’s the Somalification of America — easier to rule over a divided, chaotic empire than a functioning, high-trust, middle-class society.” “That’s their model for America: to make the whole country into a version of Somalia. Everything they do tracks back to that goal.”

Jesse Watters

345,304 次观看 • 7 个月前

Higher learning meets higher power on the all new episode of #GameKnights! Can new student Danny make it to the top of the class at #MTGStrixhaven? Tune in to find out who achieves campus glory and who flunks out: #MTG Jimmy Wong Josh Lee Kwai Rachel Weeks #sponsored Magic: The Gathering

Higher learning meets higher power on the all new episode of #GameKnights! Can new student Danny make it to the top of the class at #MTGStrixhaven? Tune in to find out who achieves campus glory and who flunks out: #MTG Jimmy Wong Josh Lee Kwai Rachel Weeks #sponsored Magic: The Gathering

The Command Zone

11,933 次观看 • 2 个月前

Over the last few months, we’ve been thinking about how to learn from “off-domain” data - data from non-robot sources like video or simulation. These data sources are not quite good enough to learn policies (even monolithic VLA models) directly, but they still contain lots of information that can be useful for generalizable robot control. How can we develop robot learning models that are able to make use of this type of data for generalizable control? In new work, that we call HAMSTER, we show that VLMs can be useful for enabling robotic learning from off-domain data, but specifically when used through hierarchical VLA architectures. We show that this class of models can learn generalizable robot policies for the real world from large-scale, off-domain data. A 🧵 (1/10)

Over the last few months, we’ve been thinking about how to learn from “off-domain” data - data from non-robot sources like video or simulation. These data sources are not quite good enough to learn policies (even monolithic VLA models) directly, but they still contain lots of information that can be useful for generalizable robot control. How can we develop robot learning models that are able to make use of this type of data for generalizable control? In new work, that we call HAMSTER, we show that VLMs can be useful for enabling robotic learning from off-domain data, but specifically when used through hierarchical VLA architectures. We show that this class of models can learn generalizable robot policies for the real world from large-scale, off-domain data. A 🧵 (1/10)

Abhishek Gupta

11,994 次观看 • 1 年前

Dads, If your son doesn’t make it to the league or go D1 life isn’t over it’s just starting. You can pretend to care, but you can’t pretend to be there.

Dads, If your son doesn’t make it to the league or go D1 life isn’t over it’s just starting. You can pretend to care, but you can’t pretend to be there.

Carl Reed Jr.

13,054 次观看 • 1 年前

Better Call Saul first two seasons were the most stressful according to Bob Odenkirk. “The pressure was just through the roof.“ “The amount of lines, can I do this at all?” “I took one acting class in college. I was learning by doing it.”

Better Call Saul first two seasons were the most stressful according to Bob Odenkirk. “The pressure was just through the roof.“ “The amount of lines, can I do this at all?” “I took one acting class in college. I was learning by doing it.”

Emir Han

158,533 次观看 • 8 个月前

A commenter recently asked me why the political class is so willing to let everything be destroyed, when all they'd be ruling over are the ashes of the Empire. For them, however, it's not the ashes. As courtiers of the ruling class, they are protected members of society that not only get saved from the destruction, but profit from it as well.

A commenter recently asked me why the political class is so willing to let everything be destroyed, when all they'd be ruling over are the ashes of the Empire. For them, however, it's not the ashes. As courtiers of the ruling class, they are protected members of society that not only get saved from the destruction, but profit from it as well.

Chris Hedges

89,724 次观看 • 5 个月前

We heard from New Yorkers about the traffic changes they want to see on 34th Street in Manhattan. The busway we’re going to add to this busy thoroughfare will make it safer AND faster so everyone can get around as they Get Stuff Done. That’s real results for the working-class people of this city!

We heard from New Yorkers about the traffic changes they want to see on 34th Street in Manhattan. The busway we’re going to add to this busy thoroughfare will make it safer AND faster so everyone can get around as they Get Stuff Done. That’s real results for the working-class people of this city!

Mayor Zohran Kwame Mamdani

128,230 次观看 • 11 个月前

Colin Anderson (former CFO, Palantir) on forward-deployed engineering; when it works and when it doesn’t. “The FD model is great in certain markets, terrible in others… you’ve got to be solving a hard enough problem.” “A world-class engineer plus their equity dilution is an expensive check. You have to be a capital allocator to pursue this model.” That’s why it doesn’t make sense for SMB contracts in the $20k–$100k range, unless it’s essentially R&D you plan to productize. At scale, though, it can pay off: “Seven-figure contracts can support it; eight or nine figures definitely can.” The key, he says, is mindset. Too many founders think of FD as a sales or customer success function. “If you’re looking at the FD model through pre/post-sales org charts, you’ve lost the plot. Start from first principles.”

Colin Anderson (former CFO, Palantir) on forward-deployed engineering; when it works and when it doesn’t. “The FD model is great in certain markets, terrible in others… you’ve got to be solving a hard enough problem.” “A world-class engineer plus their equity dilution is an expensive check. You have to be a capital allocator to pursue this model.” That’s why it doesn’t make sense for SMB contracts in the $20k–$100k range, unless it’s essentially R&D you plan to productize. At scale, though, it can pay off: “Seven-figure contracts can support it; eight or nine figures definitely can.” The key, he says, is mindset. Too many founders think of FD as a sales or customer success function. “If you’re looking at the FD model through pre/post-sales org charts, you’ve lost the plot. Start from first principles.”

TBPN

84,418 次观看 • 10 个月前