Video yükleniyor...

Video Yüklenemedi

Bu video yüklenirken bir sorun oluştu. Bu geçici bir ağ sorunundan kaynaklanıyor olabilir veya video kullanılamıyor olabilir.

Ana Sayfaya Dön

tiny-diffusion, but Japanese! I wonder how logographic languages (Japanese, Chinese, etc) compare to phonetic/alphabetic languages in generation quality and speed with character-level tokenizers. The main difference is the semantic-value-per-token. Fewer tokens are needed to express an idea, which leads to fewer AR and diffusion steps. My main question is... show more

nathan (in sf)

3,249 subscribers

54,185 görüntüleme • 7 ay önce •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

0 Yorum

Yorum bulunmuyor

Orijinal gönderinin yorumları burada görünecek

Benzer Videolar

The biggest problem humanity faces is that we're running out of people. Sounds insane, I know. Fewer people means fewer mouths to feed, more resources to go around, and less pressure on the planet. All true, until you follow it through. GDP is driven by demographics. A smaller workforce means a lower ceiling on output, and we're already carrying more debt than we can service. So governments are left with one question. How do you grow when there's no one left to do the work? There's only one answer. Robots and AI. That's why the Exponential Age is arriving faster than anyone forecast, not through choice but through necessity.

The biggest problem humanity faces is that we're running out of people. Sounds insane, I know. Fewer people means fewer mouths to feed, more resources to go around, and less pressure on the planet. All true, until you follow it through. GDP is driven by demographics. A smaller workforce means a lower ceiling on output, and we're already carrying more debt than we can service. So governments are left with one question. How do you grow when there's no one left to do the work? There's only one answer. Robots and AI. That's why the Exponential Age is arriving faster than anyone forecast, not through choice but through necessity.

Raoul Pal

246,269 görüntüleme • 1 ay önce

NEW: Pudgy Penguins' Luca Netz 🐧✳️ SAYS WHAT Ansem 🐂🀄️ IS DOING WITH HIS TOKEN PROBABLY REINVENTS HOW AIRDROPS ARE DONE FOREVER "IN HINDSIGHT IT WAS OBVIOUS, BUT THE IDEA THAT YOU HAVE AN EVERGREEN RECURRING AIRDROP IS ACTUALLY THE WAY TO DO IT AND IS THE HIGHEST EV WAY TO DISTRIBUTE TOKENS"

NEW: Pudgy Penguins' Luca Netz 🐧✳️ SAYS WHAT Ansem 🐂🀄️ IS DOING WITH HIS TOKEN PROBABLY REINVENTS HOW AIRDROPS ARE DONE FOREVER "IN HINDSIGHT IT WAS OBVIOUS, BUT THE IDEA THAT YOU HAVE AN EVERGREEN RECURRING AIRDROP IS ACTUALLY THE WAY TO DO IT AND IS THE HIGHEST EV WAY TO DISTRIBUTE TOKENS"

DEGEN NEWS

125,688 görüntüleme • 1 ay önce

The future of DeFi is DeFAI 🟧 $ANON is at the forefront of this with our main goal of making DeFi easy to access & use FOR EVERYONE (DeFi is bigger then just you and me). Which is why Hey Anon currently supports multiple languages, including Korean, Chinese, Spanish, Italian, Portuguese, Turkish, Japanese, and Finnish & we're currently working on expanding to support all languages. Currently it can understand inputs in these languages but defaults to responding in English. The team is working on enhancing language features, such as adding natural language voice support for both anon & Gemma AI.

The future of DeFi is DeFAI 🟧 $ANON is at the forefront of this with our main goal of making DeFi easy to access & use FOR EVERYONE (DeFi is bigger then just you and me). Which is why Hey Anon currently supports multiple languages, including Korean, Chinese, Spanish, Italian, Portuguese, Turkish, Japanese, and Finnish & we're currently working on expanding to support all languages. Currently it can understand inputs in these languages but defaults to responding in English. The team is working on enhancing language features, such as adding natural language voice support for both anon & Gemma AI.

Daniele 🟧 ( Meme Quant )

56,011 görüntüleme • 1 yıl önce

I was trying to find the Babymetal video on CNN that was mentioned on Reddit for quite a while, and I managed to find it on CNN J. As you can see, it has the audio dubbed in Japanese, but you can still understand it in both languages. #Babymetal #CNN #TheBrief

I was trying to find the Babymetal video on CNN that was mentioned on Reddit for quite a while, and I managed to find it on CNN J. As you can see, it has the audio dubbed in Japanese, but you can still understand it in both languages. #Babymetal #CNN #TheBrief

Babymetal Tube 🍰🎂🍫

41,544 görüntüleme • 1 yıl önce

Interestingly, as we have AI agents that run in the background, the speed of AI becomes incrementally less important than the core underlying capability level. When you could only give AI small bits of work, then the speed of response mattered a ton. The rate at which you can go back and forth with AI in real-time was the determining factor on how useful it is. But as AI agents can perform more complex and useful tasks in parallel behind the scenes, you can much more easily afford to wait longer for work to get done assuming it’s valuable. Now the focus becomes more on how you review and orchestrate the agents’ work. And the main factor is just how useful and usable is the output that the agent came back with.

Interestingly, as we have AI agents that run in the background, the speed of AI becomes incrementally less important than the core underlying capability level. When you could only give AI small bits of work, then the speed of response mattered a ton. The rate at which you can go back and forth with AI in real-time was the determining factor on how useful it is. But as AI agents can perform more complex and useful tasks in parallel behind the scenes, you can much more easily afford to wait longer for work to get done assuming it’s valuable. Now the focus becomes more on how you review and orchestrate the agents’ work. And the main factor is just how useful and usable is the output that the agent came back with.

Aaron Levie

103,811 görüntüleme • 11 ay önce

TOKENFI LAUNCHPAD IS OFFICIALLY LIVE ON MAINNET TokenFi Launchpad is a decentralized launchpad for projects that want to raise funds for their crypto tokens. It is powered by $TOKEN as its main utility token on the BNB and ETH chains, with a 2% fee charged on funds raised by every project, 50% of which is used to buy and burn $TOKEN, making it perpetually deflationary. The very first project to go live on TokenFi Launchpad is the YakDAO token sale, and it is live now. You can find information on how to participate in the YakDAO token sale here:

TOKENFI LAUNCHPAD IS OFFICIALLY LIVE ON MAINNET TokenFi Launchpad is a decentralized launchpad for projects that want to raise funds for their crypto tokens. It is powered by $TOKEN as its main utility token on the BNB and ETH chains, with a 2% fee charged on funds raised by every project, 50% of which is used to buy and burn $TOKEN, making it perpetually deflationary. The very first project to go live on TokenFi Launchpad is the YakDAO token sale, and it is live now. You can find information on how to participate in the YakDAO token sale here:

TokenFi

113,700 görüntüleme • 2 yıl önce

Tyrod Taylor says you can prolong an NFL career by how you treat people and keeping the main thing the main thing “What I would tell a younger kid or rookie coming into today’s game is how you treat people goes a long way, first and foremost.” “And keep the main thing the main thing. Football is what got you here, and football is what’s going to prolong your career.” “In today’s society, social media is a reality. It’s good in some aspects, but it also can be misleading. You never know exactly where people are, what people are going through.” “Some people early on, maybe out of ego, may try to fit in or try to keep up. And it reflects keeping the main thing the main thing, which is football.”

Tyrod Taylor says you can prolong an NFL career by how you treat people and keeping the main thing the main thing “What I would tell a younger kid or rookie coming into today’s game is how you treat people goes a long way, first and foremost.” “And keep the main thing the main thing. Football is what got you here, and football is what’s going to prolong your career.” “In today’s society, social media is a reality. It’s good in some aspects, but it also can be misleading. You never know exactly where people are, what people are going through.” “Some people early on, maybe out of ego, may try to fit in or try to keep up. And it reflects keeping the main thing the main thing, which is football.”

Josh Chambers

35,539 görüntüleme • 9 gün önce

Is entropy subjective? Entropy is core to the laws of thermodynamics and the arrow of time. And yet, some physicists argue that entropy actually depends on who is measuring it. Could it really be the case that such a fundamental aspect of the laws of physics is subjective?! I addressed this question in my talk at the Rationalist Festival (Rat Fest) 2025, organised by Conjecture Institute. Watch here or on the YouTube link below!

Is entropy subjective? Entropy is core to the laws of thermodynamics and the arrow of time. And yet, some physicists argue that entropy actually depends on who is measuring it. Could it really be the case that such a fundamental aspect of the laws of physics is subjective?! I addressed this question in my talk at the Rationalist Festival (Rat Fest) 2025, organised by Conjecture Institute. Watch here or on the YouTube link below!

Dr Maria Violaris

54,834 görüntüleme • 6 ay önce

Rewrote tiny-diffusion to be 3x smaller! Went from 951 lines to just 364, all contained in one file. As simple as possible, but not simpler. I also added a tiny GPT implementation as a comparison (312 lines, inspired by Andrej Karpathy). The two implementations are ~80% identical. The model architecture, training loop, tokenization, etc, only differ in 19 lines of code. The main differences are contained within two functions (generate and get_batch). The reason to include the GPT implementation was to show how similar autoregressive LMs are to diffusion LMs on an architectural level. Only *1* line of code in the architecture needs to be modified to support masked language diffusion instead of next-token prediction (by disabling causal masking). Link to the repo is in the comments

Rewrote tiny-diffusion to be 3x smaller! Went from 951 lines to just 364, all contained in one file. As simple as possible, but not simpler. I also added a tiny GPT implementation as a comparison (312 lines, inspired by Andrej Karpathy). The two implementations are ~80% identical. The model architecture, training loop, tokenization, etc, only differ in 19 lines of code. The main differences are contained within two functions (generate and get_batch). The reason to include the GPT implementation was to show how similar autoregressive LMs are to diffusion LMs on an architectural level. Only 1 line of code in the architecture needs to be modified to support masked language diffusion instead of next-token prediction (by disabling causal masking). Link to the repo is in the comments

Nathan Barry

161,578 görüntüleme • 7 ay önce

Live on Cline (v3.67.1) OpenAI 's GPT 5.3 Codex. The speed and token efficiency improvements are real. Here is what's new: > 25% faster than 5.2 Codex > #1 on SWE-Bench Pro (4 different programming languages) > Fewer tokens per task than any prior OpenAI model Runs cost less and finish faster. You can use it in Cline right now without setting up an API key. Select the model and try it on your repo.

Live on Cline (v3.67.1) OpenAI 's GPT 5.3 Codex. The speed and token efficiency improvements are real. Here is what's new: > 25% faster than 5.2 Codex > #1 on SWE-Bench Pro (4 different programming languages) > Fewer tokens per task than any prior OpenAI model Runs cost less and finish faster. You can use it in Cline right now without setting up an API key. Select the model and try it on your repo.

Cline

21,291 görüntüleme • 5 ay önce

Attorney Alan Dershowitz does not want to repeal the Second Amendment, but also says it clearly makes Americans less safe: “It’s in the nature of rights that they generally make the citizens less safe, that we are prepared to have a trade-off between safety and liberty…All I’m asking for in this debate is that we base that cost-benefit analysis on actual facts…If we had fewer guns, we’d have fewer deaths.”

Attorney Alan Dershowitz does not want to repeal the Second Amendment, but also says it clearly makes Americans less safe: “It’s in the nature of rights that they generally make the citizens less safe, that we are prepared to have a trade-off between safety and liberty…All I’m asking for in this debate is that we base that cost-benefit analysis on actual facts…If we had fewer guns, we’d have fewer deaths.”

Honestly with Bari Weiss

19,236 görüntüleme • 8 ay önce

Dylan on why access to the most intelligent AI models will determine who wins in every industry: "What really matters is access to the most intelligent tokens and leveraging them at the highest-value tasks. A lot of folks will want tokens, but the shitty SaaS startup in SF who is using Claude is not necessarily creating a ton of value, and therefore they're going to get priced out of tokens. Anthropic only released Mythos to certain companies for cybersecurity. You also don't want people to distill your model so you release them to a fewer and fewer customers. Those customers are now wrestling over the tokens. Token usage and therefore the additional value generated on top of those tokens aggregates among fewer and fewer companies. I don't have Mythos but you know who has Mythos, top banks. There are people like Ken Griffin of Citadel who can just sign a deal with OpenAI or Antrhopic for access to your models. I'll buy the first $10 billion worth of tokens each year. So whenever you release the model, I'll spend the first 10 billion tokens and then everyone else can get the model after that. Now, he's going to crush everyone in the markets. I can envision because I have an enterprise Anthropic contract and Anthropic people like me, they're willing to give us slightly earlier access or higher rate limits, and then my competitor doesn't have that, and I'm able to f*cking crush them."

Dylan on why access to the most intelligent AI models will determine who wins in every industry: "What really matters is access to the most intelligent tokens and leveraging them at the highest-value tasks. A lot of folks will want tokens, but the shitty SaaS startup in SF who is using Claude is not necessarily creating a ton of value, and therefore they're going to get priced out of tokens. Anthropic only released Mythos to certain companies for cybersecurity. You also don't want people to distill your model so you release them to a fewer and fewer customers. Those customers are now wrestling over the tokens. Token usage and therefore the additional value generated on top of those tokens aggregates among fewer and fewer companies. I don't have Mythos but you know who has Mythos, top banks. There are people like Ken Griffin of Citadel who can just sign a deal with OpenAI or Antrhopic for access to your models. I'll buy the first $10 billion worth of tokens each year. So whenever you release the model, I'll spend the first 10 billion tokens and then everyone else can get the model after that. Now, he's going to crush everyone in the markets. I can envision because I have an enterprise Anthropic contract and Anthropic people like me, they're willing to give us slightly earlier access or higher rate limits, and then my competitor doesn't have that, and I'm able to f*cking crush them."

Patrick OShaughnessy

36,100 görüntüleme • 3 ay önce

I have 2 goals in the years to come. 1. Stay Sober 2. Stay Hungry I may have an occasional beer or 2 but I'm getting lean and mean. I love the way I feel fasting during the day and eating meat and vegetables at night. The main thing with the sobriety is stop caring about conformity. I don't need anyone specifically. If it is to be it is up to me doing it. That's another level of freedom along with being lower body fat, more clarity and energy to think, and actually pull off ideas I see value in.

I have 2 goals in the years to come. 1. Stay Sober 2. Stay Hungry I may have an occasional beer or 2 but I'm getting lean and mean. I love the way I feel fasting during the day and eating meat and vegetables at night. The main thing with the sobriety is stop caring about conformity. I don't need anyone specifically. If it is to be it is up to me doing it. That's another level of freedom along with being lower body fat, more clarity and energy to think, and actually pull off ideas I see value in.

Jason Mauck

50,245 görüntüleme • 2 yıl önce

Kevin Garnett says he is open to working out with Victor Wembanyama in the summer: “One thing I want to speak on is Wemby’s development. If I’m him, I’m working on core strength.. I’m probably one of the strongest people you could ever run up on, and I look how I look, but when it came down to setting up on that block and then pitching and getting it off down there, yeah, I was the best at that. So I would love to work with big fella, but this generation has a preference on how they want to work out and I have to respect that.” #GoSpursGo #PorVida

Kevin Garnett says he is open to working out with Victor Wembanyama in the summer: “One thing I want to speak on is Wemby’s development. If I’m him, I’m working on core strength.. I’m probably one of the strongest people you could ever run up on, and I look how I look, but when it came down to setting up on that block and then pitching and getting it off down there, yeah, I was the best at that. So I would love to work with big fella, but this generation has a preference on how they want to work out and I have to respect that.” #GoSpursGo #PorVida

Wemby Alien Era

440,821 görüntüleme • 2 ay önce

The problem with tequila is that there are so many sh*t tequilas out there in the market. However, I will bring the best one to you. That’s why today, I’m testing all the famous tequilas. But for that, I need to know how much sugar is added in each one, how it was prepared, how it was contained, where it was produced, and of course, how it tastes. Let's see which one is the winner and which are the losers. It’s gonna be fun. Salud! Buy now:

The problem with tequila is that there are so many sh*t tequilas out there in the market. However, I will bring the best one to you. That’s why today, I’m testing all the famous tequilas. But for that, I need to know how much sugar is added in each one, how it was prepared, how it was contained, where it was produced, and of course, how it tastes. Let's see which one is the winner and which are the losers. It’s gonna be fun. Salud! Buy now:

Kevin O'Leary aka Mr. Wonderful

30,291 görüntüleme • 10 ay önce

omg Heeseung said he still has a Chinese class tomorrow... he's really working so hard on improving himself 🥹🫳🏻 💬 "I love studying languages, so I speak five languages. but honestly, the only ones I can speak properly are Korean, English, and Japanese. I'm currently studying Spanish and Chinese, but I'm not really improving. so… my goal for this year is to become better at Chinese than Ddeung-D! just you wait, my rival!" 🐹 : actually, for me too, Chinese... well, actually, as for English, I'm at a level where I can have a conversation to some extent, but as for Chinese, I'm continuously putting in effort. by the way, I also have a Chinese lesson tomorrow. 🐹 : yes, so when it comes to languages, I think repetition is always the key. If you keep working hard at it, you'll definitely see good results. five languages... wow, speaking five languages is truly amazing. I hope you know that I really recognize and appreciate how hard you're working, User 3404. yes, fighting!!!! #EVAN #에반 #HEESEUNG #희승

omg Heeseung said he still has a Chinese class tomorrow... he's really working so hard on improving himself 🥹🫳🏻 💬 "I love studying languages, so I speak five languages. but honestly, the only ones I can speak properly are Korean, English, and Japanese. I'm currently studying Spanish and Chinese, but I'm not really improving. so… my goal for this year is to become better at Chinese than Ddeung-D! just you wait, my rival!" 🐹 : actually, for me too, Chinese... well, actually, as for English, I'm at a level where I can have a conversation to some extent, but as for Chinese, I'm continuously putting in effort. by the way, I also have a Chinese lesson tomorrow. 🐹 : yes, so when it comes to languages, I think repetition is always the key. If you keep working hard at it, you'll definitely see good results. five languages... wow, speaking five languages is truly amazing. I hope you know that I really recognize and appreciate how hard you're working, User 3404. yes, fighting!!!! #EVAN #에반 #HEESEUNG #희승

마르 #RideOrDie

25,152 görüntüleme • 1 ay önce

Harry on how the album should be listened to “[…] And it became really obvious that it was music that was meant to be played and played loud. So I think everyone's different. The experience is the way that they like to experience music. Like I love listening to it alone in my car. I love listening to music with friends, and I think the album is you can have very different, uh, kind of journeys with it depending on how you listen to it. I love listening to it, like sitting still in headphones and listening to it that way. I love listening to it driving. I love being with friends.” - 102,7 KIIS FM

Harry on how the album should be listened to “[…] And it became really obvious that it was music that was meant to be played and played loud. So I think everyone's different. The experience is the way that they like to experience music. Like I love listening to it alone in my car. I love listening to music with friends, and I think the album is you can have very different, uh, kind of journeys with it depending on how you listen to it. I love listening to it, like sitting still in headphones and listening to it that way. I love listening to it driving. I love being with friends.” - 102,7 KIIS FM

HSNews

43,853 görüntüleme • 6 ay önce

I don’t know how they’re getting the quality so damn close to the actual show simply by training off of it. I remember posing this question to my professor, if diffusion models would ever be able to get one to one quality simply by training off the pixels and then manipulate it as if they actually understood the scene rather than just parroting. Back then it felt almost impossible. The assumption was that you would hit a ceiling because the model was “just” learning statistical correlations in pixel space. No semantic grounding. No true understanding. It almost feels like seedance is like an alien that can go beyond noise to replicate 1 to 1 quality sometimes.

I don’t know how they’re getting the quality so damn close to the actual show simply by training off of it. I remember posing this question to my professor, if diffusion models would ever be able to get one to one quality simply by training off the pixels and then manipulate it as if they actually understood the scene rather than just parroting. Back then it felt almost impossible. The assumption was that you would hit a ceiling because the model was “just” learning statistical correlations in pixel space. No semantic grounding. No true understanding. It almost feels like seedance is like an alien that can go beyond noise to replicate 1 to 1 quality sometimes.

Chris

130,109 görüntüleme • 5 ay önce

On the most basic level, this is how to clean historic headstones. I started this one last Saturday, and while I didn’t have time to clean it all in one day, I did get to peak at it today to see how much progress the D2 made. I’ll go back in a few more days to finish it.

On the most basic level, this is how to clean historic headstones. I started this one last Saturday, and while I didn’t have time to clean it all in one day, I did get to peak at it today to see how much progress the D2 made. I’ll go back in a few more days to finish it.

Diary of Abandonment

17,830 görüntüleme • 4 ay önce

What would Jonathan Franzen tell up-and-coming writers? "My advice is to really work on the pages. Ask yourself: Is there a way to say this in fewer words? You don't need that many details. Two sentences are all you need for a minor character. It really doesn't take much."

What would Jonathan Franzen tell up-and-coming writers? "My advice is to really work on the pages. Ask yourself: Is there a way to say this in fewer words? You don't need that many details. Two sentences are all you need for a minor character. It really doesn't take much."

David Perell

130,385 görüntüleme • 8 ay önce