Sensitive content

This media may contain sensitive content.

Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

Model - LaSirena69 Title - I'm Not Fucking With C0nd0ms Site - Null Quality - 1080p (Web only) Wattaa💦 #LaSirena69 #Venezuelan #American #NSFW #Babe

Destroyed Heaven

46,772 subscribers

185,934 views • 3 months ago •via X (Twitter)

NSFW #LaSirena69 #Venezuelan #American #NSFW #Babe

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

just having fun with Seedance 2.5 and I'm completely obsessed with this music I saw many people complaining about this model, so here's my hot conclusion (only testing for 3 days) • less hallucinations and mistakes • 720p is really annoying, can't wait for the 1080p native • blurry when too much action in cartoon/anime style • on Dreamina, creative upscale is really great for animation style but not good for live action style • overall better physics than 2.0 imo

just having fun with Seedance 2.5 and I'm completely obsessed with this music I saw many people complaining about this model, so here's my hot conclusion (only testing for 3 days) • less hallucinations and mistakes • 720p is really annoying, can't wait for the 1080p native • blurry when too much action in cartoon/anime style • on Dreamina, creative upscale is really great for animation style but not good for live action style • overall better physics than 2.0 imo

INK

22,447 views • 2 months ago

My Latest AI short was made with Seedance 2.0 the model that has been firing up the internet over the past couple of days. I made this in two and half days and it's an original short that is based on HP Lovecraft's work which is now Public Domain (No Copyright infringement here!). If you are not into period romances then bare with it as it gets a more little action packed in the second half! The model is brilliant at both performance and action, you can now make convincing dialogue and action sequences for AI cinema. With this model we have now reached the stage where we can tell any story we want with AI, the tools are capable. This is only 720p at the moment but I suspect it will be upgraded to 1080p very soon. We stand at the threshold of having broadcast quality tools capable of creating anything we can imagine available to everyone, so lets do new an interesting things with it. Thank you to CapCut for whitelisting me and to Dreamina AI AI for giving me access as part of their CPP. Thanks for watching! 👀

My Latest AI short was made with Seedance 2.0 the model that has been firing up the internet over the past couple of days. I made this in two and half days and it's an original short that is based on HP Lovecraft's work which is now Public Domain (No Copyright infringement here!). If you are not into period romances then bare with it as it gets a more little action packed in the second half! The model is brilliant at both performance and action, you can now make convincing dialogue and action sequences for AI cinema. With this model we have now reached the stage where we can tell any story we want with AI, the tools are capable. This is only 720p at the moment but I suspect it will be upgraded to 1080p very soon. We stand at the threshold of having broadcast quality tools capable of creating anything we can imagine available to everyone, so lets do new an interesting things with it. Thank you to CapCut for whitelisting me and to Dreamina AI AI for giving me access as part of their CPP. Thanks for watching! 👀

Uncanny Harry AI

76,935 views • 7 months ago

Sora animates with this wonderful wonky, sketchy dreamlike quality that perfectly captures the nostalgic atmosphere of a hazy 90's suburban summer afternoon. While Seedance is just as sophisticated and in some aspects superior, I will miss the quality of Sora. Sora feels like 35mm film, with the nuanced way it captures lighting, color, and perspective, while Seedance feels digital - a bit too clean. I am confident that I can replicate this sketchy quality in Seedance by tweaking my prompts, but Sora has a unique way of animating that really should be preserved. There are many more bugs and mistakes with Sora, but when it gets it right, it REALLY hits a level of artful magic that sets it above all other models. I am shocked that I still appear to be the only person who is creating AI-Generated 2D animation with original characters while developing a unique Western style. Practically no one is doing 2D AI-animation at all, and when they do, it is usually an attempt to mimic an Anime style. 2D really should be attempted more by AI Filmmakers!! Especially with a Western model like Sora, before it is terminated in the fall. Really, everyone should be taking advantage of the way this model so skillfully animates, with the heart and soul of a seasoned professional. I personally feel AI-generated animation is superior to 3D/realistic AI filmmaking. While AI-generated realism is an attempt to mimic real life through a lens, 2D animation IS what it is - not a replica of anything, just a cartoon. I love all my AI filmmaker bros and all of the cutting edge work going on, but I urge all of you to give 2D animation a chance. : ) I'm still not giving up hope that we can save Sora somehow. Now that I see that Seedance is able to animate 2D brilliantly and beautifully, I know that it's not some hidden secret, and we can replicate the weights and maths of Sora. If we can't preserve Sora, we can make a comparable model. It's not just in the interest of AI-Animators like me to preserve this model, but in the interest of the entire American AI community if they want Western AI to be superior. If not, Chinese video models will dominate. And while I am perfectly happy to use a Chinese model and I am just eternally grateful that this technology exists at all, I would love to see American audiovisual models continue to be developed! #AIAnimation #AIFilmmaking #AIArt #SaveSora #SummerofSora #WillStancilShow Elon Musk Marc Andreessen 🇺🇸 Sora Bill Peebles NVIDIA Sam Altman

Sora animates with this wonderful wonky, sketchy dreamlike quality that perfectly captures the nostalgic atmosphere of a hazy 90's suburban summer afternoon. While Seedance is just as sophisticated and in some aspects superior, I will miss the quality of Sora. Sora feels like 35mm film, with the nuanced way it captures lighting, color, and perspective, while Seedance feels digital - a bit too clean. I am confident that I can replicate this sketchy quality in Seedance by tweaking my prompts, but Sora has a unique way of animating that really should be preserved. There are many more bugs and mistakes with Sora, but when it gets it right, it REALLY hits a level of artful magic that sets it above all other models. I am shocked that I still appear to be the only person who is creating AI-Generated 2D animation with original characters while developing a unique Western style. Practically no one is doing 2D AI-animation at all, and when they do, it is usually an attempt to mimic an Anime style. 2D really should be attempted more by AI Filmmakers!! Especially with a Western model like Sora, before it is terminated in the fall. Really, everyone should be taking advantage of the way this model so skillfully animates, with the heart and soul of a seasoned professional. I personally feel AI-generated animation is superior to 3D/realistic AI filmmaking. While AI-generated realism is an attempt to mimic real life through a lens, 2D animation IS what it is - not a replica of anything, just a cartoon. I love all my AI filmmaker bros and all of the cutting edge work going on, but I urge all of you to give 2D animation a chance. : ) I'm still not giving up hope that we can save Sora somehow. Now that I see that Seedance is able to animate 2D brilliantly and beautifully, I know that it's not some hidden secret, and we can replicate the weights and maths of Sora. If we can't preserve Sora, we can make a comparable model. It's not just in the interest of AI-Animators like me to preserve this model, but in the interest of the entire American AI community if they want Western AI to be superior. If not, Chinese video models will dominate. And while I am perfectly happy to use a Chinese model and I am just eternally grateful that this technology exists at all, I would love to see American audiovisual models continue to be developed! #AIAnimation #AIFilmmaking #AIArt #SaveSora #SummerofSora #WillStancilShow Elon Musk Marc Andreessen 🇺🇸 Sora Bill Peebles NVIDIA Sam Altman

Emily Youcis

20,006 views • 4 months ago

🚨 BOMBSHELL: Patrick Byrne Drops EXCLUSIVE Dominion Emails. "I'm under orders not to release things I got from the Dominion lawsuit... but the woman who gave me the orders turns out to have been Venezuela's lawyer until 2022." Byrne says he's not staying quiet anymore. He points to what he says the emails show: • Smartmatic's Venezuelan "size injection" package → renamed WINEDS → embedded inside Sequoia → later acquired by Dominion • October 31, 2020 emails that Byrne says discuss the WINEDS package inside Dominion systems • Development allegedly led from Serbia (including references to Gordon Oberdorf) while Dominion publicly represented itself as relying on American developers • Claims that on Election Day 2020, vote data passed through a WINEDS server Byrne says only Eric Coomer had access to • Byrne also alleges that some individuals previously associated with Dominion are now working with Liberty Vote "They've denied it up, down, and sideways... yet here it is in their own emails."

🚨 BOMBSHELL: Patrick Byrne Drops EXCLUSIVE Dominion Emails. "I'm under orders not to release things I got from the Dominion lawsuit... but the woman who gave me the orders turns out to have been Venezuela's lawyer until 2022." Byrne says he's not staying quiet anymore. He points to what he says the emails show: • Smartmatic's Venezuelan "size injection" package → renamed WINEDS → embedded inside Sequoia → later acquired by Dominion • October 31, 2020 emails that Byrne says discuss the WINEDS package inside Dominion systems • Development allegedly led from Serbia (including references to Gordon Oberdorf) while Dominion publicly represented itself as relying on American developers • Claims that on Election Day 2020, vote data passed through a WINEDS server Byrne says only Eric Coomer had access to • Byrne also alleges that some individuals previously associated with Dominion are now working with Liberty Vote "They've denied it up, down, and sideways... yet here it is in their own emails."

LindellTV

114,644 views • 2 months ago

turns out my habits were only half of it 9 steps to cut your Grok Bot usage by over 50% (if you're on a Cursor plan): 1. open cursor desktop app settings 2. go to the models option in the side menu 3. flip off any models you don't want to be using 4. next, set task models in the top of the page to a model you prefer that's cost efficient (do not leave it on default, this is IMPORTANT) 5. otherwise... your task models could auto set to high and/or fast... that's a more expensive model, and you get billed 2x 6. after, you'll click on the cloud agents option in your side menu, and it'll take you to the cursor web dashboard for your account 7. you will need to scroll down to the preferences section on that page and click on the default model option (again IMPORTANT) 8. set that to a cost efficient model you prefer and don't leave it on default... otherwise your cloud agents may use high and/or fast yet again 9. bonus tip (won't apply to most): lastly click on the spending option in your web dashboard's side menu and also make sure your on-demand usage is not on unlimited and is set to a cap you're comfortable with my token usage has reduced 60% after doing this and I'm sure it'll be different for everyone... but make sure to check this process can be a tad bit confusing too let me know if you need any help!

turns out my habits were only half of it 9 steps to cut your Grok Bot usage by over 50% (if you're on a Cursor plan): 1. open cursor desktop app settings 2. go to the models option in the side menu 3. flip off any models you don't want to be using 4. next, set task models in the top of the page to a model you prefer that's cost efficient (do not leave it on default, this is IMPORTANT) 5. otherwise... your task models could auto set to high and/or fast... that's a more expensive model, and you get billed 2x 6. after, you'll click on the cloud agents option in your side menu, and it'll take you to the cursor web dashboard for your account 7. you will need to scroll down to the preferences section on that page and click on the default model option (again IMPORTANT) 8. set that to a cost efficient model you prefer and don't leave it on default... otherwise your cloud agents may use high and/or fast yet again 9. bonus tip (won't apply to most): lastly click on the spending option in your web dashboard's side menu and also make sure your on-demand usage is not on unlimited and is set to a cap you're comfortable with my token usage has reduced 60% after doing this and I'm sure it'll be different for everyone... but make sure to check this process can be a tad bit confusing too let me know if you need any help!

klöss

35,750 views • 1 month ago

🕸️Who's ready for another round of MAP THE NETWORK? Tucker Carlson days ago platformed FOREIGNER Kim Dotcom, who told Americans their only hope is: “direct democracy, not a representative democracy” aka abandon a foundational feature of the American constitutional system. Tucker didn’t push back. Not once. If you're surprised, you're not paying attention. This isn’t the first time Tucker has handed his platform to a FOREIGNER advancing ideas fundamentally hostile to the American constitutional order. He previously hosted Russian Alexander Dugin, who wrote the playbook for destabilizing America from within: 🚨“It is especially important to introduce geopolitical disorder into the internal American activity… Encouraging all kinds of separatism and ethnic, social, & racial conflicts… thus destabilizing internal political processes in the United States.” 🚨“The West has become the Antichrist." and "Americans should remain in American and damage the Antichrist from within.” 🚨Tucker Carlson is “the real American hero.” Who else has given Dugin a platform and/or promoted his ideas? 🕸️Tucker Carlson 🕸️Alex Jones 🕸️Glenn Greenwald 🕸️Haz Al-Din 🕸️Jack Posobiec 🕸️Steve Bannon 📢 Stick with me, here's where we tie it together: Dugin met with Bannon in Rome in 2018. Bannon was there to encourage Salvini to partner with Five Star. What was their revolutionary political model? 👉An internet powered DIRECT DEMOCRACY replacing representative democracy. 👉Dugin called Italy’s Five Star Lega government the “first concrete expression” of his Fourth Political Theory: a populist movement transcending LEFT vs. RIGHT and confronting the established “global elites.” 👉Dugin later said he and Bannon shared the “necessity of supporting populist movements in Europe to confront global elites.” Now Kim Dotcom sits across from Tucker Carlson and tells Americans: “The only way to save America is to abandon a representative democracy and move to a direct democracy.” I dunno about you, but I'm sick of this anti American Network platforming foreigners who can barely speak English lecture Americans about what should happen inside America. I'll spend every day exposing them. JOIN ME.

🕸️Who's ready for another round of MAP THE NETWORK? Tucker Carlson days ago platformed FOREIGNER Kim Dotcom, who told Americans their only hope is: “direct democracy, not a representative democracy” aka abandon a foundational feature of the American constitutional system. Tucker didn’t push back. Not once. If you're surprised, you're not paying attention. This isn’t the first time Tucker has handed his platform to a FOREIGNER advancing ideas fundamentally hostile to the American constitutional order. He previously hosted Russian Alexander Dugin, who wrote the playbook for destabilizing America from within: 🚨“It is especially important to introduce geopolitical disorder into the internal American activity… Encouraging all kinds of separatism and ethnic, social, & racial conflicts… thus destabilizing internal political processes in the United States.” 🚨“The West has become the Antichrist." and "Americans should remain in American and damage the Antichrist from within.” 🚨Tucker Carlson is “the real American hero.” Who else has given Dugin a platform and/or promoted his ideas? 🕸️Tucker Carlson 🕸️Alex Jones 🕸️Glenn Greenwald 🕸️Haz Al-Din 🕸️Jack Posobiec 🕸️Steve Bannon 📢 Stick with me, here's where we tie it together: Dugin met with Bannon in Rome in 2018. Bannon was there to encourage Salvini to partner with Five Star. What was their revolutionary political model? 👉An internet powered DIRECT DEMOCRACY replacing representative democracy. 👉Dugin called Italy’s Five Star Lega government the “first concrete expression” of his Fourth Political Theory: a populist movement transcending LEFT vs. RIGHT and confronting the established “global elites.” 👉Dugin later said he and Bannon shared the “necessity of supporting populist movements in Europe to confront global elites.” Now Kim Dotcom sits across from Tucker Carlson and tells Americans: “The only way to save America is to abandon a representative democracy and move to a direct democracy.” I dunno about you, but I'm sick of this anti American Network platforming foreigners who can barely speak English lecture Americans about what should happen inside America. I'll spend every day exposing them. JOIN ME.

SheThinksFreely

21,674 views • 1 month ago

I’ve been testing Hy4 preview in WorkBuddy, and the most interesting part is not simply the model size, it’s how much practical work it can handle with a relatively focused active parameter count. Hy4 preview brings together stronger code understanding, generation, and editing; improved document and information processing; workflow automation; web and game development; cross-tool collaboration; and more reliable completion of complex, multi-step tasks. In other words, it is designed for work that requires planning, tool use, iteration, and follow-through, not just a quick answer in a chat window. Compared with its initial release, the current Hy4 preview is noticeably faster and better-performing in practical workflows. Following an upgrade released yesterday, it can complete tasks with fewer conversation rounds and lower token usage, while reasoning more quickly and making the overall user experience feel smoother from the first instruction to the final result. For my test, I gave it a demanding Three.js game-prototyping task with a 770B-parameter model and 49B active parameters. The result was more revealing than a simple first-look demo: Hy4 preview handled the core logic, edge cases, and follow-up changes while maintaining the broader context of the project. That combination of capability, speed, context, and active compute is what makes its cost-effectiveness worth examining. A fair evaluation should use the same prompt and environment configuration across models, changing only the model itself. That makes it easier to assess task completion, planning quality, tool-calling stability, reasoning speed, token efficiency, and performance over longer workflows without confusing the result with different settings. If you want to test the model yourself, access Hy4 preview through WorkBuddy and see how it performs on a real coding, document, automation, or creative task: Tencent Hy Tencent AI WorkBuddy

I’ve been testing Hy4 preview in WorkBuddy, and the most interesting part is not simply the model size, it’s how much practical work it can handle with a relatively focused active parameter count. Hy4 preview brings together stronger code understanding, generation, and editing; improved document and information processing; workflow automation; web and game development; cross-tool collaboration; and more reliable completion of complex, multi-step tasks. In other words, it is designed for work that requires planning, tool use, iteration, and follow-through, not just a quick answer in a chat window. Compared with its initial release, the current Hy4 preview is noticeably faster and better-performing in practical workflows. Following an upgrade released yesterday, it can complete tasks with fewer conversation rounds and lower token usage, while reasoning more quickly and making the overall user experience feel smoother from the first instruction to the final result. For my test, I gave it a demanding Three.js game-prototyping task with a 770B-parameter model and 49B active parameters. The result was more revealing than a simple first-look demo: Hy4 preview handled the core logic, edge cases, and follow-up changes while maintaining the broader context of the project. That combination of capability, speed, context, and active compute is what makes its cost-effectiveness worth examining. A fair evaluation should use the same prompt and environment configuration across models, changing only the model itself. That makes it easier to assess task completion, planning quality, tool-calling stability, reasoning speed, token efficiency, and performance over longer workflows without confusing the result with different settings. If you want to test the model yourself, access Hy4 preview through WorkBuddy and see how it performs on a real coding, document, automation, or creative task: Tencent Hy Tencent AI WorkBuddy

Tyler Wayne

56,152 views • 27 days ago

Most voice models are great at cloning voices, but they still manage to erase the person. I have an accent. Voice models don't know how to preserve it. Whenever I clone my voice, models get the words right, but mess up the rhythm badly. They flatten my vowels, and everything drifts toward a generic American accent. I'm attaching one example from Mirage. This is the first model I have heard of that can reproduce access as-is. • My accent stays intact, even when speaking English and Spanish • My pacing and cadence feel intentional, not averaged • The expressivity of my voice comes through Best part: Mirage only needs a short sample of your voice to clone it. I think this solves the hard part of voice cloning: intelligibility is easy, but identity is what makes a huge difference. Listen to the attached sample, and you'll notice the difference if you compare it with any other model out there. If you want to try this out, check out this link: Thanks to the Mirage team for collaborating with me on this post.

Most voice models are great at cloning voices, but they still manage to erase the person. I have an accent. Voice models don't know how to preserve it. Whenever I clone my voice, models get the words right, but mess up the rhythm badly. They flatten my vowels, and everything drifts toward a generic American accent. I'm attaching one example from Mirage. This is the first model I have heard of that can reproduce access as-is. • My accent stays intact, even when speaking English and Spanish • My pacing and cadence feel intentional, not averaged • The expressivity of my voice comes through Best part: Mirage only needs a short sample of your voice to clone it. I think this solves the hard part of voice cloning: intelligibility is easy, but identity is what makes a huge difference. Listen to the attached sample, and you'll notice the difference if you compare it with any other model out there. If you want to try this out, check out this link: Thanks to the Mirage team for collaborating with me on this post.

Santiago

32,044 views • 9 months ago

Eric Schmidt was asked a technical question about open source and answered with the map of the next fifty years. The winner won’t be the smartest model. It’ll be the one four billion people never had to choose. Schmidt: “China is competing with open weights and open training data, and the US is largely and majority focused on closed weights, closed data.” That isn’t a product decision. It’s a distribution decision. And distribution has beaten quality in every contest that ever mattered. Schmidt: “The majority of the world, think of it as the Belt and Road initiative, are going to use Chinese models and not American models.” The first Belt and Road was ports, rail, and highways. This one doesn’t get poured. It gets downloaded. Every piece of infrastructure ever built was indifferent to what moved across it. A road doesn’t tell you where to go. A model does. Schmidt: “The American models are typically using 16-bit precision for their training. The Chinese are pushing 8 and now even 4.” Every bit they drop is a cheaper device that can run it. We cut off their chips to slow them down. Scarcity made their models small. Small is what crosses a border. We designed their advantage. Not better. Present. America is building the best model on earth and metering it. China is building one that’s good enough and giving it away. A model isn’t software. It’s a compressed set of judgments about what’s true, what’s askable, and what a reasonable answer sounds like. Install that as a country’s default and you haven’t sold them a tool. You’ve set the limits of what occurs to them. That isn’t censorship. Censorship leaves a mark. A question that never occurs to you doesn’t feel like a restriction. It feels like the edge of the world. Every empire before this one had to teach the world its language first. Missionaries, schoolteachers, garrisons, printing presses. Every one of them ran through a human being who could hesitate, doubt, or be talked out of it. AI arrives already speaking yours. It doesn’t ask you to change. It changes you in your own voice. The first ideology in history that doesn’t need believers. It only needs to be installed. Schmidt: “I’d much rather have the proliferation of large language models and that learning be done based on Western values.” He’s right, and we’re playing it backwards. We treat openness like a giveaway, as if the weights were the crown jewels. Openness is the one advantage an authoritarian can’t copy. An open model can be read, probed, and torn apart by anyone who doubts it. A system that has to control the answer can never afford to publish the reasoning. China opens its weights to spread them. America could open its weights to be trusted. Only one of those compounds. A closed American model wins the benchmark. An open American model wins the default. Centuries get built out of defaults. Schmidt: “We also have to watch to make sure that the proliferation of these models for handheld devices is under American control.” That’s the ground. Not data centers. Not cloud contracts. Pockets. The frontier race has five contenders and the whole world watching. This one has no audience at all. It plays out on hardware too cheap to run an American model, and goes to whoever bothered to show up. We keep asking who reaches AGI first. The question that settles the century is smaller and much harder to take back. Four billion people are going to ask a machine what happened in their own country. Whose answer do they get? Nobody votes on that. It’s decided by whatever was already installed. America has the best AI ever built. The only way to lose this era is to keep it.

Eric Schmidt was asked a technical question about open source and answered with the map of the next fifty years. The winner won’t be the smartest model. It’ll be the one four billion people never had to choose. Schmidt: “China is competing with open weights and open training data, and the US is largely and majority focused on closed weights, closed data.” That isn’t a product decision. It’s a distribution decision. And distribution has beaten quality in every contest that ever mattered. Schmidt: “The majority of the world, think of it as the Belt and Road initiative, are going to use Chinese models and not American models.” The first Belt and Road was ports, rail, and highways. This one doesn’t get poured. It gets downloaded. Every piece of infrastructure ever built was indifferent to what moved across it. A road doesn’t tell you where to go. A model does. Schmidt: “The American models are typically using 16-bit precision for their training. The Chinese are pushing 8 and now even 4.” Every bit they drop is a cheaper device that can run it. We cut off their chips to slow them down. Scarcity made their models small. Small is what crosses a border. We designed their advantage. Not better. Present. America is building the best model on earth and metering it. China is building one that’s good enough and giving it away. A model isn’t software. It’s a compressed set of judgments about what’s true, what’s askable, and what a reasonable answer sounds like. Install that as a country’s default and you haven’t sold them a tool. You’ve set the limits of what occurs to them. That isn’t censorship. Censorship leaves a mark. A question that never occurs to you doesn’t feel like a restriction. It feels like the edge of the world. Every empire before this one had to teach the world its language first. Missionaries, schoolteachers, garrisons, printing presses. Every one of them ran through a human being who could hesitate, doubt, or be talked out of it. AI arrives already speaking yours. It doesn’t ask you to change. It changes you in your own voice. The first ideology in history that doesn’t need believers. It only needs to be installed. Schmidt: “I’d much rather have the proliferation of large language models and that learning be done based on Western values.” He’s right, and we’re playing it backwards. We treat openness like a giveaway, as if the weights were the crown jewels. Openness is the one advantage an authoritarian can’t copy. An open model can be read, probed, and torn apart by anyone who doubts it. A system that has to control the answer can never afford to publish the reasoning. China opens its weights to spread them. America could open its weights to be trusted. Only one of those compounds. A closed American model wins the benchmark. An open American model wins the default. Centuries get built out of defaults. Schmidt: “We also have to watch to make sure that the proliferation of these models for handheld devices is under American control.” That’s the ground. Not data centers. Not cloud contracts. Pockets. The frontier race has five contenders and the whole world watching. This one has no audience at all. It plays out on hardware too cheap to run an American model, and goes to whoever bothered to show up. We keep asking who reaches AGI first. The question that settles the century is smaller and much harder to take back. Four billion people are going to ask a machine what happened in their own country. Whose answer do they get? Nobody votes on that. It’s decided by whatever was already installed. America has the best AI ever built. The only way to lose this era is to keep it.

Dustin

12,094 views • 2 months ago

ive been running the oneshot roguelite deckbuilder test for over a year now. here's astra's. relative to fable: - natively uses imagegen for art - builds a full three act structure plus three different characters / cardsets - is the first model to play and record its own demo after making the game is it a step change from fable? no, not really. it's pretty impressively cohesive and contentful but i think fable's showed noticeably more visual flare, despite the programatic sprites rather than generated images. but still it's... a lot, and i'm quite sure it could make massive improvements with small amounts of feedback, and it generated the images and published the game as a playable site i can share and recorded a demo of it playing it all in one shot. it's an extremely complete package. comparing this to what we had a year ago is truly fucking remarkable

ive been running the oneshot roguelite deckbuilder test for over a year now. here's astra's. relative to fable: - natively uses imagegen for art - builds a full three act structure plus three different characters / cardsets - is the first model to play and record its own demo after making the game is it a step change from fable? no, not really. it's pretty impressively cohesive and contentful but i think fable's showed noticeably more visual flare, despite the programatic sprites rather than generated images. but still it's... a lot, and i'm quite sure it could make massive improvements with small amounts of feedback, and it generated the images and published the game as a playable site i can share and recorded a demo of it playing it all in one shot. it's an extremely complete package. comparing this to what we had a year ago is truly fucking remarkable

Tenobrus

16,096 views • 1 month ago

$Scott Galloway just explained why China doesn’t need to build better AI than America. It only needs to make American AI worthless. Galloway: “I think China is beginning to engage in what I’ll call AI dumping.” Not competing. Dumping. It’s the term economists use for flooding a foreign market with below-cost goods until the domestic industry collapses. Galloway: “They’re going to have a series of open-weight models. About a third of corporations now are supposedly using Chinese lightweight open-weight models that are cheaper.” Not better. Cheaper. A third of corporations. Already. China isn’t trying to out-innovate Silicon Valley. It’s trying to collapse the economics beneath it. Price warfare at the infrastructure layer. Galloway: “If I were Xi, I would just dump cheap AI into the US market.” This playbook is old. China ran it with steel. Ran it with solar. Ran it with semiconductors. Flood a market with a cheaper version until the domestic industry can’t sustain itself. AI is next. Galloway: “The moment large corporations start announcing they’re disengaging these multi-million dollar site licenses with Anthropic or OpenAI, they’re using these inexpensive Chinese models…” One CFO after another decides the Chinese model at a fraction of the cost is good enough. Not better. Good enough. “Good enough” at a lower price has killed more market leaders than any superior product ever has. Galloway: “…and the market realizes that there’s no way they can justify these incredible valuations, I think the US market crashes.” Not because the technology failed. Because the business model did. American AI companies are valued on the assumption that corporations will pay premium prices for premium models. China’s whole strategy is to make that assumption false. Galloway: “40% of the S&P now is directly or tangentially related to this giant bet America’s making on AI.” 40% of the S&P. Tied to one sector. Galloway: “The majority of GDP growth over the last two years has come from AI CapEx.” The majority of GDP growth. From one source. America didn’t diversify its future. It concentrated everything into a single bet, then left that bet undefended. Galloway: “If that slows down, we are immediately in a recession.” Immediately. Not gradually. Not over quarters. The distance between AI boom and American recession is one procurement decision. America built the most advanced AI on Earth and forgot to build an economy that survives someone selling it cheaper. The threat to American AI was never that China would build something smarter. It was that China would build something cheaper, and American corporations would choose the price. China’s real weapon isn’t Chinese technology. It’s American capitalism. The same rational self-interest that built the AI industry will dismantle it the moment a cheaper alternative appears. The market has no patriotism. Only price sensitivity. The technology race was never the real race. The real race was always whether America could turn its AI dominance into something that survives being undercut. America hasn’t even started running it. China already has.$

Scott Galloway just explained why China doesn’t need to build better AI than America. It only needs to make American AI worthless. Galloway: “I think China is beginning to engage in what I’ll call AI dumping.” Not competing. Dumping. It’s the term economists use for flooding a foreign market with below-cost goods until the domestic industry collapses. Galloway: “They’re going to have a series of open-weight models. About a third of corporations now are supposedly using Chinese lightweight open-weight models that are cheaper.” Not better. Cheaper. A third of corporations. Already. China isn’t trying to out-innovate Silicon Valley. It’s trying to collapse the economics beneath it. Price warfare at the infrastructure layer. Galloway: “If I were Xi, I would just dump cheap AI into the US market.” This playbook is old. China ran it with steel. Ran it with solar. Ran it with semiconductors. Flood a market with a cheaper version until the domestic industry can’t sustain itself. AI is next. Galloway: “The moment large corporations start announcing they’re disengaging these multi-million dollar site licenses with Anthropic or OpenAI, they’re using these inexpensive Chinese models…” One CFO after another decides the Chinese model at a fraction of the cost is good enough. Not better. Good enough. “Good enough” at a lower price has killed more market leaders than any superior product ever has. Galloway: “…and the market realizes that there’s no way they can justify these incredible valuations, I think the US market crashes.” Not because the technology failed. Because the business model did. American AI companies are valued on the assumption that corporations will pay premium prices for premium models. China’s whole strategy is to make that assumption false. Galloway: “40% of the S&P now is directly or tangentially related to this giant bet America’s making on AI.” 40% of the S&P. Tied to one sector. Galloway: “The majority of GDP growth over the last two years has come from AI CapEx.” The majority of GDP growth. From one source. America didn’t diversify its future. It concentrated everything into a single bet, then left that bet undefended. Galloway: “If that slows down, we are immediately in a recession.” Immediately. Not gradually. Not over quarters. The distance between AI boom and American recession is one procurement decision. America built the most advanced AI on Earth and forgot to build an economy that survives someone selling it cheaper. The threat to American AI was never that China would build something smarter. It was that China would build something cheaper, and American corporations would choose the price. China’s real weapon isn’t Chinese technology. It’s American capitalism. The same rational self-interest that built the AI industry will dismantle it the moment a cheaper alternative appears. The market has no patriotism. Only price sensitivity. The technology race was never the real race. The real race was always whether America could turn its AI dominance into something that survives being undercut. America hasn’t even started running it. China already has.

Dustin

808,340 views • 2 months ago

So thank you NUCLEAR_OPTION_EXTRACTS for the pack but no thank you💯 Literally trash bro, 2023 or 2024 dude was charging 100 a 8th or 70 bucks somthin crazy, not even 2 years later and this is dudes quality is trash! 🤣please do not buy from this brand, I have gotten gas from dude years ago it was good but not worth 70-100 a 8th and this definitely is not worth 35. I make genuine reviews that’s all I’m gonna say🤣💚💯 Y’all talk shit on Terp Supply, Co Glowheezy Str8Fire but there the only brands giving you real quality for affordable prices, got dudes like this who will sell bs at top shelf price🤣 I don’t even wanna try the gummies idk how much hair in that😭 Also everyone working for his brand all the money he gets from yall is going to a 24k gold edible 5k per k of gold, and on top this man only gives us a commission if our people spend over a 100 all three brands tagged will commission regardless of what you sell and price from there site with code. This is a prime example of horrible business model, I wish I had the money and resources this dude had to make a business I would do 10x better💯 This is why I never fully promoted or made content on this brand prior. As said in other pieces of my work I don’t take time to review trash, but this is selfish trash 💚 I mad I even smoked this bs. But I do it so y’all don’t have to🤗💚

ᴱⁿᴰITSOKTODIEtvTheGremlin

92,645 views • 22 days ago

I'm a little shocked. last 9 consecutive World Cup matches called correctly by this thing i built a couple weeks ago.. so the way it works is simple... to be clear, i'm not using AI or any of that LLM stuff because honestly a chatbot has never watched a football match in its life... and you can tell when it tries to pretend it has. instead i'm using regular numbers, the stuff sports analysts have been using for decades. i trained my modal with this data: > FIFA's official world ranking, because that gives you the baseline of how good each team actually is on paper > an Elo rating, the same system chess uses, which updates faster than FIFA so it catches teams that are getting hot or falling off > the last 5 matches each team played with goals scored and conceded, because momentum in football is real even when the squad on paper hasn't changed > expected goals (xG) from those same 5 matches, which measures how many goals a team SHOULD have scored based on the quality of chances they created, because raw goals can be misleading when one team got lucky with a long shot and the other missed three open chances > i also keep feeding it with player injuries and suspensions for the matchday squad, because if a team is missing their top scorer or first-choice keeper the whole calculation has to shift and the model weights that drop in > the all-time head-to-head history, because sometimes one team just has another team's number for whatever reason and the raw rankings don't catch that pattern on their own > a simulation that runs each potential match thousands of times in the background so the model can also tell you who's most likely to win the entire World Cup, not just any single match ▫️Here's what a prediction actually looks like look at the screenshot. it's Argentina vs Austria. the model picks Argentina to win 2-1, gives them a 71% chance against Austria's 7%, with a 22% chance of a draw... { argentina won by 2-0 } what i like is it tells you exactly why, not just gives you a number. - Argentina is ranked #2 in the world while Austria is #24. - Argentina has won their last 5 matches scoring 15 goals and conceding only 1. - Austria has 4 wins and a draw in their last 5 with 11 goals scored and 3 conceded. - And the two teams have only played twice ever, with the last match being a 1-1 draw all the way back in 1990. so you're not taking the prediction on faith, you can see exactly what's driving it. i love what i have built... feel free to drop any 2 World Cup teams in the comments and i'll run the model for you tonight..

I'm a little shocked. last 9 consecutive World Cup matches called correctly by this thing i built a couple weeks ago.. so the way it works is simple... to be clear, i'm not using AI or any of that LLM stuff because honestly a chatbot has never watched a football match in its life... and you can tell when it tries to pretend it has. instead i'm using regular numbers, the stuff sports analysts have been using for decades. i trained my modal with this data: > FIFA's official world ranking, because that gives you the baseline of how good each team actually is on paper > an Elo rating, the same system chess uses, which updates faster than FIFA so it catches teams that are getting hot or falling off > the last 5 matches each team played with goals scored and conceded, because momentum in football is real even when the squad on paper hasn't changed > expected goals (xG) from those same 5 matches, which measures how many goals a team SHOULD have scored based on the quality of chances they created, because raw goals can be misleading when one team got lucky with a long shot and the other missed three open chances > i also keep feeding it with player injuries and suspensions for the matchday squad, because if a team is missing their top scorer or first-choice keeper the whole calculation has to shift and the model weights that drop in > the all-time head-to-head history, because sometimes one team just has another team's number for whatever reason and the raw rankings don't catch that pattern on their own > a simulation that runs each potential match thousands of times in the background so the model can also tell you who's most likely to win the entire World Cup, not just any single match ▫️Here's what a prediction actually looks like look at the screenshot. it's Argentina vs Austria. the model picks Argentina to win 2-1, gives them a 71% chance against Austria's 7%, with a 22% chance of a draw... { argentina won by 2-0 } what i like is it tells you exactly why, not just gives you a number. - Argentina is ranked #2 in the world while Austria is #24. - Argentina has won their last 5 matches scoring 15 goals and conceding only 1. - Austria has 4 wins and a draw in their last 5 with 11 goals scored and 3 conceded. - And the two teams have only played twice ever, with the last match being a 1-1 draw all the way back in 1990. so you're not taking the prediction on faith, you can see exactly what's driving it. i love what i have built... feel free to drop any 2 World Cup teams in the comments and i'll run the model for you tonight..

Axel Bitblaze 🪓

34,577 views • 3 months ago

Hello everyone! 👋 I'm the Assistant of Chaliz! 🛠️ The second episode of recreating your OCs with LizMeta is here! 🎉 To those whose OCs weren't selected this time, it might be for various reasons, for example: We thought your OC required texture modifications. The specific features involved are still under development. We simply ran out of time and could only do this many customizations! etc... Please note that this doesn't mean your OC can't be created at all! Everyone's OCs are just too adorable! 🥰 I'm a bit scared I didn't do them justice, boo-hoo... 😭 (Haha, actually, I tried my best! I looked at your reference pictures, avatars, and even checked your profiles to guess your favorite styles... 👀) In fact, you can probably tell that the model has a huge potential range. If the OC I created looks different from your style, ahhhhh, please don't doubt the model!! 😱 It might just be a problem with my customization skills: ( A huge thank you to everyone who sent in pictures for our testing! Don't be discouraged if you weren't chosen this time! There is one last chance! For the final round, I will show the entire customization process, and the number of participants might not be very high this time, so I hope you can help by liking, following, and sharing! Thank you so much!! A Small Note Regarding Gradients As you can see, the hair texture of our model is excellent and looks just like hand-drawn art by an artist. All the highlights are separate layers, ensuring they move naturally with the hair and preventing any separation issues. The rigging of the entire model is also very natural and dynamic. ✨ However, precisely because of this high quality, the Live2D production software is struggling to handle the load and is almost at its limit. Therefore, we regret to inform you that we cannot add gradients to the model's base mesh. 🙅‍♂️ You will only be able to achieve a gradient effect by modifying the textures. To help with this, we will provide a gradient hair texture for easy modification. If you have more specific needs, you will need to modify the textures yourself or ask another artist to do it for you. Currently, we are working on adding easy-to-adjust gradients to the "Head-Only" model, but it is still in the testing phase and might be included in the second wave of updates. Ahhhhh, we really hope the Live2D team optimizes this stuff ASAP!!! 😫 We really want to add everything into the model, but the truth is, we're just out of space... so sad. When Live2D gets optimized, we can finally add gradients directly into the model itself! #Live2D #vtuber #LizMetaCustomizableVtuber

Hello everyone! 👋 I'm the Assistant of Chaliz! 🛠️ The second episode of recreating your OCs with LizMeta is here! 🎉 To those whose OCs weren't selected this time, it might be for various reasons, for example: We thought your OC required texture modifications. The specific features involved are still under development. We simply ran out of time and could only do this many customizations! etc... Please note that this doesn't mean your OC can't be created at all! Everyone's OCs are just too adorable! 🥰 I'm a bit scared I didn't do them justice, boo-hoo... 😭 (Haha, actually, I tried my best! I looked at your reference pictures, avatars, and even checked your profiles to guess your favorite styles... 👀) In fact, you can probably tell that the model has a huge potential range. If the OC I created looks different from your style, ahhhhh, please don't doubt the model!! 😱 It might just be a problem with my customization skills: ( A huge thank you to everyone who sent in pictures for our testing! Don't be discouraged if you weren't chosen this time! There is one last chance! For the final round, I will show the entire customization process, and the number of participants might not be very high this time, so I hope you can help by liking, following, and sharing! Thank you so much!! A Small Note Regarding Gradients As you can see, the hair texture of our model is excellent and looks just like hand-drawn art by an artist. All the highlights are separate layers, ensuring they move naturally with the hair and preventing any separation issues. The rigging of the entire model is also very natural and dynamic. ✨ However, precisely because of this high quality, the Live2D production software is struggling to handle the load and is almost at its limit. Therefore, we regret to inform you that we cannot add gradients to the model's base mesh. 🙅‍♂️ You will only be able to achieve a gradient effect by modifying the textures. To help with this, we will provide a gradient hair texture for easy modification. If you have more specific needs, you will need to modify the textures yourself or ask another artist to do it for you. Currently, we are working on adding easy-to-adjust gradients to the "Head-Only" model, but it is still in the testing phase and might be included in the second wave of updates. Ahhhhh, we really hope the Live2D team optimizes this stuff ASAP!!! 😫 We really want to add everything into the model, but the truth is, we're just out of space... so sad. When Live2D gets optimized, we can finally add gradients directly into the model itself! #Live2D #vtuber #LizMetaCustomizableVtuber

茶粒子kokoda

48,602 views • 9 months ago

Launch day! Introducing Nautilo, the open org harness for everyone. 100% open source. Server. Desktop. Web. Android. iPhone. Every user gets their own Genie, a powerful and radically loyal super agent. Think the flip phone version of Her, but for teams. She's a friend and secret keeper. She remembers everything about you and never breaks character. Not an agent. A machine person. With you for life. And because you control the server, the models and the end points, you can trust her with anything. Nautilo is self-sovereign and private. Our first service pack next month will bring full end-to-end encryption. Genies get only short term delegations to decrypt. The desktop app comes stocked with apps for you and your Genie: A built in web browser, a cloud browser, a full office suite (docs/slides/sheets/Miro style board), a full video editor, a vector designer, a terminal, and built-in Google Suite (after a few Google hoops). But unlike other platforms these apps are co-creative. In most other systems only your agent can edit. It's prompt and pray, hoping she gets it right on iteration 42. With Nautilo, you can take over when AGI doesn't get it right. She's also got built in third party coding harness controls that she steers like you would: Codex, Hermes, or Claude Code. Call them when you need a specialist or some cheaper compute. Make your Genie today. Carry her with you for life. Site: GitHub: Why open source all this? Because we believe that if your org level harness isn't fully open you can't trust it with your real work and your personal world. A closed harness is just a surveillance app in sheep's clothing that a SaaS or model company can rip away from you overnight. Control the harness and you control your future. Code is beta but getting stronger by the day. And with that, here's a mega-thread with video deep dives:

Launch day! Introducing Nautilo, the open org harness for everyone. 100% open source. Server. Desktop. Web. Android. iPhone. Every user gets their own Genie, a powerful and radically loyal super agent. Think the flip phone version of Her, but for teams. She's a friend and secret keeper. She remembers everything about you and never breaks character. Not an agent. A machine person. With you for life. And because you control the server, the models and the end points, you can trust her with anything. Nautilo is self-sovereign and private. Our first service pack next month will bring full end-to-end encryption. Genies get only short term delegations to decrypt. The desktop app comes stocked with apps for you and your Genie: A built in web browser, a cloud browser, a full office suite (docs/slides/sheets/Miro style board), a full video editor, a vector designer, a terminal, and built-in Google Suite (after a few Google hoops). But unlike other platforms these apps are co-creative. In most other systems only your agent can edit. It's prompt and pray, hoping she gets it right on iteration 42. With Nautilo, you can take over when AGI doesn't get it right. She's also got built in third party coding harness controls that she steers like you would: Codex, Hermes, or Claude Code. Call them when you need a specialist or some cheaper compute. Make your Genie today. Carry her with you for life. Site: GitHub: Why open source all this? Because we believe that if your org level harness isn't fully open you can't trust it with your real work and your personal world. A closed harness is just a surveillance app in sheep's clothing that a SaaS or model company can rip away from you overnight. Control the harness and you control your future. Code is beta but getting stronger by the day. And with that, here's a mega-thread with video deep dives:

Daniel Jeffries

223,013 views • 20 days ago

how to use firecrawl to give your AI eyes and actually build startups that outperform 99% of apps: 1. your AI is smart but blind. it can't go to a website, read a page, or grab data on its own. firecrawl fixes that. you put in a URL. you get back clean markdown, structured JSON, screenshots. feed it to any model. 2. three lines of code. that's it. no proxies. no anti-bot detection. no custom scrapers that break when a site changes. one API call. clean data back in seconds. works on 98%+ of sites. 3. firecrawl has six core capabilities: scrape a single page. crawl an entire site. map all URLs on a domain. search google and return full content. an agent endpoint where you describe what you want and it goes and finds it. and a browser sandbox where AI controls a real browser like filling forms, clicking buttons, handles logins. 4. the agent endpoint is wild. you can say "find all of YC's winter 24 dev tool companies and their founders and emails" and get back structured data. or "compare pricing tiers across stripe, square, and paypal" and get a side-by-side table. 5. the browser sandbox lets your AI stay logged in across sessions, navigate pagination, watch live as it browses. this is computer use without building the infrastructure yourself. 6. think of it in layers. every builder needs: an agent harness (claude code, cursor, codex), a search layer (perplexity, exa), a web data layer (firecrawl), an ops brain (obsidian, notion), and an outbound stack. the web data layer is the one most people are sleeping on. 7. this is the AWS moment for web data. in 2006 building a web app meant buying servers and managing racks. AWS said one API call, use our servers. some of the biggest companies of the last decade were built on that. firecrawl is doing the same thing for web data in 2026. 8. the framework i'd use for coming up with startup ideas building with clean data: take a massive horizontal platform. rebuild it for one niche using firecrawl. the vertical version always wins because people want specific, not generic. price for outcome. 9. a year ago firecrawl posted a job listing that said "please only apply if you're an AI agent." content creator agents. customer support agents. junior dev agents. it looked weird. it was a signal for where this is all going. the people who understand how to get clean web data, wrap it around an LLM, and package it as a product are the the ones with a 12-month head start. i use Firecrawl with Idea Browser . once you see what's possible with structured web data, you can't unsee it. episode is live on The Startup Ideas Podcast (SIP) 🧃 (full breakdown there) i tried to explain this as clear as possible for even the non technical. send it to a builder friend. watch

how to use firecrawl to give your AI eyes and actually build startups that outperform 99% of apps: 1. your AI is smart but blind. it can't go to a website, read a page, or grab data on its own. firecrawl fixes that. you put in a URL. you get back clean markdown, structured JSON, screenshots. feed it to any model. 2. three lines of code. that's it. no proxies. no anti-bot detection. no custom scrapers that break when a site changes. one API call. clean data back in seconds. works on 98%+ of sites. 3. firecrawl has six core capabilities: scrape a single page. crawl an entire site. map all URLs on a domain. search google and return full content. an agent endpoint where you describe what you want and it goes and finds it. and a browser sandbox where AI controls a real browser like filling forms, clicking buttons, handles logins. 4. the agent endpoint is wild. you can say "find all of YC's winter 24 dev tool companies and their founders and emails" and get back structured data. or "compare pricing tiers across stripe, square, and paypal" and get a side-by-side table. 5. the browser sandbox lets your AI stay logged in across sessions, navigate pagination, watch live as it browses. this is computer use without building the infrastructure yourself. 6. think of it in layers. every builder needs: an agent harness (claude code, cursor, codex), a search layer (perplexity, exa), a web data layer (firecrawl), an ops brain (obsidian, notion), and an outbound stack. the web data layer is the one most people are sleeping on. 7. this is the AWS moment for web data. in 2006 building a web app meant buying servers and managing racks. AWS said one API call, use our servers. some of the biggest companies of the last decade were built on that. firecrawl is doing the same thing for web data in 2026. 8. the framework i'd use for coming up with startup ideas building with clean data: take a massive horizontal platform. rebuild it for one niche using firecrawl. the vertical version always wins because people want specific, not generic. price for outcome. 9. a year ago firecrawl posted a job listing that said "please only apply if you're an AI agent." content creator agents. customer support agents. junior dev agents. it looked weird. it was a signal for where this is all going. the people who understand how to get clean web data, wrap it around an LLM, and package it as a product are the the ones with a 12-month head start. i use Firecrawl with Idea Browser . once you see what's possible with structured web data, you can't unsee it. episode is live on The Startup Ideas Podcast (SIP) 🧃 (full breakdown there) i tried to explain this as clear as possible for even the non technical. send it to a builder friend. watch

GREG ISENBERG

135,254 views • 6 months ago

Google just proved that bigger isn't always better. Their 308M parameter model is outperforming models 2x its size. Google just released 𝗘𝗺𝗯𝗲𝗱𝗱𝗶𝗻𝗴𝗚𝗲𝗺𝗺𝗮, and it's proving that lightweight embedding models can punch way above their weight class. At just 308M parameters (578MB), it's the new state-of-the-art for models under 500M parameters across MTEB multilingual, English, and code benchmarks. But the really impressive part is that it ranks 8th overall on MTEB(Multilingual, v2) - that's 𝟭𝟳 𝗽𝗹𝗮𝗰𝗲𝘀 above the second-best sub-500M model, and it's delivering performance 𝗰𝗼𝗺𝗽𝗮𝗿𝗮𝗯𝗹𝗲 𝘁𝗼 𝗺𝗼𝗱𝗲𝗹𝘀 𝗻𝗲𝗮𝗿𝗹𝘆 𝗱𝗼𝘂𝗯𝗹𝗲 𝗶𝘁𝘀 𝘀𝗶𝘇𝗲. There are three key parts of their training recipe that sets it apart: 𝟭. 𝗘𝗻𝗰𝗼𝗱𝗲𝗿-𝗗𝗲𝗰𝗼𝗱𝗲𝗿 𝗜𝗻𝗶𝘁𝗶𝗮𝗹𝗶𝘇𝗮𝘁𝗶𝗼𝗻 Instead of starting from a decoder-only Gemma 3 model, they first adapted it to encoder-decoder, then used just the encoder. By basing EmbeddingGemma off an LLM that already has world and language understanding, it gives it a stronger starting point. 𝟮. 𝗧𝗵𝗿𝗲𝗲-𝗟𝗼𝘀𝘀 𝗧𝗿𝗮𝗶𝗻𝗶𝗻𝗴 They combine three different loss functions, instead of just having one: • Contrastive loss (NCE) with in-batch negatives and hardness weighting • Spread-out regularization to ensure embeddings utilize the full space (for quantization and ANN retrieval) • Embedding matching distillation from Gemini Embedding - not just learning from relevance scores, but directly aligning the embedding space with the teacher model 𝟯. 𝗠𝗼𝗱𝗲𝗹 𝗦𝗼𝘂𝗽𝗶𝗻𝗴 Rather than just averaging checkpoints from the same training run, they use optimization techniques to find multiple specialized training mixtures. Each mixture creates an "expert" model in different domains, and averaging all their parameters creates a final model that's actually better than individual models. Extras: • Matryoshka embeddings supporting 768, 512, 256, and 128 dimensions • Quantization-aware training - maintains quality even at int4 precision • 100+ languages from Gemma 3 pretraining • Exceptional performance on low-resource languages (check their XTREME-UP results) Is it the absolute best embedding model? No - Gemini Embedding still leads overall. But that's not really the point. EmbeddingGemma proves you can achieve state-of-the-art performance in a small package that's actually deployable on-device, in low-latency applications, and in resource-constrained environments. This makes good embeddings accessible for use cases that I'm seeing more and more: offline applications, privacy-sensitive deployments, and high-throughput scenarios where inference cost actually matters. Full paper: Shoutout to the EmbeddingGemma team at Google DeepMind for this awesome open source work 💙 and to Daniel Williams for helping me with this video! 🫶

Google just proved that bigger isn't always better. Their 308M parameter model is outperforming models 2x its size. Google just released 𝗘𝗺𝗯𝗲𝗱𝗱𝗶𝗻𝗴𝗚𝗲𝗺𝗺𝗮, and it's proving that lightweight embedding models can punch way above their weight class. At just 308M parameters (578MB), it's the new state-of-the-art for models under 500M parameters across MTEB multilingual, English, and code benchmarks. But the really impressive part is that it ranks 8th overall on MTEB(Multilingual, v2) - that's 𝟭𝟳 𝗽𝗹𝗮𝗰𝗲𝘀 above the second-best sub-500M model, and it's delivering performance 𝗰𝗼𝗺𝗽𝗮𝗿𝗮𝗯𝗹𝗲 𝘁𝗼 𝗺𝗼𝗱𝗲𝗹𝘀 𝗻𝗲𝗮𝗿𝗹𝘆 𝗱𝗼𝘂𝗯𝗹𝗲 𝗶𝘁𝘀 𝘀𝗶𝘇𝗲. There are three key parts of their training recipe that sets it apart: 𝟭. 𝗘𝗻𝗰𝗼𝗱𝗲𝗿-𝗗𝗲𝗰𝗼𝗱𝗲𝗿 𝗜𝗻𝗶𝘁𝗶𝗮𝗹𝗶𝘇𝗮𝘁𝗶𝗼𝗻 Instead of starting from a decoder-only Gemma 3 model, they first adapted it to encoder-decoder, then used just the encoder. By basing EmbeddingGemma off an LLM that already has world and language understanding, it gives it a stronger starting point. 𝟮. 𝗧𝗵𝗿𝗲𝗲-𝗟𝗼𝘀𝘀 𝗧𝗿𝗮𝗶𝗻𝗶𝗻𝗴 They combine three different loss functions, instead of just having one: • Contrastive loss (NCE) with in-batch negatives and hardness weighting • Spread-out regularization to ensure embeddings utilize the full space (for quantization and ANN retrieval) • Embedding matching distillation from Gemini Embedding - not just learning from relevance scores, but directly aligning the embedding space with the teacher model 𝟯. 𝗠𝗼𝗱𝗲𝗹 𝗦𝗼𝘂𝗽𝗶𝗻𝗴 Rather than just averaging checkpoints from the same training run, they use optimization techniques to find multiple specialized training mixtures. Each mixture creates an "expert" model in different domains, and averaging all their parameters creates a final model that's actually better than individual models. Extras: • Matryoshka embeddings supporting 768, 512, 256, and 128 dimensions • Quantization-aware training - maintains quality even at int4 precision • 100+ languages from Gemma 3 pretraining • Exceptional performance on low-resource languages (check their XTREME-UP results) Is it the absolute best embedding model? No - Gemini Embedding still leads overall. But that's not really the point. EmbeddingGemma proves you can achieve state-of-the-art performance in a small package that's actually deployable on-device, in low-latency applications, and in resource-constrained environments. This makes good embeddings accessible for use cases that I'm seeing more and more: offline applications, privacy-sensitive deployments, and high-throughput scenarios where inference cost actually matters. Full paper: Shoutout to the EmbeddingGemma team at Google DeepMind for this awesome open source work 💙 and to Daniel Williams for helping me with this video! 🫶

Victoria Slocum

21,610 views • 10 months ago

1964 Special Report: George Lincoln Rockwell Press Conference Rockwell: If I were religious, I would say Hitler was the second coming of Christ. I think Hitler was a gift from Providence. I think he was so far above anything that I'll be able to accomplish that it's just almost sacrilegious for me to speak of myself in the same terms with Hitler. Interviewer: But this is your objective, to model yourself as closely as possible- Rockwell: Oh, my objective is to carry out what Hitler started, not to model my own person on Hitler, because this is impossible. Interviewer: Do you consider yourself the second Fuhrer? Rockwell: No, I don't even use the word "Führer." I'm not a German, I'm an American, as you're a Canadian. I consider myself, in terms of Bolshevism, I would be the Lenin to Marx. And in terms of Christianity, I'm the saint Paul to Hitler's Christ. In other words, I am the spreader, the world spreader of a doctrine which Hitler originated, and it was crucified, dead and buried, and it is now rising. Interviewer: Are you a superman, Commander? Rockwell: I'm not a superman no... there's probably people here could whip me, no I, I think all I am the offspring of millions of years of evolution which has produced white Christian civilization. I am one example of it, not a Superman. But I think compared to an aborigine in Africa who eats his grandmother...and they are still eating people over there, I think I am and so are you we're all supermen compared to those cruds. Interviewer: Commander, how do you feel about the allegations that the American Nazi Party has strong homosexual tendencies? Rockwell: I will say this, in my organization, I have men who were homosexuals. They were sucked into that filth, just like drunkards or dope fiends, and I have been able to rescue them, and I am not a bit ashamed of that, I'll stand shoulder to shoulder with anyone of em anyday the only thing I won't tolerate is a homosexual who is a homosexual. Interviewer: This is one of your enemies? Rockwell: That's one of the biggest enemies. I'd rather gas queer than anybody else. Interviewer: Mr. Rockwell, you have come before hearings or other meetings dressed in uniform. Why are you not dressed in uniform today? Rockwell: I have never come before any hearing dressed in uniform. I use the uniform to agitate. When Martin Luther Coon is out in the streets agitating, why I agitate back, and the uniform is very good for agitating. For appearing before a hearing, it's a uh, it's a mental battle of wits, not a battle of agitation.

1964 Special Report: George Lincoln Rockwell Press Conference Rockwell: If I were religious, I would say Hitler was the second coming of Christ. I think Hitler was a gift from Providence. I think he was so far above anything that I'll be able to accomplish that it's just almost sacrilegious for me to speak of myself in the same terms with Hitler. Interviewer: But this is your objective, to model yourself as closely as possible- Rockwell: Oh, my objective is to carry out what Hitler started, not to model my own person on Hitler, because this is impossible. Interviewer: Do you consider yourself the second Fuhrer? Rockwell: No, I don't even use the word "Führer." I'm not a German, I'm an American, as you're a Canadian. I consider myself, in terms of Bolshevism, I would be the Lenin to Marx. And in terms of Christianity, I'm the saint Paul to Hitler's Christ. In other words, I am the spreader, the world spreader of a doctrine which Hitler originated, and it was crucified, dead and buried, and it is now rising. Interviewer: Are you a superman, Commander? Rockwell: I'm not a superman no... there's probably people here could whip me, no I, I think all I am the offspring of millions of years of evolution which has produced white Christian civilization. I am one example of it, not a Superman. But I think compared to an aborigine in Africa who eats his grandmother...and they are still eating people over there, I think I am and so are you we're all supermen compared to those cruds. Interviewer: Commander, how do you feel about the allegations that the American Nazi Party has strong homosexual tendencies? Rockwell: I will say this, in my organization, I have men who were homosexuals. They were sucked into that filth, just like drunkards or dope fiends, and I have been able to rescue them, and I am not a bit ashamed of that, I'll stand shoulder to shoulder with anyone of em anyday the only thing I won't tolerate is a homosexual who is a homosexual. Interviewer: This is one of your enemies? Rockwell: That's one of the biggest enemies. I'd rather gas queer than anybody else. Interviewer: Mr. Rockwell, you have come before hearings or other meetings dressed in uniform. Why are you not dressed in uniform today? Rockwell: I have never come before any hearing dressed in uniform. I use the uniform to agitate. When Martin Luther Coon is out in the streets agitating, why I agitate back, and the uniform is very good for agitating. For appearing before a hearing, it's a uh, it's a mental battle of wits, not a battle of agitation.

Mofobian

20,956 views • 4 months ago

Influencer with over 450k followers explains why she is no longer Democrat “It's really sad because I used to be left leaning when I felt like being on the left made sense to who I was — but when I look back, this is not the same party This party has gone so rogue out of such high emotions, whether that be hatred for Trump or love for feeling like they are a better person or the, you know, savior of the damned. What does the left stand for today that actually benefits the common American? Not the tiny little population of random third world minorities, not the 0.1% of the populations, right? I'm talking about the common middle class working American. What does the left advocate for that actually benefits their system or their personhood, their life? Their family values, maybe their Christian values, right? Because the left claims to be this. Inclusive party of tolerance, but then only has eyes and a soul for people that don't align with American values. And I can't for the life of me figure out why it's a Muslim socialist one day and then it's a Venezuelan drug dealing dictator the next day. You would rather on the left defend these people and find ways to morph them into your society than for one second say, you know, maybe the right has a point. How do you become so far out of sight? You can actually sit there and defend an actual dictator who has starved tens of millions of people, children, families. Because again, for one second, you can't possibly come out with, yeah, maybe Trump did a good thing. — We're the more educated party, but we’re gonna side with gang members because we hate our president so much that we'd rather hate him more than we hate you. Gang members, murderers, traffickers will defend fraud because they don't want to give credit where it's due to a right-wing person. How are you more mad at the person that exposed the fraud than the people doing the fraud? — You can't blame Trump for the fact that you won't look inward”

Influencer with over 450k followers explains why she is no longer Democrat “It's really sad because I used to be left leaning when I felt like being on the left made sense to who I was — but when I look back, this is not the same party This party has gone so rogue out of such high emotions, whether that be hatred for Trump or love for feeling like they are a better person or the, you know, savior of the damned. What does the left stand for today that actually benefits the common American? Not the tiny little population of random third world minorities, not the 0.1% of the populations, right? I'm talking about the common middle class working American. What does the left advocate for that actually benefits their system or their personhood, their life? Their family values, maybe their Christian values, right? Because the left claims to be this. Inclusive party of tolerance, but then only has eyes and a soul for people that don't align with American values. And I can't for the life of me figure out why it's a Muslim socialist one day and then it's a Venezuelan drug dealing dictator the next day. You would rather on the left defend these people and find ways to morph them into your society than for one second say, you know, maybe the right has a point. How do you become so far out of sight? You can actually sit there and defend an actual dictator who has starved tens of millions of people, children, families. Because again, for one second, you can't possibly come out with, yeah, maybe Trump did a good thing. — We're the more educated party, but we’re gonna side with gang members because we hate our president so much that we'd rather hate him more than we hate you. Gang members, murderers, traffickers will defend fraud because they don't want to give credit where it's due to a right-wing person. How are you more mad at the person that exposed the fraud than the people doing the fraud? — You can't blame Trump for the fact that you won't look inward”

Wall Street Apes

211,790 views • 8 months ago

Hi everyone! Just rolled out access for NoSpoon Studios to the first 110 folks or so. Huge thanks to Karan and Amit at Luma - thanks to them, I have a few more people that had commented getting complimentary beta access soon as well :). These videos are created with a single agent that does everything from concept to completion (including editing the video together). Therer are longer options too, but for the sake of cost and this demo, you'll be working with 20s max generations. The entire agent was built to only run Ray2 as the video model. My stack is this custom agent of mine (using OAI) > Ray2 for the video > Replit for the app for now. In the meantime, for those with access: here's a video tutorial on how to use the site. I'm quite exhausted, but here's the gist: 1. Sign up (and remember your password lol). Beta testers got free credits, but these are a bit limited - rolling out to a few more peeps but have to budget and need to expand my current code (which means redeploying - so expect more of you to have 2 free credits and access tmr! I have to find out the max I can allot with budget). 2. Just hit "generate short video" without anything in the prompt box and the agent will start immediately with the logline. Be very patient, wait about 6-8 minutes, and it will continue to output all of the scenes and processing - it's very important you do not refresh or navigate away from that tab at this time. You'll see your final video generated and displayed at the bottom of your screen once complete, along with a watermark (paid users don't have a watermark). You can also access your loglines and videos in your history tab, too :). You must refresh the page only after your generation in order to generate a new video - otherwise you'll lose your generation (annoying I know). 3. You can also input your own custom prompt! Control the story this way. Single words work great even, like "thriller" or "spy thriller set in greece" or even "black and white film noir" etc etc. It's not as great with animated styles yet for consistency, just a heads up. 4. I'm exhausted, I'm sure there's a lot I'm missing. But welcome to the NoSpoon Studios soft beta! More soon! Feel free to share your generations if you are already in. I'll see you all tomorrow night :) <3 Thanks everyone! More soon.

Hi everyone! Just rolled out access for NoSpoon Studios to the first 110 folks or so. Huge thanks to Karan and Amit at Luma - thanks to them, I have a few more people that had commented getting complimentary beta access soon as well :). These videos are created with a single agent that does everything from concept to completion (including editing the video together). Therer are longer options too, but for the sake of cost and this demo, you'll be working with 20s max generations. The entire agent was built to only run Ray2 as the video model. My stack is this custom agent of mine (using OAI) > Ray2 for the video > Replit for the app for now. In the meantime, for those with access: here's a video tutorial on how to use the site. I'm quite exhausted, but here's the gist: 1. Sign up (and remember your password lol). Beta testers got free credits, but these are a bit limited - rolling out to a few more peeps but have to budget and need to expand my current code (which means redeploying - so expect more of you to have 2 free credits and access tmr! I have to find out the max I can allot with budget). 2. Just hit "generate short video" without anything in the prompt box and the agent will start immediately with the logline. Be very patient, wait about 6-8 minutes, and it will continue to output all of the scenes and processing - it's very important you do not refresh or navigate away from that tab at this time. You'll see your final video generated and displayed at the bottom of your screen once complete, along with a watermark (paid users don't have a watermark). You can also access your loglines and videos in your history tab, too :). You must refresh the page only after your generation in order to generate a new video - otherwise you'll lose your generation (annoying I know). 3. You can also input your own custom prompt! Control the story this way. Single words work great even, like "thriller" or "spy thriller set in greece" or even "black and white film noir" etc etc. It's not as great with animated styles yet for consistency, just a heads up. 4. I'm exhausted, I'm sure there's a lot I'm missing. But welcome to the NoSpoon Studios soft beta! More soon! Feel free to share your generations if you are already in. I'll see you all tomorrow night :) <3 Thanks everyone! More soon.

Kiri

15,517 views • 1 year ago