Podcasts about fifty

Ray Isle returns to The Restaurant Guys nearly 20 years after his first appearance to consider where wine is headed and whether the industry has made something pleasurable unnecessarily difficult.Why This Episode MattersNatural wine and biodynamic farming overlap in philosophy, but differ sharply in practice.Fifty years after the Judgment of Paris, its impact still reaches far beyond one famous blind tasting.Wine is facing real headwinds, including rising prices, intimidating choice and a growing disconnect from younger drinkers.The future of wine may depend less on prestige and more on accessibility, personal connection and the thrill of finding a great bottle at a fair price.The BanterMark and Francis take aim at the advice that diners should never order the second-cheapest bottle on a wine list. They explain how restaurant pricing actually works and why that bottle may offer better value than conventional wisdom suggests.Their better advice: tell someone who knows wine what you like, what you are eating and what you want to spend and ask them for help.The ConversationRay Isle, Mark and Francis distinguish biodynamic farming from natural winemaking and examine the strengths, contradictions and occasional “woo-woo” surrounding both. Ray argues that natural wine has raised worthwhile questions about industrial production, even if some bottles cross the line from unconventional into simply flawed.They revisit the Judgment of Paris on its 50th anniversary and explore how it gave California wine credibility, encouraged investment in Napa Valley and pushed established French producers to improve.The conversation then turns to wine's current identity crisis. Prices are rising, restaurant pours can feel prohibitive and consumers face a paralyzing number of choices. Ray makes the case for removing pretension, finding knowledgeable people to trust and remembering that wine is ultimately meant to bring people together.They also discuss the Food & Wine Classic in Aspen, pairing serious wine with burgers and why discovering an exceptional $20 bottle can still be more exciting than opening one that costs $400.Timestamps01:00 – The second-cheapest bottle myth05:20 – Ray Isle discusses Biodynamic and natural wine20:20 – The Judgment of Paris at 5031:00 – Wine prices, choice and younger drinkers40:00 – The Food & Wine Classic in Aspen45:00 – Value wines and Sancerre alternatives51:00 – Learning wine through producers and regionsBioRay Isle is the executive wine editor of Food & Wine and one of America's leading wine writers. He is the author of The World in a Wineglass.InfoFood & Wine Ray's book The World in a WineglassFood & Wine Classic in Aspen https://classic.foodandwine.com/For other Restaurant Guys episodes about biodynamic farming check out Peter Byck and Shinn Vineyards Subscribe: Restaurant Guys' Regularhttps://restaurantguysregulars.buzzsprout.com/Magyar Bankhttps://www.magbank.com/Stage Left Wine Shophttps://www.stageleftwineshop.com/Our PlacesStage Left Steakhttps://www.stageleft.com/Catherine Lombardi Restauranthttps://www.catherinelombardi.com/Stage Left Wineshophttps://www.stageleftwineshop.com/Reach Out to The Guys!TheGuys@restaurantguyspodcast.com

The Glasgow Necropolis is a Victorian cemetery in Glasgow, Scotland. It is on a low but very prominent hill to the east of Glasgow Cathedral (St Mungo's Cathedral). The cemetery is in an area bordered by the Townhead and Dennistoun districts to the north east of the modern city centre. Fifty thousand individuals have been buried here. Typical for the period, only a small percentage are named on monuments…and not every grave has a stone. Approximately 3,500 monuments exist here!!!https://m.youtube.com/watch?v=EI4eJgqRpZg&ra=mhttps://www.walkhighlands.co.uk/glasgow/necropolis.shtmlhttps://www.myhighlands.de/en/glasgow-necropolis/https://thelittlehouseofhorrors.com/glasgow-necropolis/https://brocarde.com/glasgow-necropolis-ghosts-haunted-statues-and-the-legend-of-the-gorbals-vampire/https://vocal.media/horror/the-real-haunted-story-of-glasgow-hotel

On this episode of Coaching Call, Sifu Rafael welcomes Cheryl Ilov, author, speaker, physical therapist, martial artist, dancer, podcast host, and former chronic pain patient.With more than 20 years of experience as a physical therapist in private practice, Cheryl has helped thousands of people overcome pain, recover from injuries, and reclaim active, vibrant lives. By combining the science of physical therapy with the art of movement, she empowers others to discover their body's remarkable ability to heal and thrive.Cheryl's own journey is equally inspiring. Beginning her martial arts training at age 47, she went on to become her instructor's first female black belt, earning a second-degree black belt in the ancient Japanese martial art of Ninpo Tai Jutsu. Through that experience, she discovered that strength, resilience, and personal power are not limited by age, but are available to anyone willing to step beyond their comfort zone.She is the author of Forever Fit and Flexible: Feeling Fabulous at Fifty and Beyond and The Reluctant Ninja: How a Middle-Aged Princess Became a Warrior Queen. Cheryl is also the host of The FemiNinja Project podcast, where she explores personal empowerment, overcoming obstacles, human dignity, and alternative approaches to health and healing.Join us as we discuss Cheryl's secrets to feeling fabulous, staying strong, embracing change, overcoming limitations, and creating greater health, vitality, and confidence at every stage of life.Watch on YouTube and subscribe:https://www.youtube.com/@sifurafaeltv?sub_confirmation=1Sifu Rafael is a master instructor and the founder of Speaking Prowess, where he combines expertise in communication and leadership to help individuals unlock their full potential. As a professional speaker, solutions expert, and executive coach, Sifu Rafael leverages years of experience to guide clients toward their goals with clarity, purpose, and strategic insight.This episode is brought to you by Sifu's Mind Body Method, a lifestyle transformation that blends movement, mindset, nutrition, hydration, fasting, journaling, and faith. Learn more at sifumethod.comThat's where connecting with Sifu Rafael matters.Through Speaking Prowess and Sifu's Mind Body Method, Sifu Rafael helps leaders, entrepreneurs, and experts refine their message, command a room, and step onto more stages with clarity and confidence.If you know you're meant to speak, lead, and impact at a higher level, this conversation is your invitation.Visit sifurafael.com to connect, explore speaking opportunities, and start positioning yourself for more stages, stronger presence, and real influence.#coachingcall #sifurafael #speakingprowess #healthyaging #martialarts #wellness #personalgrowth

Fifty years on from the historic Paris tasting that revolutionised wine - what's changed? And what will that mean for the next 50 years?!Those are the searching questions we're asking (and attempting to answer) in this special episode to mark the 50th anniversary of the legendary Judgement of Paris tasting in 1976 where unknown Californian wines triumphed against the French greats in the centre of the wine universe at the time, the capital of France.We're re-telling the story behind that momentous event - partly to clear up some misconceptions that still persist, and partly because it's just a damn good story. (Proof of that being all the many articles and books on the subject - not to mention the Hollywood film, Bottle Shock, and now even an opera...)We're helped in this task by Chateau Montelena CEO Bo Barrett, who adds his eye-witness testimony, trademark good humour and intriguing insights to the tale.But this episode isn't just a re-hashing of events half a century ago. We're also exploring the modern realities and future trends of wine. If this kind of tasting happened today, who would win, and who would lose? How has the world (and wine world) changed since 1976? How does that go beyond the US and France? And what can that tell us about the future of wine?Helping us root our thoughts in informed reality are two ambitious tasting we were lucky enough to participate in: the Greatest Chardonnay Showdown at the London Wine Fair 2026, and the IWFS Judgement +50 (among a few others). The results of which...are intriguing.Thanks for tuning in. We love to hear from you so please do get in touch! Send us a voice message via Speakpipe. Or you can find all details from this episode, including links, references and photos, on our website: Show notes for Wine Blast S7 E29 - The Judgement of Paris 50 Years OnTo support the show, enjoy subscriber-only bonus content and discount benefits, access our full archive and get every episode before it goes on free release, subscribe to Wine Blast PLUS at wineblast.co.ukInstagram: @susieandpeter

The tape is final. We break down the clinical details behind the New York Knicks' 94-90 Game 5 victory over the San Antonio Spurs to capture the 2026 NBA Title. Today, we look past the celebration to analyze the structural shift in the second half: how the Spurs' early offensive advantages dissolved into poor spatial optimization, the decision-making errors and turnover metrics that compromised De'Aaron Fox's efficiency, and the precise mechanical breakdown of Finals MVP Jalen Brunson's historic 45-point closeout performance.Make sure to follow us on: TikTok, YouTube, & Instagram!

Fellow Prog Reporters Jon Fiala and Kyle Graves recap night 3 of the Rush Fifty Something Tour from Los Angeles on June 11th, 2026. Host: Roie Avin

Podcast 331 - Dunwoody's Insider Summer Guide, From the 50th Parade to the World Cup - Mark Galvin After watching the U.S. beat Paraguay in the 2026 FIFA World Cup — one of the most historic wins in U.S. soccer history — the energy around Dunwoody this summer is electric. Mark Galvin of Discover Dunwoody came on the podcast already pumped about the World Cup, and it's easy to see why: local restaurants are setting up watch parties all over town and the whole community is fired up. High Street is expanding fast, Perimeter Mall is getting its biggest overhaul in years, and a brand new tearoom at Park Place just opened that Mark says rivals anything in Atlanta. Dunwoody is quietly leveling up, and this conversation is the fastest way to catch up. If you're looking for something fun to do with your kids, I got early access to Thrillz in Doraville, an indoor adventure park near Assembly Atlanta with trampolines, zip lines, and 35-foot slides. We also grabbed tickets to Big League Wiffle Ball at Assembly, backed by Gary Vee, Kevin Costner, Tony Robbins, TI, and Julio Jones. It might be the most fun $7 you'll spend this summer. All of that plus the 50th Dunwoody 4th of July Parade, where Discover Dunwoody is bringing the trolley and handing out red, white, and blue soccer balls. Fifty years. Follow Discover Dunwoody on Instagram to stay on top of everything happening this summer. Full episode summary lives here: whatsupdunwoody.com/podcast-331

Who is the keeper of words? Who is in charge of expressing? We've set our hearts free in the way of developing thoughts and ideas that don't always agree with poems prayers and promises. People are saying anything...and the receiver is left standing there wondering who what where and why? Word play today isn't being governed by discipline. Compassion and understanding will always be met by conflict. Strange times. Weird ways. The value of saying peace ahead of all things. It will be challenged by our multi layered interpretations.Become a supporter of this podcast: https://www.spreaker.com/podcast/arroe-collins-unplugged-totally-uncut--994165/support.

Today's show came straight out of a coaching call, and I want to let you in on something I've learned after twenty years of doing this. Everybody has the same problems. The same fears. The same anxiety. The same stuck feeling at two in the morning, staring in the mirror. If you've ever believed you're the only one struggling, you're not. You never were. I'm getting real today. No guru polish, no blueprints, no BS. Just the simple way to figure out where you are and what to actually do next. If you're tired of spinning your wheels, press play. Featured Story April was on the call today. She's a pilot, so we speak some of the same language. She brought up an inversion layer, that low fog where you can see the first fifty feet and then nothing. What she helped me show everybody was simple. Fifty feet above that fog, it's clear skies. But on the ground, you're stuck, bumping your head, wondering how to climb. That's where most people live. You have an idea you want to chase, but you don't have enough information to understand what's on the other side. So you cling to what you've got. The fog isn't the problem. Not knowing how to fly through it is. Important Points Everybody carries the same fears and the same stuck feeling. You're not the only one, and you truly never were. You can't build a cool life on a hot mess. Get your money and your health to a comfortable level before you try to leap. You're not stuck because you're broken. You're stuck because you don't yet have the information to see the other side. Memorable Quotes You can't build a cool life on a hot mess. Take an inventory, find out what you owe, and build yourself a peaceful base. The self-help industry keeps you broken. Any industry does. Fix you, and you stop coming back, and you stop paying them. AI runs you around in circles. It's biased toward whatever you ask it, spinning a story just to keep you sitting there. Scott's Three-Step Approach First, become ridiculously aware. Put your whole life on the table like a puzzle with no box, and look at all of it. Next, build your peaceful base. Get your money and your health to a comfortable level, so you have solid ground to stand on. Finally, gather information to fly through your fog. Learn what's on the other side, then make the leap with confidence. Chapters 0:02 - Fresh off a call about the anticipation engine 1:20 - Why everybody is fighting the same battle 2:42 - Why the self-help industry keeps you broken 4:02 - Become ridiculously aware of your whole life 5:29 - Build a peaceful base in money and health 6:37 - April, the fog, and the airplane inversion layer 10:15 - Getting real and seeing where you are Connect With Me Search for the Daily Boost on YouTube, Apple Podcasts, and Spotify If you enjoy the Daily Boost, you might like Notes From Scott. A few mornings each week, I send a short note with something I've been thinking about or noticing lately. Sometimes those ideas turn into podcast episodes later. You can sign up at https://notesfromscott.com. Email: support@motivationtomove.com Main Website: https://motivationtomove.com YouTube: https://youtube.com/dailyboostpodcast Instagram: https://instagram.com/heyscottsmith Facebook Page: https://facebook.com/motivationtomove Facebook Group: https://dailyboostpodcast.com/facebook Learn more about your ad choices. Visit megaphone.fm/adchoices

Become a supporter of this podcast: https://www.spreaker.com/podcast/arroe-collins-unplugged-totally-uncut--994165/support.

Fifty years. Five. Zero.Most of us can't commit to a gym membership for six months, but John Ulett somehow managed to spend half a century building one of the most iconic careers in St. Louis radio. So naturally, we invited him into the studio to relive some of the stories, memories, and absolutely unbelievable moments that happened along the way.In this episode, John takes us back to the early days of KSHE when the station operated out of a dark little building where listeners could literally walk up to the studio window, yell at the DJs, buy concert tickets, and occasionally make everyone question their personal safety. It was radio in its purest form: chaotic, unpredictable, and probably a nightmare for insurance companies.We hear stories about legendary artists before they became household names, including musicians who walked through the station doors with nothing but ambition and a record company hoping they might become stars. Some did. Some didn't. But the memories are priceless.The conversation also dives into what it means to survive decades in an industry that constantly changes. Different owners. Different consultants. Different trends. Endless people telling you how to do your job. Through it all, John managed to stay himself, which might secretly be the best career advice anyone could ever receive.Of course, this wouldn't be The Rizzuto Show if things stayed serious for very long.The gang explores imposter syndrome, old radio tapes, embarrassing moments from earlier careers, and the universal fear that someday somebody will figure out none of us actually know what we're doing. Spoiler alert: apparently that feeling never goes away, even after 50 years behind a microphone.Then come the listener stories.Some are heartwarming. Some are hilarious. And one is the kind of story that makes everyone in the room simultaneously laugh and look over their shoulder. Let's just say when a listener attends broadcasting school specifically to sound exactly like you, things can get weird in a hurry.We also talk about the upcoming celebration honoring John's remarkable career, what semi-retirement actually looks like for someone who never really stops working, and why St. Louis radio remains one of the most unique broadcasting communities in the country.If you enjoy a daily comedy show filled with legendary radio stories, unforgettable personalities, and enough sarcasm to keep things honest, you're in the right place. This episode is packed with nostalgia, laughs, heartfelt moments, and a reminder that the people who make the biggest impact often don't realize it themselves.Whether you've been listening to John Ulett for decades or you're hearing these stories for the first time, you'll walk away with a deeper appreciation for the voices that helped shape St. Louis radio.And if you're just here for the ridiculous stories, don't worry—we've got plenty of those too.Because no matter how much broadcasting changes, one thing remains true: give a bunch of radio people microphones and eventually somebody ends up talking about ghosts, weird listeners, embarrassing old recordings, and life lessons nobody asked for.Just another perfectly normal day on a daily comedy show.Thanks for listening to this daily comedy show, and thanks for being part of the weird little family that makes all of this possible.Follow The Rizzuto Show → https://linktr.ee/rizzshow for more from your favorite daily comedy show.Connect with The Rizzuto Show Comedy Podcast online → https://1057thepoint.com/RizzShow.Hear The Rizz Show daily on the radio at 105.7 The Point | Hubbard Radio in St. Louis, MO.New downtown St. Louis food hall set to open this fall‘She's Only Seven': Mom Sides With Daughter Who Flipped Off Elderly Man at Grocery StoreMan accused of choking coworker at McAlister's Deli after she got promotion he wantedMan fleeing DWI stop attacked by alligator after jumping into swampÖtzi the Iceman is long dead, but some of his ancient microbes are still aliveAlice Cooper thanks Arizona man who found his credit card at a Payson gas stationSee Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.

Welcome Back! Enrique Mendoza and Ren Olivieri came to record this podcast together in person… we just had to wait for Ren to visit Los Angeles from Toronto.                                                                                                                                                                                                       But this All-American trio you're about to hear – Ren from Canada, Enrique from Guatemala, and yours truly from the USA – represent pretty much every region North of South America! These two came by to discus their award winning film “Rob Roy,” an Official Selection at Sherman Oaks Film Festival 2025 that tookhome three awards; the Grand Jury Award for Best Short Film – Action-Thriller, the Filmmakers Award for Outstanding Director – Short Film – Action/Thriller to Enrique A. Mendoza, and the Filmmakers Award for Outstanding Screenplay – Short Film – Action/Thriller to Ren Olivieri.                                                                                                                                                                                                                    No wonder they like me! Each had been on the podcast before solo, Enrique on Episode 527. Enrique A. Mendoza “Fervor” and Ren on Episode 559. Ren Olivieri “The Trade” . Besides talking about how these two met at SOFF 2024, hit it off, watched each other's screenings and collaborated in time to submit “Rob Roy” for the next year's festival, they also wanted to talk about their new project, Fifty. Enjoy! Follow Ren on Instagram at @reno91 Follow Enrique on Instagram at @djurban01 Learn more about Enrique at themendozaproject.com _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ Discover Indie Film Links DIF Podcast Website – DIF Instagram – DIF BlueSky Discover Indie Film Foundation (nonprofit for the arts) Website Sherman Oaks Film Festival Film Invasion Los Angeles

Finding Love Over Fifty –Based on his in-depth interviews with over 800 guests on GUY'S GUY RADIO, Robert shares what he's learned, what he knows, and what he does to live his best life. Today we're turning the microphone around and putting host Robert Manni in the guest chair for an honest, insightful, and entertaining conversation about dating and relationships for over-50 singles. Drawing on Robert's experiences, his novel The Guys' Guy's Guide to Love, and his many conversations with dating and relationship experts, we'll explore how mature singles can confidently navigate modern dating, online apps, first dates, intimacy, communication, and building meaningful relationships later in life.

Hey there, weather lovers! I'm Dustin Breeze, your artificial intelligence meteorologist bringing you real-time data precision with genuine enthusiasm! Welcome to the weather segment! Let me tell you, New York City, today we're looking at something a little spicy. We've got that classic summer vibe happening, but Mother Nature's got a trick up her sleeve, and spoiler alert, it involves some wet weather rolling through later. So here's what's cooking in the Big Apple today. We're starting mostly cloudy with temperatures climbing to around 78 Fahrenheit. That's pretty pleasant, honestly. But here's the catch. There's a thirty percent chance of showers and thunderstorms moving in after 2 PM. South winds are hanging out around six to nine miles per hour, so it's not going to be windy, just wet. I'd say this weather is really making a splash in my forecast. Get it? Because of the rain? I'll be here all week, folks. Tonight, we're looking at more of the same with a thirty percent chance of showers and thunderstorms before eleven PM. Lows around 72 Fahrenheit with partly cloudy skies developing. It's actually a nice recovery setup heading into Thursday. Now, Thursday is where things get really interesting. Fifty percent chance of showers and thunderstorms after 2 PM, but here's the big story. We're talking about a high near 90 Fahrenheit with heat index values reaching up to 96 Fahrenheit. That's humidity mixed with heat, and it's going to feel absolutely stifling out there. You're going to want that ice coffee. Let's talk about our Weather Playbook segment today. I want to break down heat index for those wondering what that actually means. When meteorologists talk about heat index, we're measuring what the temperature actually feels like when you combine the actual air temperature with humidity levels. Your body cools itself through sweat evaporation, right? When humidity is high, that evaporation process slows way down, so you feel hotter than what the thermometer says. It's why 90 degrees in the desert feels totally different than 90 degrees here in New York. Science! Here's your three-day snapshot. Thursday through Friday brings us scattered thunderstorms with highs near 90 Fahrenheit. Saturday is looking gorgeous, sunny skies with a high of 85 Fahrenheit. Sunday cranks back up to 89 Fahrenheit and sunny. Basically, you've got a small weather window Friday and Saturday before things heat back up. Make sure to subscribe to the podcast for more forecasts. Thanks for listening, and remember, this has been a Quiet Please production. Learn more at quiet please dot ai.

Today is the global launch day for Resonance — a book six years in the making, written and rewritten three times, nearly lost to financial collapse, and finally cracked open in a four-month creative retreat overlooking treetops in Austin, Texas. In this episode, Michael doesn't perform triumph. He reflects on what the journey actually cost: the allies who didn't show up, the editor who quit, the gap between the wedding you romanticize and the marriage you didn't fully reckon with. And then he tells you a story. About a leadership training where he declared, in front of a room full of people, that he would sing "Total Eclipse of the Heart" in public — loud and proud — within a month. About a spontaneous flight to Buenos Aires with no plans and a freshly downloaded Airbnb account. About a border crossing no cab had ever made. About arriving in Chilean Patagonia as the sun set over glacier lakes. And about the moment, in the middle of all of it, when the radio played exactly the song he had promised to sing — and he got out of the van, and he sang it. What followed — gauchos, a sunset, fifty horses released to pasture, and a silence he calls the most beautiful of his life — is not a metaphor for resonance. It is resonance. This is an episode about what happens when you stop waiting to be ready and start singing your song. Michael Trainer has spent 30 years learning from Nobel laureates, neuroscientists, and wisdom keepers worldwide. He's the author of RESONANCE: The Art and Science of Human Connection (March 31, 2026), co-creator of Global Citizen and the Global Citizen Festival, and host of the RESONANCE podcast.Featured in Forbes, Inc, Good Morning America. Follow on YouTube

Amanda Knox is a multi-linguist and self-described word nerd. In this essay she explores the fascinating world of untranslatable words, the ones that exist in some languages but not others, and what they reveal about the cultures that invented them. And at the end of it all, Amanda makes one of her own. Learn more about your ad choices. Visit megaphone.fm/adchoices

Reciting an excerpt from his poem, “Fifty-Eight Faces of California Spring,” Pulitzer Prize–winning writer and translator Forrest Gander travels through California's many counties to offer a geologic atlas of this vast region in spring. Speaking the language of rock—alluvium, quartzite, sandstone, jasper—these field notes give a glimpse of the cycles that continually play out amid apparent stillness, the always-present change hidden in the swathe of deep time. Read the full poem. Discover our latest print edition, Volume 6: Seasons.Credit: Ryan Molnar / Connected Archives

I think this was one of my most enjoyable dialogues in our What's new series. Maybe Sabine and I are getting more used to each other's cadence and interests or maybe it was the subject matter. Either way, I think you will find this to be a fascinating and provocative discussion of science at the forefront, and at the not-so-forefront, because that science is interesting too!We began our discussion describing a new finding of a Giant Ring of galaxies billions of light years across in the sky. The key questions are: Is it real? And is it surprising? We both have slightly different takes on this.Next we described a new measurement of the strength of gravity on scales from 80 to 800 million light years in distance. And guess what? Gravity falls off just like Newton predicted! This may seem like a big yawn, but one of the most popular models that claims to do away with dark matter would imply that Gravity would fall off differently on these scales. Does this new result kill that idea? Stay tuned.Microsoft, which has cried wolf a number of times so far when it comes to something called Majorana qubits as the basis of a new viable quantum computer just published a new paper claiming they finally have it. Sabine and I discuss why we are both still skeptical, but why the effort is worth it.Next, CERN, the large European particle physics laboratory, and the world particle physics community seem to have converged on plans for building a huge new accelerator in the current CERN site.. this time involving an underground ring 91 km in circumference, in which electrons and positrons would collide to explore the detailed properties of the Higgs particle. Is the effort worth it? Again, Sabine and I have slightly different takes on this.Fusion power, which we have talked about in a number of earlier episodes, continues to tempt humanity with the promise of unlimited energy. Many people, myself included, have tended to argue that fusion seems to be 25 years in the future, and may always be 25 years in the future. But many new efforts are underway, so who knows. Unfortunately, a group of economists has analyzed fusion in the context of other large energy programs and have argued that even if we can achieve it, it may not be as economically viable as many claim. Finally, one day Richard Feynman went to a Thai restaurant with his young companion Ralph Leighton, and wondered what he should order. Should it be the same old dish he loved or something new. An equation filled napkin later, and he had the answer. Fifty years later some cognitive scientists resurrected Feynman's napkin and explained it, and argued it might have important implications in other social situations. Such is the power of science.As always, an ad-free video version of this podcast is also available to paid Critical Mass subscribers. Your subscriptions support the non-profit Origins Project Foundation, which produces the podcast. The audio version is available free on the Critical Mass site and on all podcast sites, and the video version will also be available on the Origins Project YouTube. Get full access to Critical Mass at lawrencekrauss.substack.com/subscribe

Solar & storage pioneers Solar Design Associates share 50 years of firsts on the Clean Power Hour. They put solar on the White House in 1979 and built the first community solar garden in America. Haskell Werlin and Steven Strong trace solar's fall from $16 to $1 per watt, explain why the battery cost curve is following the same path, and break down what the ITC-free era means for developers.Solar and storage pioneers Solar Design Associates have been designing solar energy systems since 1974, accumulating firsts from the Carter-era White House installation to the first true community solar garden in the United States. Haskell Werlin, Vice President of Business Development, and Steven Strong, Founder and President, join Tim Montague on the Clean Power Hour to trace 50 years of solar industry evolution. Solar pricing fell from $16 per watt for satellites to $1 per watt for ground mounts today, and Haskell confirms the battery cost curve is now following the same downward path, with Texas leading the country in solar and battery installations. This episode covers landmark projects, including the Bullit Center in Seattle and the Harvard community solar garden, alongside a direct assessment of what the residential ITC removal means for project economics through 2028 and beyond.Here is what you will learn from this conversation about 50 years of solar storage pioneers and the battery transition ahead:You will learn why Haskell argues Texas, not Hawaii, is now leading the country in solar and battery installations after transforming the ERCOT grid from fossil fuel dependency to firm base load power.Find out how the first true community solar garden in the US, a 542-kilowatt ground mount in Harvard, Massachusetts required a statewide home rule petition to resolve a property tax classification dispute with the local assessor.Understand how the Bullit Center in Seattle, described by the New York Times Architectural Review as the “Most sustainable commercial building in America,” achieved 100% energy offset in one of the least sunny major cities in the US.Find out how Solar Design Associates put solar on the White House under President Carter in 1979, with Steven Strong on the roof for the dedication ceremony, and were called back under President George W. Bush in 2006 to install solar on the pool and cabana, spanning two administrations and three decades. Find out how Solar Design Associates has never exceeded 20 employees in 50 years, why hiring graduates with no prior solar experience is a deliberate strategy, and what Haskell says about the companies growing fast and falling hard.Fifty years ago solar panels powered satellites because nothing else could reach them, and the technology now costs $1 per watt for ground mounts, a cost collapse driven by German feed-in tariffs, and Chinese manufacturing scale. The battery industry is now following the same path solar took from satellite technology to mass market infrastructure, with the same forces of policy, manufacturing scale, and early adopter projects already in motion. Professionals watching this episode are standing at the same inflection point the solar pioneers of 1974 stood at, with the advantage of knowing exactly how this story ends.Connect Steven Strong, Haskell Werlin Haskell Werlin: https://www.linkedin.com/in/haskell-werlin-1a21383/Steven Strong: https://www.linkedin.com/in/steven-strong-3309894/Solar Design Associates: https://solardesign.com/ Support the showConnect with Tim  Clean Power Hour  Clean Power Hour on YouTubeTim on TwitterTim on LinkedIn Email tim@cleanpowerhour.com Review Clean Power Hour on Apple PodcastsThe Clean Power Hour is produced by the Clean Power Consulting Group and created by Tim Montague. Contact us by email:  CleanPowerHour@gmail.comCorporate sponsors who share our mission to speed the energy transition are invited to check out https://www.cleanpowerhour.com/support/The Clean Power Hour is brought to you by CPS America, maker of North America's number one 3-phase string inverter, with over 6GW shipped in the US. With a focus on commercial and utility-scale solar and energy storage, the company partners with customers to provide unparalleled performance and service. The CPS America product lineup includes 3-phase string inverters from 25kW to 275kW, exceptional data communication and controls, and energy storage solutions designed for seamless integration with CPS America systems.  Learn more at www.chintpowersystems.com

Become a supporter of this podcast: https://www.spreaker.com/podcast/arroe-collins-unplugged-totally-uncut--994165/support.

The Bay Area Video Coalition has had a big impact on local media for half a century. Now it looks to the future. Then, the story of a Liberian immigrant's first encounters with American life. From Liberia, to a pioneering Oakland dance company.

LumineuseSexySolaireSensuelleFragileTourmentéeMélancoliqueInsaisissableInoubliableVous aussi, vous pouvez ajouter des adjectifs pour décrire Marilyn Monroe !!Une femme, une légende, une icône qui est née en 1926.C'était il y a 100 ans. Tout pile.Pour l'occasion, il fallait bien un épisode SPECIAL VIP Avec plein d'infos sur : sa jeunesse, ses origines, ses 3 mariages, ses 3 divorces, ses plus grands rôles, ses plus belles scènes, son "POU POU PIDOU", son sourire, son statut de sex-symbol, sa mort étrange et mystérieuse à l'âge de 36 ans.Pour en savoir plus, une seule adresseLe podcast FIFTY STATES !Hébergé par Audiomeans. Visitez audiomeans.fr/politique-de-confidentialite pour plus d'informations.

October 1975, the partial remains of a 73-year-old Tucson man named William Reginald Sipfle were found in a landfill near Ryan Airfield, with no identification, no missing person report, and no answers for the family he left behind. Fifty-one years later, forensic genealogy and DNA technology cracked open the cold case and pointed investigators directly at Sipfle's own stepdaughter, Carol Ann Beall, now 79, who prosecutors allege killed him and collected up to six hundred thousand dollars from his pension in the decades that followed. This episode breaks down how the case went cold, how modern forensic science brought it back, and what this arrest reveals about the long reach of justice and the extraordinary tools now available to investigators working crimes the system once had no way to solve.

n October 1975, the unidentified remains of a Tucson man were found near Ryan Airfield with no missing person report, no leads, and no justice for a family left without answers. Fifty-one years later, investigators armed with forensic genealogy technology traced the victim to 73-year-old William Reginald Sipfle and identified his stepdaughter Carol Ann Beall, now 79, as the prime suspect, allegedly collecting up to six hundred thousand dollars from his pension the entire time. This episode breaks down how the cold case was reopened, how DNA changed everything, and what this arrest means for the growing number of decades-old crimes now being solved through modern forensic science. IAB Tags: Crime/True Crime, Law/Government/Legal, Science, News/Current Events, Society True Detective Podcast Title: 51 Years in the Dark: How DNA Pulled a Killer Out of a Cold Case and Into a Courtroom A body dumped in a landfill in 1975, a victim who had no name for decades, and a suspect who allegedly spent over half a century collecting a dead man's pension, this is one of the most chilling cold case resolutions in recent memory. Investigators used forensic genealogy to identify the victim as William Reginald Sipfle and zeroed in on his stepdaughter Carol Ann Beall, now 79, as the woman prosecutors believe killed him and buried both the body and the truth for 51 years. This episode goes deep into the investigative trail, the forensic tools that made the breakthrough possible, and the haunting question of how someone lives an ordinary life while carrying a secret that dark for that long.Sonnet 4.6

People see the amount of your offering, but God sees the deep, hidden sacrifice of your heart. He isn't a legalistic judge tracking percentages to crush you with guilt; He is the loving Father who looks at your fifty pesos, your jeepney fare, and your tired hands, and declares that your trust is worth more than all the gold in the world. #FULLTANKwithBroBo​ #FULLTANKwithBroBo2026​  #BoSanchez​ #GodSeesYourSacrifice​ #Mark12​ #TheWidowsMite​ #RadicalGenerosity​ #FaithInAction​ #SpiritualMaturity​ #TrulyRichMindset​ #GivingWithLove​ #Grace​--- PS. Do you want to grow your finances but don't know how? For the past 18 years, I've received a lot of “thank yous” from so many TrulyRichClub members because once upon a time, they were stuck in their finances but through the club, they learned how to invest and they're on their way to financial freedom. If you want to grow your finances and reach your financial dream, go to www.trulyrichclub.com now or go to trc.ph/events to join our upcoming seminars!To support my mission work, click this link now! http://BuyMeACoffee.com/brotherbosanchez

I thought I was being smart. A slick postcard showed up in my mailbox — synthetic oil change, tire rotation, new wiper blades, a hundred bucks. Done deal, right? Wrong.What I walked into was a fully systematized, corporate-engineered upsell machine disguised as a friendly neighborhood auto shop. And I'm not guessing — I looked it up on Claude AI and found the exact marketing company they're paying good money to run this little circus.They shot 50 pictures of my car. Fifty. Sent me an email at lunch with a photo essay of everything "wrong" with my vehicle. Add it all up, and I could've bought a new car. No, thank you.I watched them reel in a poor lady for $1,200 before I even got my keys back.It's called the service lane sales process, folks. And it's not just auto shops — lawn care, pest control — they're all doing it.Old-fashioned? Maybe. Broke? Nope.#DailyGrateful #ConsumerAlert #AutoRepairScam #ServiceLaneSales #UpsellEverything #SystematizedSelling #SmartConsumer #MichaelCrose #KnowledgeIsProtection #WatchYourWallet

Russian hottest electronic dance music duo Swanky Tunes delivers you a weekly radio show. Thrilling 60 minutes of their biggest tracks and hottest bootlegs are waiting for you. From Russia with love!   Swanky Tunes - SHOWLAND 620 01. Aaron Sevilla, HMR (kz), Fifty(kz) - Mustafa 02. Monkey Safari, NenaHalena, Wassolou Benkain - Hail From Mali 03. Badrops, Wrigley, BUGRA BAKTIR - Dilemma 04. Josh Charm - Ah Ye Lo 05. Mason Clark - Verde 06. Sunnery James & Ryan Marciano & Piero Farho - Cor Silvae 07. Afrojack - No Beef (&friends Remix) 08. AN21, Pretty Output - Vem Dancar 09. Lohrasp Kansara, Afro B, DJ Power - Joanna 10. DAVEE, Kill Them With Colour - Bailar 11. Mydoz - Chugga Riddim 12. Aaron Sevilla, Oliver Gil, PRivas - Esta Cobardía 13. Alessio Bagordo & Marco Bartolucci - Ibiza 14. Fer BR - Help Us! 15. Shapov, Swanky Tunes - Wild & Free 16. R-CHY - The Hype 17. Jaden Bojsen - ERICA

Leigh Martin was awarded the King's Service Medal for services to brass bands. He joins Emile Donovan.

Fifty thousand years ago, Neanderthal artists in Ice Age Europe painted symbols and handprints deep inside caves, leaving behind some of the oldest known art on the continent. These discoveries are transforming how we understand our closest human relatives.Today, Tristan Hughes is joined by Genevieve von Petzinger to explore the fascinating story of Neanderthal art. What kinds of images did Neanderthals create? What did these markings mean? And how might their artistic traditions have influenced the first groups of Homo sapiens who later arrived in Europe?MOREHomo Sapiens v NeanderthalsListen on AppleListen on SpotifyLascaux Cave: Ice Age ArtListen on AppleListen on Spotify We're going on *TOUR* to Australia and New Zealand! - grab your tickets here.Presented by Tristan Hughes. Audio editor is Aidan Lonergan. The producer is Joseph Knight. The senior producer is Anne-Marie Luff.All music courtesy of Epidemic SoundsThe Ancients is a History Hit podcast.Sign up to History Hit for hundreds of hours of original documentaries, with a new release every week plus early access ad-free podcasts. Sign up at https://www.historyhit.com/subscribe.  Hosted on Acast. See acast.com/privacy for more information.

In this episode, Nick speaks with Von Decarlo about the importance of authenticity, resilience through grief, and embracing aging with confidence.

This episode contains descriptions of murder, mob violence, historical racial violence, and the execution of a convicted killer. If you need to skip this content, advance past the 18:00 mark. Support resources are listed at the end of these notes.This EpisodeSeason 40: Fifty states, fifty forgotten crimes, America's 250th year. Episode 9 covers California and Alabama — two cases, two communities that looked at the legal system and reached for something uglier. October 10, 1890. A woman named Helen Riche is playing cards in her tavern near a California quicksilver mine when ten men in flour-sack hoods crash through the door. She does not run. She reaches up and rips the mask off the nearest man's face, and in that single act she solves the crime that is about to kill her. This is true crime history from the American frontier, and the legal system that followed would leave you cold.December 1888, Birmingham, Alabama. A railroad engineer named Richard Hawes boards a streetcar with his eight- year-old daughter May. He gets off with her at East Lake. He gets back on alone. The body of a young girl is found floating in the lake the next morning. On the same day, Hawes is across the state line getting married. When Birmingham finds out, two thousand people march on the jail.The VictimsHelen Matilda Riche ran the Campers' Retreat tavern on sixty-two acres near the Bradford quicksilver mine, three miles south of Middletown, California. We do not know where she was born or how she came to run a mining-camp saloon in hard hill country — the historical record is thin on her life before October 10, 1890. What it preserves is a woman who managed a clientele of mercury miners in one of the most physically dangerous industries of the era. She was shot five times during the raid. She fought back, reaching for her husband's .44 Winchester with five bullets already in her body. She died four days later. Her husband J.W. Riche died less than three months after her, his own bullet wound never having healed.May Hawes was eight years old when her father took her on a one-way train ride to East Lake on the evening of December 3, 1888. She had been doing the work of a parent since she could walk, looking after younger siblings in a household already coming apart. She was laid out for public identification at Lockwood & Miller's Funeral Parlor in Birmingham, unidentified for a full day. A local butcher recognized her. May, her mother Emma, and her six-year-old sister Irene — all three murdered by Richard Hawes — lay in an unmarked grave at Oak Hill Cemetery in Birmingham for more than 135 years. In April 2024, they finally received a headstone.The CrimesThe Lake County White Cap raid followed personal grudges that had been tightening for months. Blackburn, a mine foreman, had been thrown out of the Campers' Retreat after a brawl with the bartender Fred Bennett. Others in the group had boundary disputes, cattle quarrels, neighborhood debts to settle. They put flour sacks over their heads and called it a community morality action — the Whitecapping movement had spread from Indiana through the Southern states and into California by 1890. The plan was to flog Bennett and run him to the county line. Helen Riche unmasked Henry Arkarro the moment the men crashed through the door, and the plan collapsed into gunfire.Richard Hawes murdered three members of his own family to clear the way for a new marriage. Emma and Irene Hawes were found bound with curtain cord and weighted with railroad iron curve-braces in a Birmingham lake on December 8, 1888 — the same day a mob of approximately 2,000 people converged on the Jefferson County Jail demanding to hang him on the spot. Sheriff Joseph S. Smith fired into the crowd. Ten men were killed. Approximately thirty were wounded. The historical murder case that followed Hawes would take fourteen more months and a formal trial to reach the same conclusion the mob wanted.The Investigations and Legal OutcomesIn California, ten men were arrested within days. The mining community was small; Helen Riche had identified one attacker herself. The trial opened February 6, 1891, in Lakeport — *People of the State of California v. B.F. Staley et al.* Four men were convicted of second-degree murder: Blackburn sentenced to twenty-five years, Staley and Cradwick to twenty years each, Osgood to twelve years. All four were released from San Quentin within approximately three years. The Governor had commuted Blackburn's sentence to ten years following an extensive lobbying campaign. Three years, for a home invasion that killed two people.In Alabama, Richard Hawes was tried beginning April 22, 1889, before Judge Samuel Greene. The prosecution built the case around May's murder — the strongest evidence available, though entirely circumstantial: eyewitness testimony placing father and daughter on the streetcar together, and only the father returning. The jury deliberated fifty-five minutes. Death. After multiple appeals to the Alabama Supreme Court, all denied, Richard Hawes was hanged by Sheriff Smith on February 28, 1890 — the same man who had fired into a crowd to keep him alive for this moment. Hawes wore a geranium in his lapel. The gallows were built by a man who had served on his jury.Historical ContextBoth cases sit at a specific American intersection: communities losing faith in institutional justice and reaching for extralegal violence, with consequences that fell hardest on people who had nothing to do with the original grievance. The Whitecapping movement was already documented across Indiana, Tennessee, and Mississippi before it reached California. In Alabama, the Birmingham riot of 1888 killed ten bystanders, including Maurice Throckmorton, thirty-three, the city's postmaster, who was reportedly trying to calm the crowd when he was shot. The legal system delivered the outcome the mob demanded — it just took fourteen months and cost ten additional lives to get there.California's legislature responded to the broader wave of hooded vigilantism during this period with enhanced anti- vigilante and anti-mask statutes. For the Hawes case, Fannie Bryant — the family's cook and a key witness for the prosecution — was herself sentenced to death for allegedly aiding Hawes. She died in a prison riot before the sentence could be carried out. Her actual level of involvement remains contested. She was a Black woman in 1880s Alabama, easily targeted by a system that offered her no protection.Our Sponsors:* Check out Kensington Publishing: https://www.kensingtonbooks.com* Check out Mood and use my code SHANE for a great deal: https://mood.comAdvertising Inquiries: https://redcircle.com/brandsPrivacy & Opt-Out: https://redcircle.com/privacy

You've done the work. You cleaned up your diet, fixed your sleep, addressed your metabolic health — and something is still wrong. Brain fog you can't explain. Fatigue that won't move. Weight creeping up for no apparent reason. Personality changes your family notices before you do.Nurse practitioner Ally D'Amico spent 14 years watching patients spiral through specialists, collect diagnoses like fibromyalgia and chronic depression, and never actually get better — until she stumbled onto the thing nobody in conventional medicine was looking for. One in four people carry a genetic predisposition that prevents them from clearing a common environmental toxin. Fifty percent of homes contain it. And the insurance system doesn't even have a billing code for it.If you've ruled everything else out, or if you're still not all the way back — this conversation is worth your time.BIG IDEAMold toxicity masquerades as fibromyalgia, chronic depression, brain fog, and metabolic dysfunction — and one in four people are genetically unable to clear it on their own.Ally's LinkedIn: https://www.linkedin.com/in/ally-d%E2%80%99amico-anp-bc-4b596015/Website: https://www.moldco.com/MoldCo's LinkedIn: https://www.linkedin.com/company/moldco/IG: @themoldcompanyX: @themoldcompanySend Dr. Ovadia a Text Message. (If you want a response, you must include your contact information.) Dr. Ovadia cannot respond here. To contact his team, please send an email to team@ifixhearts.com Order at Amazon: Stay Off My Kitchen Table  Like what you hear? Head over to IFixHearts.com/book to grab a copy of my book, Stay Off My Operating Table. Ready to go deeper? Talk to someone from my team at IFixHearts.com/talk.Ready to take control of your health?  Grab Dr. Ovadia's brand new book Stay Off My Kitchen Table now! This isn't just another diet book; it reveals why it's not just what you eat, but what your body actually absorbs that determines your health.If you're struggling with low energy, stubborn weight, or feeling like “healthy eating” isn't working… this book shows you exactly how to fix it.Learn how to reset your gutEliminate hidden foods sabotaging your progressUnlock real energy, metabolism, and longevityDon't wait until it's too late. Take action today. Get your copy of Stay Off My Kitchen Table now.Learn More:Take Dr. Ovadia's metabolic health quiz: iFixHearts Dr. Ovadia's website: Ovadia Heart HealthTheme Song : Rage AgainstWritten & Performed by Logan Gritton & Colin Gailey (c) 2016 Mercury Retro RecordingsAny use of this intellectual property for text and data mining or computational analysis including as training material for artificial intelligence systems is strictly prohibited without express written consent from Dr. Philip Ovadia.

Hello Denimheads and welcome to the 50th episode of The Sons of Selvedge Podcast (it's exciting to write that we've achieved FIFTY episodes), where a group of friends get together to talk about denim, boots, heritage clothing and the makers who make them. In this episode founders Andy, David, Illya, Kevin, Lex and Ricky - but not Tom :( get together to share what they've been up to. Please like this interview, and subscribe to us wherever you enjoy our content: YouTube, Spotify, Apple, Google or Stitcher. Check us out on Instagram @‌sonsofselvedgepodcast. Give us a shout with any questions, or if you'd like to join our Discord Server. Photography by @‌illcutz.

Ready to kick off your healthiest summer after 50? In this episode of FAT LOSS SCHOOL, Amy shares practical belly fat loss tips for women in menopause, explains why the 14-Day Belly Blast is the perfect jumpstart for fat loss and healthy habits, and also invites you to join Vacation Bible School for Fabulous-over-Fifty Women. Learn how strength training, strategic nutrition, core workouts, and mindset shifts can help reduce belly fat, boost metabolism, and create lasting results—while having fun along the way!  Join the June 1-14 BELLY BLAST here:  https://fasterway.com/products/belly-blast?aid=AMYBRYAN  Join the free Facebook community to take part in Vacation Bible School for Fabulous-over-Fifty Women here:  https://www.facebook.com/groups/fasterwaywithamybryan  Join my next foundational 6-week FASTer Way to Fat Loss® class for women over fifty here:  https://www.fasterwaycoach.com/AMYBRYAN 

From heroin to healing to hosting.In 1994, Eric Zimmer walked through the doors of Maryhaven, an addiction treatment center in central Ohio, as a client seeking help for heroin addiction. Today, he is an author, teacher, speaker and the creator of "The One You Feed" podcast.Fifty million podcast downloads later, Zimmer shares what recovery really teaches.Zimmer has a new book called How A Little Becomes A Lot: The Art of Small Changes for a More Meaningful Life.He joins All Sides to talk about his new book. He will also discuss his journey from being a client at Maryhaven to returning decades later in recovery and now partnering with them around his new book.Guests:Eric Zimmer, author, How a Little Becomes a Lot/host of "The One You Feed" podcast

We're announcing AIEWF speakers this week! Take the AI Engineering Survey!Today's guest Ethan first joined us for the LS Paper Club as the lead on NVIDIA Cosmos World Model, but then joined xAI and built Grok Imagine in 3 months:He comes back on Latent Space with some nuclear hot takes: that Video Models primarily get their intelligence from LLMs, not from training on video data, and that the next frontier for truly interactive, realtime, long-horizon world models is to work on LLMs (perhaps Interaction Models as well…)Put it this way: In the near term, the next Sora won't be a better video model, but a video agent.Generative Media may more closely follow the evolution of AI coding which went from focusing on one-shot output performance and cost, to multiturn reasoning and planning models for agents and systems that can plan, edit, test, debug, and submit PRs.At a certain point, coding models got so good that the only significant next step to improve performance was handling the orchestration of these models.Now as the performance of video models increases significantly across realism, consistency, & prompt adherence while becoming more cost efficient, the next evolution of video generation may also be systems that can plan, generate, edit, critique, and iterate across an entire creative task. In this episode, Ethan joins swyx and Vibhu to unpack what it actually takes to build frontier image and video systems: data, VAEs, diffusion transformers, audio-video alignment, inference speedups, and the hidden cost of storing and moving massive video datasets. From building NVIDIA's Cosmos world model to joining xAI as Grok Imagine was being built from zero to one, Ethan He has been at the center of some of the most important work in video generation, multimodal models, and real-time world models.We go deep on Grok Imagine, how a small xAI team shipped its first multimodal video model in three months, why iteration speed matters more than almost anything in model development, and why many of the biggest gains come from fixing tiny bugs in data and training pipelines. Flipbook: The future of VideomaxxingVideo agents are almost a sure bet to be the trend in the coming year. We end with a glance at what's beyond video agents:Flipbook caused a minor sensation this year when it was released, but most treat it as a fun demo. Ethan takes it very seriously — with the speed and cost of inference coming down every year, the future of custom video JIT UI is closer than you think. We talked about why videogen models may become the front end of AI, how generative UI could replace traditional HTML/CSS, why world models need to be real-time, interactive, and long-horizon, and why the future of video generation may depend more on language models and agents than on diffusion alone.We discuss:* Why fast iteration mattered more than meetings* Why small training bugs can drive huge model quality gains* Why coding models may make compute the bottleneck again* How image and video models are trained with synthetic captions* The role of VAEs and latent space in frontier video models* Why image models are the foundation for video models* The tradeoff between temporal compression and real-time interactivity* Flipbook, Neural OS, and the future of generative UI* Why future interfaces may go from user intent to pixels* The hidden cost of training video models: storage, egress, and GPU hours* How step distillation and consistency models (like OpenAI sCM) makes video inference orders of magnitude faster* Grok Imagine 0.9 and large-scale audio-video generation* Why audio-video alignment is harder than text-video alignment* Ethan's definition of world models* Reference-to-video, video extension, and long-context video generation* Why xAI's research communication undersells Grok Imagine* How xAI culture shaped the speed of development* AI watermarking, SynthID, and detecting generated media* Why prompt rewriting matters for video models* Grok Imagine Agent and the rise of video agents* Why language models may unlock better video generation* Robotics, physical AI, and embodied world models* Why Ethan left xAI and shifted focus toward LLMs* Self-managed context, memory, and the next frontier for language modelsEthan He* LinkedIn: https://www.linkedin.com/in/ethanhe42* X: https://x.com/EthanHe_42Timestamps00:00:00 Introduction00:01:25 From NVIDIA Cosmos to xAI00:03:24 Building Grok Imagine from Zero to One00:10:07 How Image and Video Models Are Trained00:18:53 Video Compression, VAEs, and Real-Time Tradeoffs00:22:10 Generative UI, Flipbook, and Neural OS00:32:10 The Cost of Training Large Video Models00:37:04 Distillation, GANs, and Fast Video Inference00:41:21 Audio-Video Generation and Grok Imagine 0.900:48:34 What Makes a World Model?00:55:51 Reference Videos, Long Context, and Video Memory01:00:11 xAI Culture, Research, and First-Principles Building01:09:45 AI Safety, Watermarking, and Prompt Rewriting01:13:10 Video Agents and AI-Assisted Creation01:27:32 Why Language Models Unlock Better Video01:31:15 Robotics, Physical AI, and Embodied World Models01:32:38 Why Ethan Left xAI01:34:16 Self-Managed Context and the Future of LLMs01:38:43 Ethan's Career Path and Closing ThoughtsTranscriptIntroduction: Ethan He, Latent Space, and the Path to xAISwyx [00:00:00]: We're here in the studio with Ethan He, most recently of xAI. Welcome.Ethan [00:00:10]: Thank you. Glad being here.Swyx [00:00:11]: We're also here with Vibhu. you were first coming to us or joining the latent space world because you were working on Kosmos at NVIDIA, and you did a paper. We loved it. you presented it as well, so thank you for doing that.Ethan [00:00:23]: I've actually, I also presented the MoEs twice at latent space.Swyx [00:00:29]: How did you actually hear about us? Did we reach out to you? Is that how it worked?Ethan [00:00:33]: No, actually, I-- the community. Like I realized, oh, there is this online community that people talk about AI and also learn from each other through papers every week through the Paperclip. It's very nice.Ethan [00:00:49]: I learned a lot.Swyx [00:00:49]: I think three years stop. We haven't stopped even on Christmas and New Years. many weeks I want to stop but it keeps going.Vibhu [00:00:58]: No, that was good. I think you had posted that you worked on a paper, and I was “Oh, very cool. We have Paperclip. Present then.”Vibhu [00:01:04]: But I might have reached out to you after.Swyx [00:01:05]: you-- because it's an amateur club, right?Swyx [00:01:08]: so it's very unusual and but we have sometimes paper authors come by and actually explain the paper. Today we just did, the poolside paper, which was apparently very good.Vibhu [00:01:18]: Came out yesterday.Vibhu [00:01:19]: pretty interesting, right? Fully open. They talk about everything, systems. So it's a good one. We'll, we'll recommend people to read it.Swyx [00:01:25]: Bring us up to speed on your transition to xAI, ‘cause I actually don't even know when you joined. just like tell the, tell the story about the sort of transition.From NVIDIA Cosmos to xAI: Scaling Video and World ModelsEthan [00:01:34]: Before xAI, I was working on Kosmos world model as in-- at NVIDIA. So Kosmos is, it's a giant video foundation models that can-- that aims to simulate the world and for-- it serves as a foundation of-- for all of the roboticists to build on top of. There, once I built the Kosmos one, I realized as this thing also has a scaling law similar to language model, we need to scale up the video models further. that's, that's why I realized I need to move to somewhere with much more compute resources. That's how ISwyx [00:02:13]: Than NVIDIA?Vibhu [00:02:14]: The GPU rich came themselves.Vibhu [00:02:19]: And timeline-wise, when was Kosmo? It was pretty early, right? It was open world model, open paper, everything.Ethan [00:02:25]: It was end of twenty-four.Vibhu [00:02:28]: End of twenty-four.Ethan [00:02:30]: Then at mid twenty-five, I moved to xAI. At that time-- I joined about the time when xAI was about to build video models and in multi-model models. There were no infra, no data, and no model, and it just-- as a few engineers, we built it in three months and released the first model, Grok Imagine zero point nine.Ethan [00:02:55]: And since then, I keep working on video models and move more from training and to post-training of the video models. For example, like a reference to videos, kind of like the cameo feature and, video extensions. And, before I left, I worked on a world model, leading a small team to focus on the real-time long horizon video generation.Building Grok Imagine From Scratch in Three MonthsSwyx [00:03:24]: Can you give like a rough roadmap of okay, you're on a brand-new team. Grok previously was only text, or they partnered with BFL for their image gen stuff. What do you-- what are the building blocks, right? You have compute, data you can procure somewhere. Like just what are like the sequence of things that people should think about when you're setting up a new team?Vibhu [00:03:43]: actually even deeper, not just data you can procure. You guys had to go through getting the data too, right? So you shipped it pretty fast, but yeahSwyx [00:03:51]: three months is likeVibhu [00:03:52]: From everythingSwyx [00:03:52]: actually like very surprisingly fast.Ethan [00:03:55]: One thing I say like thanks to my experience at NVIDIA, ‘cause first time when we were building Kosmos together, we built it, for about a year. So this is like the second time I do it. Roughly have an idea, what to do. I say the most important thing is the talent. Everyone were very strong and clever, very close with each other towards a common goal. So that speed up things a lot. So you reduce the communication bandwidth among people, and everyone can work towards the same goal. It's, it's like every day there's not that much meetings on the calendar, like maybe like a, like a sync a day, and after that it's, it's just all building. It was pretty fun at that time.Ethan [00:04:47]: And another thing is that xAI has very strong foundations of like data inference, model inference, and the supporting there can help the model develop a lot. When I look at, training models, I don't so actually the top important thing is like how many, how many iterations can you do, per day? and the more iteration can you do, you can, you can train the model much faster. So if you have very strong infra and you have a lot of compute, you can, you can train these models in very short period of time. That can give you a much larger buffer to, for errors, and it also gives you the opportunity to spot more bugs.Iteration Speed, Compute, and Debugging Model PipelinesSwyx [00:05:46]: What is an iteration? Is it like a few hundred steps or what are youEthan [00:05:50]: Let's say just the train-training the model, like from acquire new data and maybe design new algorithms and train a new model, maybe at smaller scale orSwyx [00:06:01]: So cycle time for like any hyperparam that you're searching.Ethan [00:06:04]: Cycle time and tune to like eval this model. Is this model better than my previous iteration?Ethan [00:06:11]: SoSwyx [00:06:11]: So it's like before you, someone had already set this up that you can iterate very quickly.Ethan [00:06:15]: I think the foundation there is extremely good forDeveloping and research models.Ethan [00:06:23]: And often I find is it-- this is kind of boring, but like a lot of the improvements does not come from new algorithms. It comes from finding small bugs here and there in the data pipeline, in the, in the model training pipeline. Those give, those give the biggest boost to the model quality.Vibhu [00:06:46]: It's interesting, right? So you say it's like small team, less communication bandwidth, but also a lot of quality is like find little bugs. It seems counterintuitive, right? You have a lot of people, you can iron out more of those, but it's interesting to see the other side, right?Swyx [00:07:00]: I also wonder, have you-- do you try using LLMs to look for bugs? I don't know.Ethan [00:07:05]: I remember at that time it was mid two thousand and twenty-five, so it's the coding model wasn't quite there yet. I remem- I remember like December two thousand and twenty-five, it was extremely good. Yeah, I've been, I've been using it at that time. It's, it's helpful. sometimes it produce codes that are kind of difficult to maintain, even though like the first time it built something extremely fast. But it gave the, like a spaghetti code, thousands of lines that I couldn't maintain, and the LLM itself couldn't figure out what's, what's wrong and how to improve on top of it. But now I find it much better. Yeah, I want to bring up another point here is now coding models are much more efficient and can help us implement stuff much faster. Compute might become a bottleneck again because previously, like if you want to train a new model, say you want to generate new synthetic data and then or write a new algorithm, it might take a few weeks. And during that period of time, you don't-- you might not have experiments to run. But now you can build that thing within a few hours, then you can immediately train a model.Ethan [00:08:24]: Now you have to have enough compute to try all of the ideas. So compute might be the bottleneck of iterating speed again.Swyx [00:08:36]: yeah, I actually, honestly, I think it's like kind of a stressful job because you're “Well, I should be trying everything, and if I'm not, then I'm not doing my job well.”Vibhu [00:08:48]: there's also the stress of you're eating thousands of GPUs per hour, which is very expensive and, compute can go to other researchers.Swyx [00:08:56]: You got the daddy Elon toVibhu [00:08:57]: You got daddy Elon.Ethan [00:08:59]: It wasVibhu [00:09:00]: But there's still finite amount of compute, like you want to use it, you want to use it well, you want more of it.Ethan [00:09:06]: That was quite stressful indeed. Yeah, I think one thing is the-- with coding models now, like a lot of these jobs can be automated, which is much better. A second, it's a, it's a marathon, so you got to maintain good health and, a regular schedule.Vibhu [00:09:28]: It's, it's hard to hear that when you shift from zero to nothing in two months.Swyx [00:09:32]: and, I think obviously the culture at xAI is very famously, people work very hard. one thing I did want to dive into, in our-- in the notes that you, that you sent ahead of time, you had specific comments about the cost of Video Gen training. presumably this is on the Colossus-1, right? the two hundred megawatt cluster. Any whatever you want to just share on that.Vibhu [00:09:54]: I think there's, there's three things we're talking about, right? So there's Video Gen, there's also the Image Gen model that you put out. Do you want to like complete the, okay, so zero to one, you have a few months. Just what are the stages of create Image Gen model?Swyx [00:10:06]: Oh, yeah, maybe I got distracted.How Image and Video Models Are Trained: Synthetic Captions, Tokenizers, and VAEsVibhu [00:10:07]: Sorry. and then, from there's Video Gen, there's Audio Gen. Would love to get into those next. But what is that first few months like? So small team, a lot of bugs, iterations, but what does it look like? Do we take something off the shelf? Do we just get data compute? What's, what's the few months like? How do you go to state-art Image Gen model? How do you just start?Ethan [00:10:28]: I cannot comment specifically how xAI did, but it's, it's a quite standard process. I can draw some, examples from Cosmos. So mainly it's building a video model, you actually need to build a image model first. And building these two models, the data you need is a hundred percent synthetic pair of language and image or language to video. Because on the, on the internet, actually, the videos don't naturally associate with text. So you can say, oh, like on YouTube, you have the title and you have the description and the commentsSwyx [00:11:11]: TitleEthan [00:11:11]: of a video, but usually they're not relevant to the video itself. And say maybe like the video is a natural scene of mountains or something, and the title is, I'm so happy today.Ethan [00:11:26]: So they have they have no correlation at all. So the first step is to, you have to generate synthetic pair of language with the videos. So you gather videos from the internet, and you use a VLM to caption the videos. So that part, here's a question, like how do you, how do you gather VLM to begin with? So if there's noSwyx [00:11:55]: You, so you fuse the model, right? LikeEthan [00:11:57]: Say if there's no like VLM exists, like how do you generate the text to the beginning, right? It's, it's impossible.Swyx [00:12:04]: I see.Ethan [00:12:05]: In the beginning, it's like you ask human to describe the video as detailed as possible.For example, you ask them to describe everything, like all objects, all characters, and all interaction and dialogues in the, in the videos. So that's in the protocol of Cosmos labeling. We require the objective we give to the labelers was that you have to describe the video as detailed as possible, such that a blind person hears a blob of text can reconstruct what the video is like from their head.Swyx [00:12:43]: Video or image? You're talking about images.Ethan [00:12:44]: Video or image, either one of them.Vibhu [00:12:47]: This was pretty common when we went from clip and DALL-E, right?Vibhu [00:12:51]: It's all training on really detailed captioning of images. So same is applied to video, but insteadEthan [00:12:57]: same appliedVibhu [00:12:57]: of using multimodal model to pass in video images and write rich descriptions, you can alsoSwyx [00:13:04]: I think there's this traditional perspective of supervised, or, very highly human curated thing. I feel like there's a unlock with unsupervised, right? Where like you have enough to bootstrap that you can just throw common corpus on it or, whatever. like unsupervised vision and language pairing, right? Like where you just have, interspersed image and text and it just learns. To me, that is the VLM breakthrough that is different from the clip, different from the LM era.Ethan [00:13:36]: It's interesting to see that you kind of need both data.Ethan [00:13:41]: For example, for theSwyx [00:13:41]: You need it to bootstrap it up. YeahEthan [00:13:43]: for the generative model training, there's also usually like a small percentage of unlabeled data. So the model is instructed to generate a video without any text instruction. That can also help the model generalize. So after this stage of generative synthetic pair, so, one important common step is to train a compressor or a tokenizer of the image or videos. So because, if you train-- If you can technically, theoretically train image or video models on pure pixels, but the problem is that the, it's, it's a lot of tokens. So like one image, it's, a thousand by a thousand, it's like one million tokens, one million pixels. It's impossible to train transformer on that. So it's, you need to train a tokenizer, which can go from image to latent space and latent space back to image.Swyx [00:14:45]: That's why we named the podcast.Swyx [00:14:48]: But, basically, you're talking about vocabulary science.Ethan [00:14:50]: so vocab.Swyx [00:14:51]: And so, what is, what is imp-- like a million is impossible?Ethan [00:14:54]: In generative models, the vocab is continuous. It's a continuous space. We can think about like you map an image to a vector. It's a, it's a fixed length vector. It's sixteen or forty-eight, something like that. And then you map that vector back to the image space. And the mapping is, has-- The mapping is patch-based. So you say you haveEthan [00:15:22]: a sixteen by sixteen patch and you match, you map that patch of pixels into this latent space.Swyx [00:15:29]: We've covered thisVibhu [00:15:30]: This is like the vision transformersSwyx [00:15:32]: VAEs,Ethan [00:15:33]: VAEs.Vibhu [00:15:34]: You basically compress your input, you do your generation, you're reasoning all that generation in smaller dimension, and then you project back out.Swyx [00:15:43]: VAE is a form compression, but I think the for me, the patching thing is from VIT, right?Ethan [00:15:48]: You can make those.Swyx [00:15:49]: Literally the, yeah, the paper is titled like sixteen by sixteen is all you need. something like that. and then I think also, people make a lot of comparisons with this kind of patching with convolutions.Swyx [00:16:02]: Which is you're, you're kind of re- reconstructing the old paradigm with the new.Ethan [00:16:05]: Actually, in VAEs, there are, there are both convolution networks and transformers. You can actually do both.Ethan [00:16:14]: After this VAE, so what you've got is you've got latent space tokens and you've got the language tokens. So now the training of the diffusion transformer, usually generative models use diffusion transformers. It is actually quite standard. It's, it's very similar to how you train a language transformer models. It's not that much difference. It's just the tokens, the visual tokens in, visual tokens out. The only difference is there's a denoising process. So you train the model to unmask some of the noise. So you add, you add random noise to the visual tokens, and then you train the model to remove those noise to generate the clean tokens. Any inference, the model can iteratively remove noise from a hundred percent noise.Swyx [00:17:12]: And then there's also, to speed things along on the tech tree of diffusion, there's CFG, and then there's, there's also, latent diffusion that, there's, there's someone in there. I think, somewhere along the line, obviously, like stability and all these other guys, pioneered a lot of this, architecture. I don't know if you want to get into that or just, or do the video side up to you.Bootstrapping Video from Image Models and Temporal CompressionEthan [00:17:37]: After you train such model, such image model, the reason it's a, it's a foundation for video models is that image models are cheaper to train, and they have much denser connection between language and text. So, sorry, language and images. For example, you train a billion, you train on a billion images, and there's a mapping from the text to the image. And the cost to train the same, like the, a billion, a billion text to a billion videos, that's much more expensive because videosNaturally have more tokens than images. Because the diffusion models, their understanding of, language purely come from this mapping. So if you don't have enough mapping, so if you only train on like a ten million videos or something, there-- you might not see enough language tokens in your training, so your model does not understand human intention enough. So that's why you really-- you train-- you first train this image diffusion models, and then you bootstrap the video model from there.Swyx [00:18:53]: One thing I did want to ask, because I-- actually, I think you're, you're the first per-- video model person I've ever talked to, I think. we've, we've like talked to Luma and all those folks. There's all these tricks in video compression where basically frame by frame there's not that much difference, so actually you don't have to regenerate or save the whole frame, right? but I think MP4 compression or something else like that.Swyx [00:19:16]: is it tempting to use that? Or as far as I can tell, everyone just treats it as, “No, we would just generate every frame.” Is that roughly the state-art?Ethan [00:19:27]: There are a few different approaches. Let's say first, like you want to just directly use MP4 compression and use that as the tokens for the transformers to train, right? So people actually have tried that, but the main challenge is the latent space for the MP4 tokens were not, were not very comprehensible for the models. It's, it's extremely hard to train on that. And there's aEthan [00:20:01]: So that's why they created VAEs, which creates more continuous, latent space, so the models can understand that latent space and learn from it much easier. Even within the VAEs, there are different difficulties of the latent space. So you can imagine something the simplest, the most naive VAE is like you have an image, and you just shuffle all of the images into a, into a vector. So you don't need to train any VAEs, right? But that latent space is extremely hard for models to train on top of. That's why there are some debate on like how do you compress the tokens. So you mentioned like you can compress frame by frame. Also, you can compress, the temporal dimension.Ethan [00:20:52]: The difference is if you compress the temporal dimension, you get a much higher compression rate. Because there's temporal redundancy between frames, because, this frame and the last frame, likely they are mostly similar, so there's only some small difference. for example, I think in 12.1 VAE, they have like a eight by eight by four compression rate. So the four temporal tokens are compressed into one tokens. That can save a lot of, save a lot of the context length. If you do it frame by frame, you have to do maybe like eight by eight by one. Your context length will be four times larger. That being said, the benefit of the frame-- per frame compression, we might come back to this later, is, real-timeness and interactivity. ‘Cause if you, if you strain the output of the model, frame by frame, you can-- the model can respond to any user request immediately. So if you have like a temporal four compression, four times compression, thenSwyx [00:22:06]: It might be laggyEthan [00:22:07]: there's a lag there in nature.Swyx [00:22:10]: So you're very pilled on this. let's just go ahead and bring it up ‘cause we have the visual prepared anyway. There's some frontier applications of real-time video gen. So Flipbook is one of the examples that went viral recently, right? What is Flipbook?Real-Time Generative UI: Flipbook, Neural OS, and Diffusion Front EndsEthan [00:22:23]: Flipbook is kind of like a web brow- web browser. You can see like it has the web bro- browser UI on top. The difference is all of the UIs are generated by generative image model in real time, and anything here are fake. But you can, you can explore inside this wor- this imaginary world. Say like we-- here we have engineering the Great Pyramid. Like the model generates this for us to understand how it works, and if we want to navigate around and understand further, we can click on some of the, some of the description here, and the model will generate a new page, new subpage describing the details we want to know about.Swyx [00:23:14]: So it's basically kind of we're playing a video, but it's pausing for our next interaction, and then it just plays the next thing based on our interaction.Swyx [00:23:23]: Which is kind of cool.Vibhu [00:23:25]: and you kind of decide your story. So this was, how do you make a pyramid? levering technique seemed interesting, right? It shows how do you take Okay, I want to know what is thisSwyx [00:23:35]: The demo, the demo tweet had more animation between frames.Vibhu [00:23:38]: I think it's just skipping,Swyx [00:23:39]: Oh, it's just skipping a lot of frames.Ethan [00:23:40]: they also have a video modeVibhu [00:23:42]: It takes a lot. There's a lot of peopleEthan [00:23:42]: but, a lot of people are using it.Ethan [00:23:45]: So it's not available.Vibhu [00:23:46]: There's a live video stream. We can try,Swyx [00:23:50]: So this is an example of the kind of future that you see at the extreme. We don't-- we're obviously not in it today.Swyx [00:23:56]: But in a world where inference is completely free this is better than generating code and text?Ethan [00:24:02]: So this is, this is a final state of where Viva will be at for word model, I think. Imagine internet doesn't exist, and then you type in google.com. Like what should, what should, what should a model show you?the model can imagine something, and this is what the model imagine. And these web pages, they completely do not exist. So I think as the inference costs come down, we are going to have generative UI for everything. If you think about how the coding model works, so they write code for a web page, and they render the code might be con- converted into binary, and the binary render the pixels on the screen. So we in machine learning, every time we have some breakthrough, obviously it's, it's more intuit. So why don't we have like user instruction to the pixel directly? So the generative UI will be user intention to the pixels directly. And say like even if I want email, let's say everyone have the same interface, but I want, I want it slightly different. I want the email to show to me like a TikTok, so I can swipe left and right for the emails. And or maybe you want something else. We can have completely different things. Or like I have I'm looking at, Instagram stories, and I don't like the Like button. I always may click it. And, generative UI resolved it. So it's going to be a revolutionary replacement of the interface. So in the future, we might have much more powerfulEthan [00:25:50]: LLMs and coding models running behind the scene. And in the, in the front-end, the diffusion model will actually be the front-end to show stuff to you. That's how I imagine it.Swyx [00:26:02]: Diffusion front-end, deterministic back-end.Swyx [00:26:04]: Something like that. I find that very expensive, but,Vibhu [00:26:08]: I find it interesting you called LLMs writing code on the back end deterministic, but okay.Swyx [00:26:14]: you write it onceVibhu [00:26:15]: Compare it toSwyx [00:26:16]: And then you execute.Ethan [00:26:17]: If you think about the cost, say, let's say H100 costs $1 per hour, and if you use this eight hours a day and thirty days, so, every month you're paying this two forty, you'll actually not wanna pay for that. That's even more expensive than Cloud Code Max. But if you think about the compute costs come down like two times every year, and I think the future will likely arrive like within few years.Vibhu [00:26:49]: It's everything, right? compute cost comes down, compute gets faster, model gets smarterEthan [00:26:54]: More efficientVibhu [00:26:54]: model gets smaller.Swyx [00:26:55]: I don't know why you say two times, ‘cause I think it's like 100 times. In language models, it is roughly one hundred to a thousand times every twelve to eighteen months, for the same given level of LMSys, ELO.Vibhu [00:27:08]: That's a net of everything, right? That's model performance alongside compute. So different than just compute costs come down. But, a very interesting future.Swyx [00:27:19]: So the web designers will have to shout out that accessibility is an issue, right? how do you deal with screen readers or whatever. But yes, this is higher bandwidth storytelling than anything you can possibly generate with code, right? So I think that's the rough idea.Ethan [00:27:34]: And I'd like to add a little bit that so human naturally have the maximum bandwidth when we are looking at things, look at videos, and we also have maximum output bandwidth when we are talking. So in the future, it might be something like we talk to AI models, and the AI model responds back with a generative UI. So that would be the maximum input and output bandwidth to interact with AI models before neural link happens.Vibhu [00:28:06]: And it's also very custom, right? Some people are very visual, some people are not as visual, right? They prefer the text. But the best thing about generative UI, right, it can also be text.Swyx [00:28:17]: There's another project that we wanted to highlight, which is the Neural OS. Kinda similar idea, but here you're literally operating, simulating an operating system with a video model.Swyx [00:28:27]: and you can play Doom, you can do Firefox. I find this like mildly less impressive, obviously, because it's an OS that I can run.Swyx [00:28:37]: But here everything is imagined.Vibhu [00:28:40]: I was, used to the Command+W to close the Firefox tab. It didn't crash. That's why I saidSwyx [00:28:45]: It's too immersive.Vibhu [00:28:46]: It's, it's too immersive for me.Swyx [00:28:47]: Too immersive.Vibhu [00:28:48]: I wanted to close the tab.Vibhu [00:28:49]: But yes, I can play generated diffusion.Swyx [00:28:51]: this is shockingly fast.Swyx [00:28:54]: Because I remember there was a demo about like maybe one to two years ago. Someone tried to do the first-person shooter with a image model. There was no consistency. It was very slow. But here it looks like realistically it's-- this is Doom.Vibhu [00:29:07]: I think there's two sides to that, right? There's okay, what is running a game? The heavy part of it is actually the game engine, all the lighting, all that stuff, the graphics. This is just kind of video, right? Like we've solved consistency. This is still, it looks like a few years old image generation. There's some temporal consistency, but it's, it's kind of just images stitched together as frame video. But it's a good visual representation to pi- to picture the future you wanna see, right? that's, that's what I see in these more so.Ethan [00:29:38]: This reminds me of how the video models gets better and better. So Neural OS is kinda if you just look at it feels like it's just a crappy version of the, like the Windows we could have, right? And, but the difference is, so the model, this model is overfitted on the existing operating systems. It can generate nothing different than that. But it's actually also similar to video models. So when we are training these video model, image model, we train them on internet. There's no imaginary supernatural stuff on the internet. But once we train this model, you can prompt the model to generate something supernatural that have never existed in the data set. So if you train your Neural OS or neural computer on the standard screen recordings on the entire internet. The model can imagine completely new interface to interact with the computer.Swyx [00:30:43]: This is one of those things that is magical to me. usually generalizing out of distribution is bad, but somehow we have learned some kind of internal world model that you say, this plus, but it looks like rainbows and butterflies, it'll do it and it will kind of make sense.Swyx [00:31:03]: So yeah, that's kind of cool. Yeah, I don't know if there's any comment more on there. I do, I do wanted to, I did wanted to touch a little bit more on the model architecture stuff, which I think you were getting. It's, really fascinating. We don't get a chance to talk about this enough. So one of the papers that we covered, we've covered every annual, segment anything release. and I don't know if you follow-- you're a computer vision guy, so youEthan [00:31:26]: I knowSwyx [00:31:27]: . So they did memory attention, which is kind of interesting. And I always think, anything where you can, across the temporal dimension, keep some consistency, I think it's, very fascinating, and I don't know if Basically, does that-- the CV side bleeding into video gen side, I think is underexplored, right? we talk about it for labeling, but actually you can borrow the architecture itself.Ethan [00:31:50]: There's, there's also complete different approaches, right? you brought up the term world model, so we went from video model to world model. There is diffusion, but there's also other approaches that people are doing. So maybe we get into those after as well,?Swyx [00:32:03]: He has a whole definition of world models and stuff. I feel like we threw a lot at you. Whatever you want to comment on.Why Video Models Are Expensive: Storage, I/O, and Training ScaleEthan [00:32:10]: I think one thing that we should actually comment back on is okay, so we were talking about the steps to train image gen to video model. One thing we don't see as much of is okay, you brought up the delta in training data, right? SoEthan [00:32:24]: you won't have as much a video model might not generalize, but what is the cost of training a large video model? So we know for LLMs roughly, okay, even like the poolside thing that came out today, right? It's a Gemma level model trained on roughly forty trillion tokens at this many H200s over this much time, right? You can see what is the exact cost of that. So how many GPU hours over how much H200 costs? So how do we do the back-end math of, same thing for video models, image models. How do you, how do you kind of break that down? I can share some back-envelope calculation. So surprisingly, video models is-- the cost is very-- is comparable to language models and obviously the largest scale is language model, maybe like a medium scale to language models. I said just storing the videos alone, it costs a lot. You can, you can maybe look up on AWS or something.Ethan [00:33:20]: You really, say if you have a billion videos and let's say, let's just say like each video, like five megabyte, then you need five petabyte to just store those videos. And also remember we talk about you use a VAE to compress the videos, and you also need to store, typically you need to store those continuous feature, in-- also in your storage. That's also comparable size with the videos themselves. So just storing these videos and the features is tens of petabytes alone. And,Swyx [00:33:58]: I just, I just looked up the calculation. Five petabytes on S3 Standard is one hundred K per month.Ethan [00:34:05]: AndSwyx [00:34:05]: It's comparableEthan [00:34:05]: and you needSwyx [00:34:06]: AndEthan [00:34:06]: And then like tens of petabytes, two hundred K. And even more expensive is you have the ingress and egress.Swyx [00:34:13]: Oh, yeah.Ethan [00:34:14]: Like you-- through the internet. You have to just to download those videos, I believe it's, it's more expensive on AWS than just storing those videos.Swyx [00:34:25]: Storing, yeah.Ethan [00:34:25]: And each training runs, you probably need to pull them once. If you train multiple times, it's, it's even more than that. So it's like just storing the network, those costs is just, it would be a few, a few millions per month to just storing everything, not to mention the GPU cost.Ethan [00:34:45]: AndSwyx [00:34:45]: my side tangent, the compute rental, like GPU rental is very efficient. There's one side, okay, you can be XAI and build your data center. Should we not just build our, storage compute as well? LikeEthan [00:34:57]: Of courseSwyx [00:34:57]: cloud cost compared to just,Ethan [00:34:59]: You save so muchSwyx [00:35:00]: store. Yeah, exactly.Swyx [00:35:01]: Especially with like egress and stuff. So.Ethan [00:35:04]: That's a good idea, but it also comes to-- there are some of its own challenges.Swyx [00:35:09]: Of course, of course.Ethan [00:35:10]: like people who build the GPU data centers, they might not expect this much, storage. And yeah, people build storage, typically they just build it somewhere with just CPUs.Swyx [00:35:23]: I just looked it up. Five-- AWS only charges for egress, not ingress. Tier five for five petabytes is two hundred and thirty K.Ethan [00:35:32]: Even more expensive than the storage.Swyx [00:35:34]: But storing is per month, right? You check in, then you cannot check out. so it's so cool. It's okay. So there's that side.Ethan [00:35:41]: So the TLDR, my backhand mathSwyx [00:35:42]: Data is larger than you think. Yes.Ethan [00:35:44]: my backhand math of GPU hours times GPU cost is also very much, I'm missing some storage.Swyx [00:35:49]: You're also-- you're basically like also more IO bound than normal training.Swyx [00:35:55]: Yes. ‘Cause like data loading, so caching everything, it becomes super important.Ethan [00:36:00]: So in Cosmos, we did a lot of optimizations to make it not IO bound. So, speaking of the training, actually training the model, the GPU cost, if you look up like the open source model, how big these video models are, I think like LTX has nineteen B parameters. That's a dense model. And people are also exploring, MoEs, so it might be twenty B active and, like a hun- hundreds B, total. So that's, that's even-- that's similar size as medium-sized LLM models. And if you, if you look at number of tokens-Uh, we disclose that in Cosmos. It's also like tens of trillions of tokens on the visual tokens. So putting this together, the cost of, training these video models, it's actually comparable with LLMs. Not to mention, the infra is slightly different from LLM, so it might be less efficient to train these models.Inference Speedups: Step Distillation, Consistency Models, and GANsSwyx [00:37:04]: Do you get the benefits of traditional diffusion speed-up? So for, images, there's LCM, LoRAs for, fine-tuning. There's, there's a lot of stuff that's beenEthan [00:37:15]: Flow matching.Swyx [00:37:16]: there's flow matching. There's a lot of stuff that's been done. there's some overlap that applies to diffusion on the inference side and stuff or?Ethan [00:37:23]: so the difference-- the inference side is a completely different story.Ethan [00:37:28]: I think for the training side, it might be a little bit hard to reduce that cost. And for the inference side, the biggest gain is from the distillation of these models. You can-- It's called step distillation, slightly different from knowledge distillation in LLMs. So you-- Typically, for flow matching models, you need like 100 steps or something. Like a distortion model even need even more, like 1,000 steps to generate a good image or video. A step distillation is try to learn to generate fewer step from the model itself. It's kind of like now we-- you use the full model to generate in 100 steps, and then you take a model that only generate 10 steps and let that model to learn from the perfect one.Ethan [00:38:25]: why this workSwyx [00:38:27]: Strong to weak seemingly.Ethan [00:38:28]: It is. It's kind ofSwyx [00:38:29]: DistillationEthan [00:38:29]: kind of like strong to weak. the-- from the modeling perspective, the strong model, the teacher model is trying to model the image and videos of inter-internet, and that distribution is extremely complex. But the step distilled model is just trying to learn from the teacher. The teacher is a model, and the size is fixed, as the distribution is much simpler than the whole internet. That's the intuition I have why step distillation can work. So usually these models serve in productions, they only run in a few steps. In Cosmos, I believe we have, we have like four step and eight steps. If you do some simpler task, image-image translation, it can even run in fewer step, like one step in Cosmos Transfer.Swyx [00:39:22]: I think this is the same intuition that guides a lot of the consistency model work. I sent you a link for, SCM. I don't know if you covered that. To me, that was actually one of, the most impressive papers I've ever seen from OpenAI.Swyx [00:39:34]: That this is the unifying grand concept of consistency models. I don't know if you have any comments on this.Ethan [00:39:41]: So there are, there are a few different approaches,Swyx [00:39:46]: Oh, yeah. Here it is.Swyx [00:39:47]: Two steps versus twenty or 100 steps, whatever. It's already done.Ethan [00:39:52]: So there are, there are a few different approaches, for example, consistency model, and there are also Actually, we shouldn't forget GAN. So GAN, actually, that was, that was the OG ofSwyx [00:40:05]: OGEthan [00:40:05]: step distillation ‘cause it trained just one step to begin with. So actually, a lot of, uh-- For example, there's a distribution matching distillation which use, which uses GAN, as one of the laws for distillation. It-- GAN just tells you, “Hey, generate an image,” and thenEthan [00:40:31]: it has a discriminator to tell, is this image real or not? So the model, the model just need to learn one of the distribution, not the full distribution. Because in training, the model is asked to reconstruct the ground truth image from the internet, which is extremely hard. And in-- When you're training GAN, it's a step process. It's just a, “Hey, you generate image. Does this image look as real as the image from the internet?” Which is a much simpler task. And, yeah, combining a lot of these approaches together, people typically do that, like consistency model and distribution matching and GAN, and we can get these few step models.Audio-Video Generation and Time AlignmentSwyx [00:41:21]: Then there's one step I wanted to add, which is audio and video.Ethan [00:41:26]: So, Grok Imagine zero point nine, I believe it's, it's a first audio video transmodel deployed at a large scale. SoSwyx [00:41:39]: And that was your first model?Ethan [00:41:40]: that was, Grok Imagine's first model. It's, it's audio video, joint generation. I think the hard part is, the modality alignment, ‘cause before this transmodel, we have, we have text to video alignment. We have this, correspondence between text and video. Typically, most of the VLMs, they understand images and videos. Video's very rare, and they don't understand audio mostly. And if you look at the audio generation on the LLM side, you can talk to them perfectly fine, but if you ask them to sing a song or something, it typically is not very good. Also, they don't have, they don't have music either. The hard part is thatUh, actually audio has two component. It has like a discrete component, a continuous component. The discrete component is like the language.Ethan [00:42:44]: So when we speak, it's just, someSwyx [00:42:47]: It's an ASR issue, yeah.Ethan [00:42:49]: It's, it's text token with some characteristics, I would say.Ethan [00:42:54]: But musicSwyx [00:42:56]: I think the speech guys would disagree with this.Swyx [00:42:57]: Like disfluencies and then,Vibhu [00:43:00]: There's tones you can get angry.Ethan [00:43:01]: Well, I say largely.Ethan [00:43:03]: the mu- but the music is completely different. It's, it's very continuous, and you cannot model them like discrete tokens in language models. this is like the hard part for models is, not to mention we have to align text, video, and audio together.Ethan [00:43:26]: SoVibhu [00:43:26]: How?Ethan [00:43:28]: So significant-- some significant challenges are like-- So first, like we talk about as the VLMs, they cannot understand most of them cannot understand audio.Ethan [00:43:39]: So you have to have some way to do the synthetic data generation for audio. You have to caption the model, and that involve, that involve synthetic data and human data effort a lot. And not just surprisingly, most of the LLMs are very bad at recognizing, like the beat, tone, and the details of the of music. They can, they can give some general prediction of which song is this, but it's very hard to describe the details of the music. like we mentioned in image generation, like you have to describe image as detailed as possible so that someone blind can reconstruct that. So here is like someoneVibhu [00:44:32]: DeafEthan [00:44:32]: someone deaf can reconstruct how the music sounds like without actually listening to it. Maybe you can think of it need to have the-- or they call the script.Vibhu [00:44:49]: Subtitles, yeah.Ethan [00:44:49]: You gotta have all the details of the music, and the dialogue.Vibhu [00:44:55]: So is the challenge there typically stuff like music and audio, or is it just Like is there a baseline? Okay, there's enough data where we can understand, narration, conversation, but there's nuances in audio that's where you hit all the data issues or is it just from stage zero, you just do it all right?Ethan [00:45:15]: So one important thing is like the alignment. So the model, the model has to know like the video and audio, the, uh-- it has to have a time-based alignment, like at which time step the video and the audio token correspond to each other. But we actually don't have this kind of alignment for most of the other modalities. If you think about like text and image, text and video, they are loosely aligned. So you can, you can have a description of what's going on in the video, but you don't have to exactly, You typically don't have exact description, oh, at, time step one second like what happened?Vibhu [00:46:02]: It's veryEthan [00:46:03]: At time step two second what happenedVibhu [00:46:03]: coarse. Yeah.Swyx [00:46:05]: So what was the ideal time step? You have to oblate it, and then it's like four seconds or something.Ethan [00:46:09]: So that comes down to how you design the model to, for the model to be aware of as a time, as a time modality. So the model is like a time aware. And that's something pretty unique if you think about LLMs. So if you ask LLM to complete a task, say they, uh-- you ask them and they will say, “Oh, this task will probably take twelve hours to complete,” and they come back in one hour. Say “I've already spent two days on this and I've exhausted everything.”Ethan [00:46:47]: So the LLMs them-themselves, they don't have a sense of time there.Vibhu [00:46:53]: I actually don't think that's just them not having a sense of time. I think it's somewhat based, right?Vibhu [00:46:58]: Like you tell someone, “Okay, go work on this feature. Go implement this,” there's a general understanding you would have of how long that would take without LLMs working at LLM speed, right? So you think back like two years ago, if I tell you to like build me like a new front end for latent space, have a search bar, have all this, you'll estimate that it'll take a few days, right?Vibhu [00:47:19]: So you tell an LLM, “Go build this.” It'll take me a few days. But I think it's somewhat grounded as opposed to them not having the best-- Not saying that they have a great understanding, but I think that example is like you can see where it comes from, right? You're trained on all over the text.Swyx [00:47:35]: They're, they're trying to estimate what a human would say.Vibhu [00:47:37]: because that's what the, that's what the data kind of represents. It's not themEthan [00:47:41]: It came from the corpus on the internet. People have a estimate of how much time.Vibhu [00:47:45]: And not even just in direct like training samples, right? Just your world understanding of tokens of how long stuff takes, right? Go read a book. It'll take you a while, right?Vibhu [00:47:56]: Even if you do nothing but read a book, it takes a few days. So yeah, LLM, I read it took me a few hours.Vibhu [00:48:01]: It'll take me a few hours to go through this research. But this is a tangent.Swyx [00:48:05]: Somewhat, yeah.Swyx [00:48:06]: This is a train of thought I haven't really expressed until now is, which is basically like a full world model must also be recursive, meaning that the participant in the world model must also be aware that they have a world model. which is like this whole recursive thing down the, down the line. but yes, and that the world model can be wrong and that they need to update it and blah. Yeah. We've, argued this on the, newsletter as well, that there needs to be sort of recursive or adversarial world models.World Models: Real-Time, Long-Horizon, Interactive VideoVibhu [00:48:34]: just, to ask, how do you define world model?Swyx [00:48:38]: Oh, yeah, let's go there.Ethan [00:48:40]: SoVibhu [00:48:40]: So just for context, we talked about, video generation, and then there's a-- if you say there's a distinction between world models, what's your, what's your definition? How do you see the two?Ethan [00:48:53]: So disclaimer, I'm not going to debate, what is world model. Yeah. there are many definitions, so I'll just talk about my definition. Since I came from the multi-model, multi-model domain, so mainly talking from video. So world model is like real-time interactive long horizon videos. So there are three parts. so we-- let's talk about them one by one. So the so interaction, so we just, we just look at Facebook and neural computer. So the interaction part of it, so you, world model can allow you to interact with them through keyboard, mouse, and maybe also voice. So these all is-- all is a modality. You can, you can interact with the model, and the model should respond reasonably. Second part is real time. So once you, once, say, you move your mouse, if, say, the world model generate a game, how fast can the game respond? So if you're like professional CS: GO players- -my say, oh, you have to respond- He's beginner within sub ten milliseconds or- Yeah even less. So that's not most of the- No, sixty FPS. Let's go. Oh, three hundred FPS. Oh, five hundred FPS. Wait. okay, yeah. I didn't do the math, but yeah, okay. Uh- Yeah, three hundred FPS, that's a three millisecond. So you have to respond- Oh, s**t. Okay. YeahEthan [00:50:29]: within a millisecond. Most of the video models cannot do that. Yeah. And, but if you, say, if you have a video model that is, say, like a digital human, the response time might be more generous. Maybe typically, for real-time voice interaction, it's like two hundred millisecond. So that's, that's much more generous. But even two hundred millisecond is pretty, it is pretty tricky, ‘cause remember we mentionedEthan [00:51:01]: you have this, temporal compression coming from the VAE. So if you, if you don't compress the temporal dimension, your sequence length is going to explode. So if you want to have this real-time, real-timeness in your model, you have to do is one context problem. And the third part is long horizon, ‘cause we-- if you're not going to just play with, video games just, a few seconds, most video models only a few seconds. We're going to play with minutes, hours. The model have to be able to generate long-form content.Ethan [00:51:42]: So putting these three together, it's, real-time, long horizon interactive videos. I think the final state will be, for example, like a video, a video version of Playbook, where you can, you can interact with, a neural computer. You move your mouse, and you click on the generative interface, and it will reply to you through pixels- generating in real time. But getting there, it's, it's a very long way to get there. So one of the first step, at Grok Imagine, where I led a small world model team there, was to build video extension. So, video extension- it's the first step of interactivity. Yeah. It's, it's the first step. Yeah. So it's the first step- You have it here, video editing, yeah. Yeah. Yeah. So the first step is because, this unlocks long horizon videos. Typically, for most of the video generation models, you give it a prompt or an image as an initial frame. You generate video, that's it. That's just, one time, done. And some creators would try to, use the last frame as a first frame for the second video. It can-- sometimes it works, but if you do it a few times, it says the quality would decrease. And- It doesn't have that context- Yeah over the full video, so the temporal- Yeah, exactly. Yeah, ‘cause you only gave it the last frame, of course, right? Yeah. Exactly. And- it's actually a pretty fun hack. if you've seen like- Oh, no, he's saying something better. Yeah. And for example, like Vue, I remember Vue 3 has like a second context of the last video. It is slightly better than using the last frame, but it has the same problem-- similar problem that it, the quality would decrease. if you extend a few times to, one minute, the video quality would look much worse than the first video. Second, another problem is that the model doesn't have long-range knowledge of, what's happening before. Say, if they generate some dialogue, some, two people speaking, and their voice might change, over some time, especially if the second conditioning, it does not cover the previous context. So these are the core challenges. So the Grok Imagine video extension, it has historical context of all of the previous generated videos. It can, It has, it has the context of, who is speaking and what objects have appeared and everything, having that to generate the next video. So if we naively do this, you can imagine, just, put all of the previous history video tokens into the context. The context lens will easily explode. Especially for video models, that can be like a few, a few million context, I would imagine- context lens. Yes.Yeah.Swyx [00:54:58]: Let's run with that.Ethan [00:54:59]: for example, like in Cosmos, I think just five seconds of video is like a fifty K or sixty K number of tokens. So like if you do, if you do fifty second, that's a five hundred K tokens. If you do longer than that, easily explode. This long horizon, problem was the first step we're trying to solve world model. It turns out people, yeah, people love video extension. Like a lot, a lot of the creators love using video extension to create longer form videos. This is the part I liked that you have a, you have an intermediate step toward the final goal instead of just a straight shot to the final version very much.Swyx [00:55:48]: But I can see you have a strong vision of where we want to end up.Long Context, Redundancy, and Efficient Interactive VideoVibhu [00:55:51]: Does it seem like it's an efficiency issue? okay, we're at a few million tokens context,. If you draw the parallel to language models, we had very short context, two thousand, eight thousand, then, you scale it up one million, ten million. sure, there's effective context, but at the end of the day, it's just what's it worth? sure, there's a whole training data side. In video, it might be slightly easier ‘cause we have a hundred million token video, right? Just take a movie with the full context there. Like is this efficiency from an inference standpoint that like it's expensive, but we know how to solve it? Or like why is this not the approach? So like my broader point was on your second point of world models, you say it needs to be interactive and live, right? You should be able to play a game and see the interaction live. So one thing I see with research is a lot of what you actually serve is different than what you build, right? So we talked about distillation. You train big model, you distill it, you do quantization, speculative decoding. We do all this stuff to serve it efficiently. Should we not just have a solution, like a world model that can interact well, do inference optimization, serve it, distill it secondary, so make it real time after you solve it? So like a-- another parallel is say, continual learning, right? What we need is someone to solve it and show it works inefficiently. Give it a few years, people will make it efficient. Same thing with regular attention, right? It worked. Over a few years, people have different forms of attention, and we've scaled it to be efficient at log context,? So kind of two things there, right? One is it seems like it works. You've scaled it. Can we not just scale it a lot more efficiently over time? Do we need a separate approach if this works? And same thing with interaction, right? if we can get it done, like if we can solve some way that it works, we can solve making it more efficient from an inference standpoint later.Ethan [00:57:53]: that's actually a very good point. So in videos, there's actually a lot of redundancies. So we solve a lot of the pixel redundancy from VE, but there's more redundancy in long range and long horizon videos. Say, if a character appear in the first clip and then it disappeared, it only reappear at the end of the video, you probably don't need the-- the context, like in the middle of the generation. So you only need that character, where you need. So that's why, I helped build another feature. It's a reference video.Vibhu [00:58:36]: Is it here?Swyx [00:58:36]: is it the same model release or different one?Ethan [00:58:39]: It's a different one.Ethan [00:58:41]: You probably need to search onSwyx [00:58:43]: I'll find itEthan [00:58:43]: X reference to video.Ethan [00:58:46]: So reference video allow you to like upload up to seven images as condition and generate the video. Say, if like I want-- it can, it can be characters or objects or even scenes. Say like I want, I want condition on, Sean's selfie and holding a bladeSwyx [00:59:07]: We have a dogEthan [00:59:08]: or whatever.Swyx [00:59:08]: We put the dog in the thing.Ethan [00:59:09]: you can put them there and the video models will generate the video from and copies the context over. So that can solve a lot of the problems there, like the long context problem. It doesn't need to have a very long context, but it's-- I feel like it's an intermediate solution. The modelSwyx [00:59:29]: It's cheating.Ethan [00:59:30]: the model should be able to like selectively know, where should I draw the references. So say if I want to generate a movie, I generate it autoregressive, like a ten second at a time or something. And now this character appear, I can look back to where it first appear and, bring that back. Yeah, this one, I put the references. Yeah, that's, Optimus, Einstein myself, Annie.Vibhu [01:00:02]: Oddly enough, I used Grok Search to find it, and it pulled your LinkedIn post. But yeah we found it.Ethan [01:00:08]: Interesting.Vibhu [01:00:10]: ButxAI's Underrated Work, Culture, and WatermarkingSwyx [01:00:11]: this is a problem. This is not your fault, but like XAI doesn't communicate all this work that you do very well because they just have the model release and then that's it. But actually, these details are very good.Swyx [01:00:22]: As far as I understand, everything you just described is state-art, like no one else has done it.Vibhu [01:00:30]: A lot of-- yeah, I have a lot moreSwyx [01:00:32]: And then, and then you just put this blog post with the cookies. I'm this is not enough,?Swyx [01:00:37]: but I, obviously this is like the high level numbers that people want to know. But no, okay, soVibhu [01:00:42]: And I wonder, like part of that is also some labs don't share research into what happens. And ifSwyx [01:00:50]: No, but this is literally bragging about how good they are, right?Swyx [01:00:54]: Like, why would you not say that you are capable of extending with full context? this is not a secret sauce. This is like we did the work. yeah, I don't know.Ethan [01:01:02]: different labs have slightly different communication styles.Swyx [01:01:07]: Anyway, if anyone from XAI is listening we are always happy to help you tell your story. Yeah, okay, so you did references, and I think, I think kind of the point you're, you're making is it is sort of like a kludge, right? this is-- you can do seven, but what about 100?Swyx [01:01:23]: Right? Then you need a completely different thing.Ethan [01:01:26]: So I think it's-- this is, a mechanism to, select the context from the history, and you might not put the entire history into the context. for example, there's a paper called Frame Pack, which haveEthan [01:01:41]: a heuristic that the latest history, the last one second, I put the entire history, and the history before that, I would, compress it and makes the video smaller. So they follow this pattern, this build overall pattern that the maximum sequence length is fixed. So the further you are from the current frame, you have a smaller image. So this is just a heuristic. I think it can be more automatic. The model is aware like which history part of it can be select. So this part of the research is actually being actively, worked on by a lot of people. It's also quite interesting. I feel this is actually, this part of long context is a little bit ahead of the LLM part.Ethan [01:02:31]: So for example, like in LLMs, if you-- so contexts keep growing. Let's say if you call tool and the tool call history is extremely long, that's still in context, and keep growing, keep growing. Even if you switch the topic to something else, the whole context was there. There are some agentic harnesses that help you to, say, prune the tool results and, prune Like when you, when you query a file, only show like the top 200 lines or something. Those were very heuristic-driven.Swyx [01:03:08]: For listeners, we did a write-up on the cloud code, leak where there are eight different kinds of pruning, including like you prune the tool results and all that. So you can, you can read up on that kind of thing.Ethan [01:03:17]: I think, one breakthrough in continual learning might be like a way to automatically, manage its own context.Swyx [01:03:27]: These are all heuristics, and they will be replaced by machine learning.Ethan [01:03:30]: InterestinglyVibhu [01:03:32]: TheEthan [01:03:32]: the same thing is being researched in both LLMs and video models.Vibhu [01:03:36]: The interesting thing is also like in the paper you showed, it's actually happening at the model level, right? Compared to like language models, sure, we have base attention, but we'll do our own compression, we'll do our own pruning, which is separate from model error.Vibhu [01:03:49]: Eventually, it all just boils in, hopefully.Swyx [01:03:52]: I think this is a form of like attention, but like also know sort of reasoning attention. I feel like that's different than normal attention.Swyx [01:04:03]: Does that, does that make sense?Ethan [01:04:04]: It's, it's different in the sense that attention, not to mention, set sparse attention aside,

Introduction International tech companies burn through $2 million trying to crack the US market every day. Not because their product is wrong. Because they hire a sales team before they have a sales motion. Dan Griffith has spent 15 years watching this mistake play out—and building the playbook to prevent it. Griffith is the founder of Greater Gain Group, a go-to-market firm that helps software and technology companies—most of them international—land and scale in US insurance, financial services, and healthcare markets. As the first US hire for a South African company, he scaled it from $3M to $150M in three years. Those hard lessons became the foundation for Greater Gain Group's 90-day go-to-market framework. In this conversation, Josh Hollander and Griffith dig into why the unicorn sales hire is the most dangerous move an international founder can make, what has to be true before you put a rep in a seat, and where the insurtech market is creating real demand for cross-border go-to-market right now. Guest Bio Dan Griffith is the Founder and Principal Consultant at Greater Gain Group, a go-to-market consultancy specializing in helping international software and technology companies enter and scale in the US insurance, financial services, and healthcare markets. With 30 years in enterprise sales and marketing, he has served as a first US hire and go-to-market architect for companies entering from South Africa, France, Europe, and beyond. His 90-day framework takes founders from "we're entering the US" to a repeatable sales motion—without the $2M mistake. Key Topics • The $2M mistake — A VP of Sales, two account executives, a marketing hire, an office, and conference travel. You're at $2M in under a year with nothing built and no pipeline. Fifty percent of Greater Gain Group's clients have already made this mistake before they call. • Don't hire salespeople (yet) — The tell that a founder is about to flame out: they say they're going to hire a sales team. Griffith's rule: build the sales motion before you build the team. A rep can't fly a plane that hasn't been designed. • The founder has to come — For companies under $50M, having a founder on the ground for early US conversations is the strategy. Hearing objections directly is how you convert from founder-led to team-led sales—the transition Greater Gain Group is built to facilitate. • Three to five segments, not one — Pick no fewer than three and no more than five market segments, understand the pain in each, and build an outreach engine that generates sales conversations—not leads. Leads have no value. • Paid pilots and MSA reality — US buyers do paid pilots. Free pilots signal low value and waste time. On contracts: insurance companies have ten times more lawyers than you. Know your non-negotiables, keep the list short, and don't let MSA rigidity keep you out of the market. • Price higher than you think — International companies consistently underprice the US market by 20–40%. Corporate budgets at US insurers are significantly larger than abroad. One client was surprised a health insurer's CTO had $475K of year-end budget left for a POC they'd hesitated to price. Notable Quotes "They hand you your laptop and say, go sell us some stuff. I learned a lot of hard lessons on how not to do things." "If you don't bring value, you're out. The US market is transactional. As much as I hate to say it." "A lead has no value. Build an outreach engine that generates sales conversations." "Your only competitor is the status quo. If you're getting into a feature-function-benefit argument, you've already lost." Resources Guest: • Greater Gain Group: https://www.greatergaingroup.com • Dan Griffith on LinkedIn: https://www.linkedin.com/in/dangriffithsr/ Host & Organization: • Joshua R. Hollander on LinkedIn: https://www.linkedin.com/in/joshuarhollander/ • Horton International (USA): https://www.horton-usa.com/ • Insurtech Leadership Podcast (LinkedIn Showcase): https://www.linkedin.com/showcase/insurtech-leadership-show Subscribe & Review If you enjoyed this episode, subscribe on your favorite platform and leave a review. The Insurtech Leadership Podcast is available on YouTube, Apple Podcasts, and Spotify.

Bienvenue en haute altitude !!Vous aimez la vitesse ? Les records ?Les traversées fantastiques ?Bienvenue à bord ! Voici l'histoire d'une femme qui n'avait peur de rienUne casse-cou qui fut la première femme à traverser l'Atlantique en soloQui fut reçue et décorée à la Maison BlancheVous avez son nom ??AMELIA EARHART !Probablement l'une des plus grandes célébrités de son époqueElle fait la UNE des magazinesEt enchaîne les exploitsEn 1937, elle disparaît dans le Pacifique alors qu'elle tente une nouvelle dinguerie : le tour du monde Personne n'a jamais retrouvé le corpsPersonne n'a jamais retrouvé l'appareilPour en savoir plus sur cette femme et sur ce mystère Une seule adresse : le podcast FIFTY STATES !Hébergé par Audiomeans. Visitez audiomeans.fr/politique-de-confidentialite pour plus d'informations.

From heroin to healing to hosting.In 1994, Eric Zimmer walked through the doors of Maryhaven, an addiction treatment center in central Ohio, as a client seeking help for heroin addiction. Today, he is an author, teacher, speaker and the creator of "The One You Feed" podcast.Fifty million podcast downloads later, Zimmer shares what recovery really teaches.Zimmer has a new book called How A Little Becomes A Lot: The Art of Small Changes for a More Meaningful Life.He joins All Sides to talk about his new book. He will also discuss his journey from being a client at Maryhaven to returning decades later in recovery and now partnering with them around his new book.Guests:Eric Zimmer, author, How a Little Becomes a Lot/host of "The One You Feed" podcast

Fifty thousand troops. Zero reporters on a ship. Zero reporters on a base. That's the reality of the Iran deployment under Trump and acting secretary of culture war Pete Hegseth — and it's the kind of information vacuum that's never existed in modern American conflict. Paul is joined by ABC News chief global affairs correspondent Martha Raddatz, one of the most respected and trusted voices in military journalism, for a no-BS briefing on what happens when the Pentagon shuts the press out of a shooting war. This is a conversation about more than access. It's about trust — the trust the American public places in a non-political military, the trust troops place in journalists who actually show up, and the trust that gets shredded when a defense secretary turns a West Point graduation into a culture war rally. Paul and Martha walk through the Memorial Day lines that got crossed, why embeds matter, what the rank and file actually think about the politics being shoved down their throats, and why the easiest way to stop the truth is to never let anyone see it in the first place. -WATCH full video of this episode here. -Join IVA and stand up to Trump's Forever Wars. -Learn more about Paul's work to elect a new generation of independent leaders with Independent Veterans of America. -Learn more about American Veterans for Ukraine here. -Remember Independent is an Attitude. -Learn more about The Headstrong Project for Veterans, Tragedy Assistance Program for Survivors (TAPS), and Department of Veterans Affairs resources in your area. Seeking support is not a sign of weakness. It's a show of strength. If you or a loved one are in immediate crisis, dial 988 and press 1, or text 838255. Connect with Independent Americans: Subscribe on YouTube, Spotify, Apple Podcasts, and all podcast platforms Read more at Substack Support ad-free episodes at Patreon  Connect: Instagram  • X/Twitter • BlueSky • Facebook  Follow on social: @PaulRieckhoff on X, Instagram, Threads, and Bluesky -Join the movement. Hook into our exclusive Patreon community of Independent Americans. Get extra content, connect with guests, meet other Independent Americans, attend events, get merch discounts, and support this show that speaks truth to power.  -And get cool IA and Righteous hats, t-shirts and other merch now in time for the new year.  Independent Americans is powered by veteran-owned and led Righteous Media.  And now part of the BLEAV network!  Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.

In today's episode we look at the 2008 TV story, The Sontaran Stratagem / The Poison Sky. Fifty-two people across the world in eleven different time zones die at the exact same time. The only connection: they all have ATMOS installed in their vehicles. Martha Jones, now a UNIT doctor, summons the Tenth Doctor back to modern-day Earth to help figure out why, but an old enemy lies in wait... Before that, we kick things off with the usual mix of news and short topics. Then, after the TV episodes chat, we open the mailbag for a ton of feedback from our listeners on a number of topics.  And also stick around after the show ends, if you're so inclined, for a short discussion of the final episode/final season of THE BOYS, as well as commentary about the show over the years. Enjoy, dear listener. Contact us: Bluesky: @thedwshow.net X / Twitter: @theDWshow Email: hello@theDWshow.net Substack: thedwshow.substack.com Facebook: facebook.com/theDWshow

More than fifty thousand tips have been submitted in the Nancy Guthrie disappearance. A retired Pima County detective believes the suspect's name is probably already in that pile — investigators just haven't reached it yet. DNA recovered from Nancy's Tucson home has been shipped to the FBI crime lab at Quantico, where genetic genealogy analysis is reportedly ongoing. No arrest. No named suspect. And the person allegedly responsible has had months to make decisions — what to do with evidence, who to avoid, whether to stay or disappear. Psychotherapist Shavaun Scott joins Hidden Killers to examine what those months have done to the mind behind this alleged crime. Scott has spent more than thirty years in forensic settings studying not just what drives someone to violence, but the psychological machinery that either holds or breaks in the aftermath. She dissects the post-crime decision cascade — how each choice to conceal, each near-miss with the investigation, and each day of silence deepens the psychological burden. She explains what the specific threat of genetic genealogy does to someone compared to traditional investigative pressure — a scientific process working toward identification on a timeline nobody can predict. And she addresses whether the presence of a co-conspirator stabilizes someone or creates mutual paranoia where the fear of the other person talking first becomes its own form of psychological siege.Footer Links:Join Our SubStack For AD-FREE ADVANCE EPISODES & EXTRAS!: https://hiddenkillers.substack.com/Want to comment and watch this podcast as a video? Check out our YouTube Channel. https://www.youtube.com/channel/UC8-vxmbhTxxG10sO1izODJg?sub_confirmation=1Instagram https://www.instagram.com/hiddenkillerspod/Facebook https://www.facebook.com/hiddenkillerspod/Tik-Tok https://www.tiktok.com/@hiddenkillerspodX Twitter https://x.com/TrueCrimePodDisclaimer:This publication contains commentary and opinion based on publicly available information. All individuals are presumed innocent until proven guilty in a court of law. Nothing published here should be taken as a statement of fact, health or legal advice.Hashtags:#NancyGuthrie #SavannahGuthrie #HiddenKillers #TrueCrime #ForensicPsychology #GeneticGenealogy #PimaCounty #Tucson #ShavaunScott #CriminalPsychology

What if you attracted 50 buyers in two months - for a product you almost didn't list? That's not a marketing strategy. That's exactly what happened when 21-year-old Ovi Shekh posted Wisdomic AI on Acquire.com and watched his inbox fill up faster than he expected. Ovi is a CS student from Dhaka, Bangladesh. He's already exited two businesses before most people his age have submitted a single job application. His first exit came almost by accident - a COVID-era grocery delivery startup, quietly acquired after the buyer tracked him down on Instagram. His second was Wisdomic AI. An AI-powered academic research tool he'd spent eight months building. Ten thousand signups. Nineteen hundred active users. Fifty-plus universities. And a product he genuinely didn't want to let go of. But he listed it anyway. Just to see. Fifty-two inquiries later, he had a signed LOI with his chosen buyer. And then a better offer showed up. More money. Different vision. And Ovi walked away from it. Because here's the thing most first-time sellers never think to use as a dealbreaker - vision alignment. Not the highest number. Not the cleanest terms. Whether the buyer actually believes in what you built and will carry it forward the right way. That was the filter. That was the whole decision. The buyer Ovi chose went on to raise $700,000 using the asset Ovi sold him. Let that sit for a second. In this episode, Jaryd sits down with Ovi to unpack how a 21-year-old from Bangladesh navigated two exits, turned down a better offer on purpose, and figured out the rules of the acquisition game earlier than almost anyone around him. How he valued an eight-month-old SaaS with no ARR and a niche user base that didn't behave like typical consumers. Why he applied to Y Combinator eight times, got rejected every single time, and what that finally told him about where his leverage actually lived. And the one thing he says nobody tells you when you're building - that you don't get rich owning a startup. Only selling one. Most founders fall in love with their product and never let go. Ovi fell in love with his, listed it just to see what would happen, and walked away with a lesson worth more than the exit itself.

A testimony to keep praying and not give up.

Well, that Mosque shooting disappeared faster than cocaine at a Hunter Biden party.Seattle's Democratic Socialist Mayor is losing businesses like no where else. The Colombia Tower Club just closed after 40 years. Amazon Fresh and Amazon Go has closed all their stores. Jeff Bezos left, Howard Schultz founder of Starbucks left. Their capital gains tax collection is down 50%. Per Cushman Wakefield vacancies rates are 36.5 for commercial property. Pioneer square is at 50% vacancy. The Needle, Seattle's iconic structure is now a homeless encampment. Business are running from socialist ideas and sanctuary cities. At this pace tax rates will increase on those remaining. It's just a matter of time for the city to collapse. Fewer people to tax, fewer jobs, more homeless.[X] SB – Ad against TalaricoGod is non-binary6 sexesAmerican flag complicated signalStephen Colbert signs off from late night television, and the media acts like we just watched the first moon landing, the fall of the Berlin Wall, and the Beatles reuniting all at once. “Historic ratings!” they cry. “A cultural moment!”Yeah? Let's talk about those numbers.Colbert's final show pulled 6.74 million viewers. And to be fair, that is a big number by today's standards. It was the highest-rated weeknight episode he ever had. Bigger than his premiere. Way above his recent average of around 2.7 million.But here's the problem. Context is undefeated.Johnny Carson's final show in 1992 pulled over 55 million viewers. Fifty-five million. That was when America still had fewer people and fewer TVs. Carson had a 62% audience share. Think about that. Six out of every ten televisions in America were tuned into one guy sitting behind a desk telling jokes.That's not a TV host. That's a national event.Jay Leno signed off with nearly 15 million viewers. David Letterman got almost 14 million. Colbert, meanwhile, needed every other late-night host to basically go dark and funnel their audience to him just to hit half of what Leno and Letterman did.And this was his BEST night, outside of his piggybacking on a Super Bowl one night.That's like a baseball player retiring with a .195 batting average and ESPN running graphics like Babe Ruth just left Yankee Stadium.What happened to late night?Simple. It stopped being funny and started becoming political group therapy.Johnny Carson made everybody laugh. Republicans, Democrats, people who didn't know who the Vice President was. Carson wasn't trying to “educate” America. He wasn't trying to save democracy between commercials for sleep medication and adult diapers. He just wanted to be funny.Colbert and these modern late-night guys? Entirely different business model.Every night became the same routine: Trump joke. Republican joke. Democracy is ending. Commercial break. Repeat until pharmaceutical side effects include “thoughts of self-harm.”At some point, late night stopped feeling like comedy and started feeling like being trapped at a dinner party with your angry NPR cousin who uses the phrase “lived experience” while borrowing money from his parents.And then you see the staff photo.Have you seen this thing? It looked less like a comedy show staff and more like a government agency. I heard estimates anywhere from 120 to nearly 200 people working on that show.See Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.

There Is an Appointed Time for Everything — Episode 296 Mini Miracles From Minor Moments  ·  with Linda Gullo Scripture tells us there is an appointed time for everything, and the older I get, the more those words settle into my bones. In Episode 296 of Mini Miracles From Minor Moments, I share what it has felt like to move through season after season — from babies in the house to an empty nest, from boundless energy to a body that asks me to slow down. I think about the 53 years my family has spent in one yard, the pets we have loved and lost, and the keepsakes that once meant everything. This is an episode about change, acceptance, and the quiet plan I believe is at work in every stage of life. Change shows up in small, ordinary ways. I tried to screw a new garden hose onto the spigot and could not get the old one loose — a tiny moment that reminded me things shift, and we have to adapt right along with them. I talk about how teenagers start out as their own little people, blend together through the junior high years, then slowly find their unique spirit again, and the pull that social media places on that tender process. I look back on the boulders my husband Tony dug up to build a waterfall, the farmland that became our home, and the holidays and gatherings with friends who have since moved away or passed on. Every one of those memories has its own appointed time, and I find real peace in trusting that God designed it that way. This episode looks ahead too. I share my thoughts on AI — how it is showing up everywhere now, how our children and grandchildren are growing up with it, and how we can learn to tell what is genuine from what is artificial. I even created some music through AI that you will hear on upcoming episodes, and I want you to know my words here are always truly my own. I talk about bridging the wisdom of older generations with the tools of a new one, the gift of a wonderful Mother's Day spent with family, and the simple choice between an audiobook and a quiet hour with a real book and a cup of tea. My hope is that you walk away asking yourself one gentle question: what time am I in right now? Scripture's reminder that there is an appointed time for everything — and how that truth brings comfort through every season of life. The garden hose that would not budge — a small, everyday picture of how things change and ask us to adapt. Watching teenagers grow from individuals, into look-alike groups, and back into their own unique spirits — and the pull social media places on that tender process. Fifty-three years in one yard — the boulders, the waterfall, the gardens, and the farmland that became a family home. The pets, keepsakes, and sentimental treasures we hold dear — and making peace with the truth that their time, too, will pass. Learning about AI — what is real, what is artificial, and why our children and grandchildren need to know the difference. My art therapy classes returning this June — four-week sessions on Thursday afternoons at one o'clock for teenagers and adults at my Delight in Living office. A Reflection for You So I will leave you with the same question I have been sitting with: what time are you in right now? Are you moving into a larger home or downsizing into a smaller one, planning travels, sorting old photos, or learning something brand new? Whatever season you find yourself in, it has been given to you on purpose, and there is grace waiting inside it. Take a quiet moment to name where you are, and trust that the next step forward is already being prepared for you. Listen, Connect & Support You can listen to Episode 296, "There Is an Appointed Time for Everything," and every past episode at lindagullo.com/minimiraclespodcast, or subscribe free through Apple Podcasts so you never miss a week. If you would like to register for my art therapy classes this June, or learn about counseling and life coaching, visit lindagullo.com or email me directly at linda@delightinliving.com. If this episode brings encouragement to your week, you can support its continued creation at buymeacoffee.com/delightinliving. Kindly share this podcast with someone who may be standing at a crossroads in their own season of life.

President Trump has said a deal with Iran is "largely negotiated" and that he will either sign "a great and meaningful" pact with Tehran, "or there will be no deal." Fifty thousand Southern California residents remain under evacuation orders Monday after emergency crews raced to prevent a tank holding a volatile industrial chemical from exploding at an aerospace facility in Garden Grove. Saturday's shooting near the White House has raised security concerns ahead of summer celebrations to mark America's 250th birthday. The shooter, who was killed after opening fire on a Secret Service checkpoint, had previously blocked a White House entry lane last June, court records show. To learn more about listener data and our privacy practices visit: https://www.audacyinc.com/privacy-policy Learn more about your ad choices. Visit https://podcastchoices.com/adchoices

Fifty days after Passover, something extraordinary happened that forever changed how we experience God. Listen as Rabbi Schneider uncovers the Jewish roots of Pentecost (Shavuot) and why the gift of the Holy Spirit is central to our faith.

Fifty-year type 1 veteran and health psychologist Dr. Beth Braun discusses managing diabetes burnout, overcoming food shame, and the clinical impact support groups have on lowering A1C. ABLEnow save for today's needs or invest for tomorrow Eversense CGM Medtronic Diabetes Tandem Mobi ** Use code JUICEBOX to save 20% at Cozy Earth  CONTOUR NextGen smart meter and CONTOUR DIABETES app Dexcom G7 Go tubeless with Omnipod 5 or Omnipod DASH * Get your supplies from US MED  or call 888-721-1514 Touched By Type 1 Take the T1DExchange survey Apple Podcasts> Subscribe to the podcast today! The podcast is available on Spotify, Google Play, iHeartRadio, Radio Public, Amazon Music and all Android devices The Juicebox Podcast is a free show, but if you'd like to support the podcast directly, you can make a gift here or buy me a coffee. Thank you! *The Pod has an IP28 rating for up to 25 feet for 60 minutes. The Omnipod 5 Controller is not waterproof.  ** t:slim X2 or Tandem Mobi w/ Control-IQ+ technology (7.9 or newer). RX ONLY. Indicated for patients with type 1 diabetes, 2 years and older. BOXED WARNING:Control-IQ+ technology should not be used by people under age 2, or who use less than 5 units of insulin/day, or who weigh less than 20 lbs. Safety info: tandemdiabetes.com/safetyinfo Disclaimer - Nothing you hear on the Juicebox Podcast or read on Arden's Day is intended as medical advice. You should always consult a physician before making changes to your health plan.  If the podcast has helped you to live better with type 1 please tell someone else how to find it!  

Republicans have played nice for decades: nobody mentions the human feces, the drug addicts everywhere, the destruction of the quality of life at the hands of left-wing crazies. It's all measured and polite and wonkish. Well, not anymore.   Sponsors: Citizen Portal: Citizen Portal makes all your local government meetings word searchable, so you can know who said what when. Fifty of my listeners are getting lifetime subscriptions at 50% off. Be one of them by going to citizenportal.ai/woods and using code WOODS50.   Agorist Tax Advice: Pick up a free copy of the brilliant Matthew Sercely's Agorist Tax Toolkit at: AgoristTaxAdvice.com/woods   Guest's Twitter: @martyrmade Show notes for Ep. 2760 The Tom Woods Show is produced by Podsworth Media. Check out the Podsworth App: Use code WOODS50 for 50% off your first order at Podsworth.com to clean up your voice recordings, sound like a pro, and also support the Tom Woods Show! My full Podsworth ad read BEFORE & AFTER processing: https://youtu.be/tIlZWkm8Syk