POPULARITY
Categories
741 episodes with tldr
255 episodes with tldr
98 episodes with tldr
144 episodes with tldr
62 episodes with tldr
47 episodes with tldr
53 episodes with tldr
32 episodes with tldr
40 episodes with tldr
36 episodes with tldr
16 episodes with tldr
16 episodes with tldr
38 episodes with tldr
13 episodes with tldr
13 episodes with tldr
7 episodes with tldr
5 episodes with tldr
17 episodes with tldr
17 episodes with tldr
7 episodes with tldr
4 episodes with tldr
15 episodes with tldr
6 episodes with tldr
8 episodes with tldr
6 episodes with tldr
4 episodes with tldr
8 episodes with tldr
7 episodes with tldr
8 episodes with tldr
5 episodes with tldr
13 episodes with tldr
5 episodes with tldr
4 episodes with tldr
5 episodes with tldr
3 episodes with tldr
8 episodes with tldr
4 episodes with tldr
5 episodes with tldr
3 episodes with tldr
10 episodes with tldr
3 episodes with tldr
4 episodes with tldr
9 episodes with tldr
8 episodes with tldr
4 episodes with tldr
2 episodes with tldr
4 episodes with tldr
4 episodes with tldr
2 episodes with tldr
5 episodes with tldr
5 episodes with tldr
3 episodes with tldr
3 episodes with tldr
4 episodes with tldr
Michele lost 120 pounds—but the deeper transformation was learning to live without the food noise, social pressure and old identity that had shaped her relationship with sugar for decades. Episode #273
The difference between being a researcher people rely on and a researcher people build withI have spent a lot of my career being the person who shows up to the meeting, sits at the table, and somehow still feels like a guest at it. You know the version of this, where everyone is aware research exists, they will happily call you when something needs validating, and yet none of the actual product decisions seem to happen anywhere near you. I have come to think this is one of the quietest career traps in our field, and I do not think we talk about it enough, so I sat down with Priyanka Kuvalekar, a Senior UX Researcher at Microsoft who leads research for Teams Calling and agentic AI experiences, to get into the part of the job that has nothing to do with method and everything to do with people.Priyanka has eight years in the field, a background as an architect before tech, and a way of describing cross-functional work that made me nod so hard I almost knocked the mic. This one is for anyone who has ever felt like the team's resident data nerd and wanted to be something more than that.What we get into:* A relationship means people know you are available, a partnership means people build with you. This was the line that organized the whole conversation for me. Priyanka draws the distinction cleanly by defining how a relationship is the PM passively knowing research exists and reaching for you when they want more of it, and a partnership is being invited into the room where the decisions actually happen, from the small “there's a bug, we need feedback” moments all the way up to “we have a brand new idea, where does research fit.” The shift she describes is from waiting to be invited to proactively shaping where the product is going, and I think a huge number of researchers are sitting in the relationship box right now without realizing there is another box entirely.* Sometimes the most strategic thing you can do is act like a therapist. Priyanka said she will literally sit with her cross-functional partners and work to understand what motivates them, what they are worried about, what keeps them up at night, where they see the product going, and what risks they genuinely want answers on. She called herself a therapist at one point, and I loved that, since the listening she is describing is not soft or passive, it is reconnaissance. When you actually know what your PM is afraid of and what your designers are carrying in terms of product history, you can position your research where it will land hardest, and you stop guessing at where to make an impact.* You get proactive by questioning the decision, not just delivering the study. The move Priyanka described that I want every researcher to steal is using the five whys on your own stakeholders. When a PM brings a decision built on generally available knowledge rather than verified sources, she asks why they are taking it, where the data came from, and what the consequences of being wrong actually are. She asks for the roadmap, the six-month plan, the reasoning behind the sequencing, and what comes out of that is the context you genuinely cannot make good calls without. The reframe here is knowing what your team is building for, not just what they have asked you to test.* The data does not change, the translation does. This is the one Priyanka named as her core takeaway, and it is the difference between a report that lands in a deck and a report that moves a decision. The same finding gets framed in tradeoffs and risk for PMs, in feasibility for engineers, and in business vision and risk-of-inaction for executive leadership, all without diluting what the research actually says. She pairs this with a research point of view in every report, where the insight comes with a recommendation, a severity rating, and a clear justification of why a high is a high and what happens if the team ignores it. Translating is not softening, it is meeting each person in the language they already think in.* Tell people how NOT to use your research, out loud, every time. This is the bit I have rarely heard anyone articulate as clearly as Priyanka did. She has watched teams take one insight out of an entire study and bend it into whatever they already wanted to do, and she has seen people pull inspiration from unrelated reports in other departments and treat it as evidence for their own product. Her answer is to keep evangelizing after the readout, give a clear TLDR of what the research says, and then state plainly what the research does not say. Disclaimers on how not to use the data sound almost paranoid until you have had a stakeholder cheerfully misquote you in a leadership meeting, and then they sound like the most reasonable thing in the world.* You can say no without saying no, and you can stay visible without becoming a yes-researcher. We both confessed to the same arc, going from reactive to overcommitted to burnt out, and Priyanka had the most practical fix I have heard for it. When her team was packed for four months on foundational work and the usability and “does this button work” requests kept coming, she opened a 30-minute weekly research office hour where designers could bring quick questions and she would help launch something low-lift on a testing platform, which built real goodwill without blowing up her roadmap. She pairs that with weekly check-ins where she shares even tiny progress, so the question her team asks shifts from “what is that researcher even doing” to “what is coming next from them,” and that small reframe is what buys her the time to do thorough work without disappearing for eight weeks at a stretch.None of this works as a one-time performance. The therapist conversations, the office hours, the weekly progress updates, the disclaimers on every readout, all of it is maintenance work that you have to keep doing even on the weeks you are slammed and would rather hide behind a study. Priyanka has been sharing weekly updates for over a year, and I think that consistency is the actual product, not any single clever framing. If you take one thing from this episode, take the relationship-to-partnership distinction, and then go do the unglamorous, repeated work that earns the second one.Watch or listen to the full conversation above. Priyanka is genuinely lovely and far more generous with her hard-won lessons than she needed to be, so go give her a follow.Find Priyanka: * LinkedIn* InstagramInterested in sponsoring or advertising on this podcast? I'm always looking to partner with brands and businesses that align with my audience. Book a call or email me at nikki@userresearchacademy.com to learn more about sponsorship opportunities!The views and opinions expressed by the guests on this podcast are their own and do not necessarily reflect the views, positions, or policies of the host, the podcast, or any affiliated organizations or sponsors. This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.userresearchstrategist.com/subscribe
This week on The First Degree, we're doing something a little different. Consider this episode your CHAOS crash course. Tom O'Neill's groundbreaking book CHAOS: Charles Manson, the CIA, and the Secret History of the Sixties has sparked years of conversation, debate, and controversy. But with this week's Deep Dive featuring an in-depth conversation with Tom himself, we wanted to make sure everyone could join us, whether you've read the book or not. In this episode, we break down the major players, the official Manson narrative, the questions O'Neill spent more than two decades investigating, and the evidence that turned CHAOS into one of the most talked-about true crime books of the last decade. Whether you're completely new to CHAOS or just need a refresher before our interview with Tom O'Neill, this is your TLDR.
We start this episode with a movie review (I'm sorry, what?), before providing an update on the Diving Playhouse saga we brought you two weeks ago. The TLDR is that the lease has been terminated, with thousands turning up to Hyde Park in Sydney to protest the decision.We then discuss what we can do to save our music venues from closing, with the news that MoshPit Bar will be closing in Sydney. It was one of just 19 venues in the state that put on 100 shows or more last year, so year, it's a pretty devastating blow.Its closure also raises questions over grant funding, with the venue receiving more than $150,000 in state and federal grants over the last few years, and still closing down.We pose more questions than answers in this one, but we hope you're along for the ride.Also please remember to rate us five stars if you can be bothered. Thank you
The cross-national correlation between gender equality and lower fertility is exceptionally strong (r ≈ 0.81). After the 1960s, a unique mating regime spread across parts of the world—with female emancipation, individual mate choice, and effective birth control—followed by a continuing rise in singlehood and declining fertility. Almost all women still want to reproduce, but many struggle to find a good-enough partner. This article argues from an evolutionary perspective that many men's utility to “free women” has been so diminished that solving the fertility crisis by increasing pair-bonding rates seems unfeasible. A viable means for aiding the survival of low-fertility nations could be to provide women with the economic and social resources necessary for them to conclude that having children alone makes for a better life than remaining childless. Such policies would likely exacerbate male marginalization, but new technologies are on the horizon that could offer men reproductive equality.
The stretch of TLDR main line episodes continues with a DC title and a Mad Cave OGN that's getting a re-release! Plus, all your weekly reviews!
This week on Overtime on Inferno, we try to avoid talking about BLAST Bounty for as long as possible. To the surprise of no one, online CS still sucks.Mini-games are back with another edition of TLDR tenable, and we also create our free agent dream teams. Though none of the rosters we built could ever compete with this new Liquid team led by JT. Yes, you heard us right, Liquid genuinely look fantastic right now.Join the discord:https://discord.gg/X3jU4djxUK
We would love to connect with you and you can send us a message here.
The stretch of TLDR main line episodes continues with an IDW title and an Oni title that's set to return in August! Plus, all your weekly reviews!
Sign up to TLDR Party and get episodes of Andymonium every single fortnight: https://toolong.news/s/burnhamThe UK has ANOTHER prime minister, so you bet TLDR are going to talk about it! Welcome to the first episode of Andymonium! Ben and Georgina are here to run through everything that has been going on in the last week of insane UK politics.
We would love to connect with you and you can send us a message here.
On this episode of Run the Numbers, CJ breaks down Cumberland Farms' IPO filing and the surprising business behind it.—SPONSORS:RightRev is an automated revenue recognition platform that lets your product team ship new pricing without asking finance for permission, and your sales team close deals without creating downstream chaos. Check out their free tool at calculator.rightrev.com It scores your rev rec process, shows what's exposing you to risk, and tells you exactly where to focus before it bites you in the rear end. Check it out at https://calculator.rightrev.comPulley is an equity management platform that lets you issue options, model dilution, and complete 409As without your cap table turning into a spreadsheet disaster. Founders raising, hiring, and scaling use Pulley to keep equity clean and stay focused on building. Learn more or request a demo at https://pulley.com/mostlymetricsRillet is an AI-native ERP built for modern finance teams that want to replace NetSuite and close faster. With revenue recognition, close management, multi-entity support, and native Stripe and Salesforce integrations, Rillet helps scaling companies run their finance stack in one place. Hundreds of teams, including Windsurf and Mercor, use Rillet to make the zero-day close real. Book a demo at https://www.rillet.com/cjMaximor is an autonomous finance platform that runs order-to-cash, procure-to-pay, the close, cash management, and reporting on self-learning agents instead of a dozen disconnected tools. One PE-backed customer cut their close in half, took audit findings from seven to zero, and cut back-office costs by 70% in six months. You pay for outcomes, not seats. See it at https://www.maximor.ai/Brex is an intelligent finance platform with AI-powered agents that capture expenses automatically, enforce policy before the spend happens, and close your books in minutes instead of weeks. 35,000+ companies like OpenAI, Coinbase, Anthropic, and DoorDash already run on Brex. It's time to get Brex AF. Learn more at https://www.brex.com/metricsAnrok is the sales tax platform that watches your exposure everywhere, automates compliance, and flags risk before it turns into a surprise back-tax letter from a state you've never set foot in. Companies like Anthropic, Notion, and Vanta already trust Anrok to stay ahead of rules that move faster than any spreadsheet can. Talk to a sales tax expert for a personalized exposure estimate at https://www.anrok.com/rtn—LINKS: Mostly Talent: https://mostlymetrics.typeform.com/to/cLTxtAsNCJ: https://www.linkedin.com/in/cj-gustafson-13140948/Mostly metrics: https://www.mostlymetrics.com—TIMESTAMPS:0:00 Preview and Intro2:57 Key metrics4:28 Europe now out-earns America5:01 $5.8B debt at 8x leverage6:20 TLDR: snacks, gas, borrowed money7:58 Fuel vs. inside: the gross profit photo finish8:58 COCO, CONCO, and OTHER explained10:57 Sponsors — Maximor | Brex | Anrok14:02 Why scale matters: four reasons16:02 The US convenience store market17:42 EVs: long the birth rate18:05 Back to the debt19:10 The sale leaseback20:24 Sponsors — RightRev | Pulley | Rillet23:26 Does the deleveraging math work?24:27 The growth thesis25:06 600–700 stores become Cumberland Farms26:16 The chicken fryer ROI27:00 Coffee and loyalty: 6M members in 13 months28:15 Red flag 1: five-for-five COSO failures28:41 Red flag 2: entire C-suite is new29:04 Red flag 3: lending to their own parent29:42 Red flag 4: Big Tobacco funds loyalty29:57 Red flag 5: foreign issuer, Cayman charter30:10 Red flag 6: one supplier is 31% of costs30:32 Red flag 7: balance sheet is stale30:45 Cap table: TDR Capital and the Issa brothers31:55 Casey's vs. ARCO33:47 Goldman got bumped34:08 Massachusetts banned the pump clip until 201534:57 No health insurance disclosure for 16K US workers35:30 Robert Swan (ex-Intel CEO) is on the board36:08 Credits
We would love to connect with you and you can send us a message here.
We would love to connect with you and you can send us a message here.
Camp SundayWe would love to connect with you and you can send us a message here.
TWiV explains research showing how temperate phages enhance bacterial host fitness via RNA-guided flagellar remodelling, and that IgA antibody is necessary and sufficient to prevent norovirus infection in mice. Hosts: Vincent Racaniello, Rich Condit, Brianne Barker, and Jolene Ramsey Subscribe (free): Apple Podcasts, RSS, email Become a patron of TWiV! Links for this episode Support science education at MicrobeTV Positions in Rosenfeld Lab (email) Fred Murphy passes away (X) US has lost its hold on measles elimination (CIDRAP) What's in the genetic code of America's measles outbreaks (ProPublica) Did Pete Hegseth single-handedly raise the R₀ of flu? (Inside Medicine) Fred Murphy's Virus Images and Foundations of Virology (UTMB) Temperate phages and RNA-guided flagellar remodelling (Nat Micro) Eukaryotic RNA guided system (Nature) IgA prevents norovirus infection in mice (Sci Transl Med) Craig Wilen's letter to TWiV (TWiV 492) Timestamps by Jolene Ramsey. Thanks! Picks of the Week Brianne – APod June 26: Milky Way Urban Style Rich – Your birth order affects your future, but not for the reason you think; Dr. McCoy Jolene – Bakteriopolis paper and site Vincent – The Claude Shutdown Is a Total Sh*tshow Listener Pick Martin – Ancient text charred by Vesuvius read for first time in 2,000 years Intro music is by Ronald Jenkees Send your virology questions and comments to twiv@microbe.tv Content in this podcast should not be construed as medical advice.
You can get a perfect workout plan and still never get in shape. The science is right there. The exercises are correct. But knowing what to do and actually doing it when you don't feel like it are two completely different problems. That's why personal trainers exist — not to hand you better information, but to make sure you show up and follow through. Investing works the same way. In this episode of Off The Wall, David B. Armstrong, CFA and Nate Tonsager, CFA, CIPM make the case that behavior — not strategy — is the real variable in investment returns. They break down dollar-cost averaging as a structural accountability tool, walk through what the Vanguard lump-sum data actually says (and the hidden assumption most people skip over), and show how pre-set triggers turn market sell-offs into buying opportunities instead of reasons to freeze. The TLDR: the best investment strategy isn't the one with the highest expected return. It's the one you'll actually stick with when the market is down and your instincts are telling you to do something about it. Because in investing, you don't win in the spreadsheet. You win in the chair. Please see important podcast disclosure information at https://monumentwealthmanagement.com/disclosures Episode Timeline/Key Highlights: 0:33 — Vinyl And Setup 2:42 — Volatility Makes Cash Decisions Hard 3:06 — Dollar Cost Averaging Explained Simply 5:35 — Why 401(k) Autopilot Works 6:57 — Vanguard Data And The Hidden Assumption 10:10 — DCA Triggers To Buy Selloffs 12:30 — Experience Changes How Pullbacks Feel 15:06 — Stop Hunting For The Perfect Bottom 17:50 — Missing Best Days And Political Noise 24:20 — Time Horizons And Cash As An Asset 29:57 — Dalbar Study And The Value Of Advice 24:20 — Wrap-Up Plus AMA Call For Questions Connect with Monument Wealth Management: Visit our website: https://monumentwealthmanagement.com/ Follow us on Instagram: https://www.instagram.com/monumentwealth/# Connect on LinkedIn: https://www.linkedin.com/company/monument-wealth-management/ Connect on Facebook: https://www.facebook.com/MonumentWealthManagement Connect on YouTube: https://www.youtube.com/user/MonumentWealth#Fit Subscribe to our Private Wealth Newsletter: https://monumentwealthmanagement.com/subscribe/ Check out our Between Sips Podcast: Where Money Meets Meaning Because money without meaning never feels like wealth. https://monumentwealthmanagement.com/between-sips-podcast/ About "Off the Wall": Markets are noisy. Your time is limited. Off The Wall cuts through the clutter. Hosts David B. Armstrong, CFA and Nate Tonsager, CFA, CIPM bring you straightforward, candid insights about what's really moving markets and why it matters for successful investors. From economic shifts to portfolio positioning, we break down the complexities so you can invest with intention and stay grounded when headlines and life feels chaotic. Learn more about our hosts on our website at https://monumentwealthmanagement.com
Starmergeddon is BACK?! Maybe! Hopefully! If 400 of you sign up to TLDR Party, TLDR will bring back their UK politics podcast over on TLDR Party, but only if you sign up using THIS LINK: https://toolong.news/s/burnhamFor now though, Ben and Georgina run through what the hell has actually happened leading up to Starmer's resignation, and what's next for Westminster.
We would love to connect with you and you can send us a message here.
Have you stopped checking the headlines in the morning? Are you changing the channel once the news comes on? Do you find yourself feeling anxious when hearing about current events? If so, you're in packed company. This year's Digital News Report found that engagement and trust in legacy media continues to fall, as online content creators and AI dominate digital platforms. Not only has the medium changed, but also interest. This year's report found that 45% of Canadian respondents actively avoid the news. Host Catherine Jette speaks to Craig Robertson, a co-author of the Digital News Report to discuss why Canadians are turning away from the headlines, and how media companies can keep up as interest peaks elsewhere. We love feedback at The Big Story, as well as suggestions for future episodes. You can find us:Through email at hello@thebigstorypodcast.ca Or @thebigstory.bsky.social on Bluesky
The Senior Care Industry Netcast w/ Valerie V RN BSN & Dawn Fiala
Send us Fan MailGoogle is starting to answer home care questions for families before they ever click a website, and that shift changes everything about how we market, write, and respond. We walk through what “practical AI” really looks like for a home care agency that still wants to lead with empathy, trust, and real human guidance, not a cold automated experience.We share how we're using AI for website development and content writing as a first draft, then tightening it with human review so it doesn't sound generic, miss local details, or overpromise. We dig into the questions families ask most often like home care cost, payment options, and whether Medicare or Medicaid pays for home care and explain why your site needs clear, easy-to-quote answers. You'll also hear why TLDR summaries, structured FAQ pages, and better long-tail keyword content can improve visibility in Google AI Overviews and other AI search tools, helping your agency get referenced first.Then we get hands-on with AI chat for home care websites: how it can respond instantly after hours, capture inquiry details, and forward full chat transcripts to the right person on your team. We lay out the guardrails that matter most including no medical advice, no pricing quotes, no caregiver availability promises, and clear emergency messaging. Finally, we talk about AI phone answering and where it may fit best as overflow support and call routing, plus why you should start small, test carefully, and keep humans in the loop.Subscribe for more practical home care marketing and sales training, share this with an agency owner who's curious about AI, and leave a review with the biggest question you want AI to help you handle.Continuum Mastery Circle IntroVisit our website at https://asnhomecaremarketing.comGet Your 11 Free Home Care Marketing Guides: https://bit.ly/homecarerev
This week on Two Parents & A Podcast, happy Wednesday (so weird it's not Thurs?!) - we are OFFICIALLY 3 episodes a week… AND tomorrow we officially launch our SUBSTACK!!!
We would love to connect with you and you can send us a message here.
Send us Fan MailThe same friction that quietly kills conversions on your group sales page is killing your reps' email responses too — same problem, just a different channel. In Part 2 of the Group Sales 101 series, Jeremy Neisser walks through the four-question friction test for cold outreach, the caveman test that exposes a hard-to-read email in 10 seconds, and a real-world rewrite that tripled one rep's response rate without changing the offer. Practical, tactical, and ready to apply to your team's outreach this week.KEY TOPICS COVERED- How cognitive load starts in the inbox — not on your website — and why most reps miss it- The four questions every cold group sales email must answer in under 10 seconds- Why long emails get deleted, not read, by busy HR directors and office managers- Writing first lines that signal relevance to specific buyer personas- Using 3–4 short bullets instead of paragraphs to describe what's included- The case for showing pricing early — and the response-rate data that backs it up- One email, one ask: writing CTAs that get answered in 10 seconds- The caveman test — a 10-second readability gut check that flags friction fast- A real example: cutting one rep's 5-paragraph cold email to 3 sentences tripled her response rate- Using Claude or ChatGPT to stress-test your outreach against the friction test before you sendTIMESTAMPS[00:00] – Welcome and recap of Episode 168 on group sales personalization[00:35] – Why this episode builds on Episode 154 (reducing friction on your group sales page)[01:10] – The four buyer questions: Is this for me? What do I get? What does it cost? What's next?[01:55] – Same friction, different channel: applying website thinking to email outreach[02:39] – How cognitive load starts at the email or voicemail — not the landing page[03:05] – Anatomy of a typical group sales cold email — and why it fails[03:35] – TLDR: when buyers do mental gymnastics, they don't respond[04:02] – Real example: a 250-company outreach that produced only 5–10 responses[04:35] – How cutting that email to 3 sentences tripled the response rate[05:01] – The four-question test for your cold emails[05:30] – Writing a relevant first line: HR directors vs. youth pastors vs. corporate buyers[06:24] – Question 2: What do I get? Use 3–4 bullets, drop the jargon[07:23] – Why "dedicated group area" loses to "everyone sits together"[08:00] – Question 3: What does it cost? Why pricing transparency increases responses[08:19] – When pricing is missing, buyers fill the gap with a number that's too high[09:00] – A pricing A/B test that proved transparent pricing wins[09:42] – Question 4: What do I do next? Why most CTAs fail[10:11] – "Let me know if you're interested" is not a call to action[10:45] – One email, one ask: writing low-friction CTAs[11:39] – The caveman test: 10-second clarity check[12:08] – Real example: handing a cold email to a kitchen worker, groundskeeper, and usher[13:03] – Using Claude or ChatGPT to run the friction and caveman tests on your emails[14:00] – Episode takeaways[14:28] – Preview of Part 3: how great reps place groups strategically[15:50] – Why confidence and small wins change the culture of a sales team[16:42] – Closing CTA: ratings, reviews, and how to get in touchCALL TO ACTIONIf group sales is on your plate this season, share Episode 169 with your reps and run their current cold email through the four-question friction test together. Questions or want a second set of eyes on your outreach? Reach out at sportsmarketingmachine.com to schedule a call.QUOTE PULLSJeremy Neisser: "Confused people don't respond."Jeremy Neisser: "The long email isn't a thorough email. It's a hard email. And hard emails don't get read."Jeremy Neisser: "When there's no pricing at all, the buyer's brain fills in the gaps. And almost every single time, they're going to come up with a number that's too high."Jeremy Neisser: "One email, one ask. The more decisions you force a buyer to make, the less likely they are to make any of them."Links mentioned:Episode 168 - Group Sales 101 — Part 1- Personalization Wins Group SalesEpisode 154 - How to Make Your Group Sales Page Easier to Buy FromEpisode Web pageYour Reps Should Be Closing… Not ProspectingMost teams only reach about 20% of their local market.The other 80%?They're businesses, churches, schools, youth teams, and organizations that simply haven't been contacted yet.That's where we come in.We become an extension of your team, delivering qualified group sales opportunities directly to your reps.Fill out the form on this page and let's see if we can help.Episode PageRegister for the upcoming July 23 webinar: REGISTERRevelocity Sports Sports Marketing Machine on LinkedInSports Marketing Machine on InstagramBook a call with Jeremy from Sports Marketing Machine
Returning to the show — but appearing on TLDR for the first time as a pair — Zack Kaplan and John J. Pearson discuss their upcoming title at Dark Horse, THE SMART DIVISION.Final Order Cutoff Date: June 29Release Date: August 5
All the insight and fun of GameFace in 1/3 of the time! Analysis of the most crucial games at the Xbox Showcase, Summer Game Fest, and the Nintendo Direct!
We would love to connect with you and you can send us a message here.
The TLDR here is simple: 'BEEF' Is the Best Show on Netflix right now.But hop in anyway because Ebube's summoned AJ and Soonen for a union that's decidedly NOT fueled by road rage or deteriorating romance. The trio get into everything from who's actually in the right this season, to what the show has to say about the systems that feed the human conditionThey talk about how this season is as much about class warfare as individual conflicts (4:27), in line with what this season of the show is truly about (6:31); as well as the richness of the performances (9:49), Josh and Lindsay's delusion (19:49), how power changes people (23:35) and why it's important for young men to have solid role models (30:02).Art does mirror life after all.You can support us here.Also available on YouTube.Host: Ebube UbochiGuests: AJ & SoonenProduction by: Ebube Ubochi
Everybody has books, emails, and even text threads they intended to read but never got around to because it just felt like too much. So, you skimmed it. You got a sense of it. You can fake your way through a conversation about it. But it feels like so much work to dive in and dissect it. And that might be ok with a high school reading assignment, a text thread with golf buddies, or an all-staff email from HR. But what about a message from God, explaining the purpose of life and how to get the most out of it. Isn't that what the Bible is? Maybe. But where do you start and how do you make sense of it? Is it even possible for an average person to get something significant out of this ancient book on a daily basis?
We would love to connect with us and you can send us a message here.
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,
People say reading the Bible changed their life. But how? You've been doing it for a while, and it hasn't changed much of anything. You downloaded the app. You subscribed to the verse of the day. You bought a study Bible. You've even highlighted a couple things. And all that is great. But you still struggle with anxiety. You're still addicted to your phone. You still yell at your kids from time to time. You still wonder if you've wasted your life every year on your birthday. I thought Scripture was supposed to change you, turn you into a new person, and make your life better. Why isn't it working for you? Is the whole thing a scam? Are you just doing it wrong? What gives? If you're new with us, let us know how we can be praying for you, we invite you to fill out an online Connect Card by visiting https://southhillschurch.churchcenter.com/people/forms/91550—If you are looking for what is next for you, we invite you to fill out an online “Next Steps” card by visiting https://southhillschurch.churchcenter.com/people/forms/672517To give with us select the Give tab on the Church Center App or visit https://southhills.org/giving/ and select the Corona Fund or Corona BOW Fund—Visit our Linktree to find out more about everything mentioned in today's message or follow along with the message slides:https://linktr.ee/SouthHillsCorona —To RSVP for On-Campus Events select the Events tab on the Church Center App or visit https://southhills.org/corona/ TL:DR - Guest Speaker: Eric West “Why Isn't It Working?” 5.31.26 -
Everybody has books, emails, and even text threads they intended to read but never got around to because it just felt like too much. So, you skimmed it. You got a sense of it. You can fake your way through a conversation about it. But it feels like so much work to dive in and dissect it. And that might be ok with a high school reading assignment, a text thread with golf buddies, or an all-staff email from HR. But what about a message from God, explaining the purpose of life and how to get the most out of it. Isn't that what the Bible is? Maybe. But where do you start and how do you make sense of it? Is it even possible for an average person to get something significant out of this ancient book on a daily basis?
All the fun and insight of GameFace in 1/3 of the time with improved production values! Reviews of LEGO Batman: Legacy of the Dark Knight and Yoshi and the Mysterious Book! Plus, Not E3/Summer Game Fest 2026 previews for PlayStation, Xbox, Nintendo, and third-party!
Hilary Dubin is co-CEO and head of Jones' digital product & behavioral support program. She graduated from the University of Pennsylvania magna cum laude, majoring in cognitive science with a concentration in computation and cognition, an honors thesis on the effects of gender, realism, and role of virtual agents, and a minor from Wharton in consumer psychology. She worked in David Brainard's visual neuroscience lab for 3 years and published 4 papers and supplementary materials on illumination discrimination (color perception). After Penn, she was selected as one of ten Americans to be a Ventures Fellow in the Excel Ventures incubator program in Tel Aviv, and continued on to be the inaugural member, and later program lead, of the US Associate Product Manager Program at Atlassian. She worked as a product manager at Atlassian for 5 years, ultimately as Head of Confluence Editions & Admin Experience where she launched Confluence Premium & Free into multi-million dollar product offerings with 2M+ users. She hired & managed two PMs and lead a team of over 30 developers. Prior to founding Jones, she and Caroline founded Cozier together, a sleep & loungewear brand designing ethical, effortlessly chic garments for every/body. Hilary started vaping casually in 2017 when the JUUL seemed relatively harmless and fun. When the world went on lockdown in 2020, her casual vaping habit became a daily crutch for coping with stress and working from home. After over a year of unsuccessful cold-turkey quit attempts, she finally kicked her vaping habit in 2022 when Caroline suggested she try NRT. Outside of work, Hilary loves hiking, backcountry skiing, trying to find the best burger in NYC, and playing with other people's dogs. In This Conversation We Discuss: [00:00] Intro [02:34] Creating products from personal pain points [06:52] Sponsor: Klaviyo [08:59] Meeting potential customers where they are [10:47] Adapting products based on user feedback [13:48] Testing market demand with waitlists [16:02] Sponsor: Electric Eye [17:10] Maximizing personal networks for growth [18:34] Gathering behavioral data in early days [19:52] Callouts [20:02] Launching a product to engaged audiences [22:09] Sponsor: Intelligems [24:09] Pivoting marketing to bridge early limitations [26:24] Driving organic traffic with relatable content [30:33] Adding modern value to traditional products Resources: Subscribe to Honest Ecommerce on Youtube Nicotime mints and social app to quit vaping quitwithjones.com/ Follow Hilary Dubin linkedin.com/in/hilary-dubin-374156b4/ Follow Caroline Vasquez Huber linkedin.com/in/caroline-vasquez-huber Book a demo today at intelligems.io/ Schedule an intro call with one of our experts electriceye.io/connect Get your free demo klaviyo.com/honest If you're enjoying the show, we'd love it if you left Honest Ecommerce a review on Apple Podcasts. It makes a huge impact on the success of the podcast, and we love reading every one of your reviews!
Nobody accomplishes anything truly impressive overnight. They do it over time. I've never heard of a doctor who finished med school in one sitting. Or a gold medal gymnast who picked up the pommel horse that morning. No. Each of these people broke their big goal into smaller steps, and inched their way forward, little by little. It wasn't accidental. It was intentional. And if you want Scripture to change to become the basis for how you think, talk, and live, it's going to happen in much the same way. There's a path to becoming a doctor. There's a path to becoming an Olympian. What's the path to becoming a Jesus follower reliant on Scripture? And is it realistic for someone like you?If you're new with us, let us know how we can be praying for you, we invite you to fill out an online Connect Card by visiting https://southhillschurch.churchcenter.com/people/forms/91550—If you are looking for what is next for you, we invite you to fill out an online “Next Steps” card by visiting https://southhillschurch.churchcenter.com/people/forms/672517To give with us select the Give tab on the Church Center App or visit https://southhills.org/giving/ and select the Corona Fund or Corona BOW Fund—Visit our Linktree to find out more about everything mentioned in today's message or follow along with the message slides:https://linktr.ee/SouthHillsCorona —To RSVP for On-Campus Events select the Events tab on the Church Center App or visit https://southhills.org/corona/ TL:DR - Adam Smith “From Roulette To Routine” 5.24.26 -
Join Alison and TL;DR as we look at feminism, the only answer to the chaos of our times!
All the fun and insight of GameFace in 1/3 of the time! Reviews of Forza Horizon 6, Subnautica 2, and Directive 8020! Plus, PlayStation officially leaves PC, remakes of Halo 2 and Halo 3 are on the way, and much more!
Trail TMZ is back.Host and award-winning correspondent Allison Mercer dive into one of the strangest weeks of trail running that we have seen in a while. From the Satisfy and Adidas backlash and influencer culture debates, to doping discussions around Cam Hanes, Sage Canaday and clean sport, plus the growing role of social media in shaping running culture.We also talk FKTs, Will Peterson's Appalachian Trail attempt, upcoming Pacific Crest Trail action, why controversy dominates attention online, and whether running is losing the things that made it special in the first place.Topics:• Satisfy backlash and brand culture• Influencer running and authenticity• Cam Hanes, Sage Canaday, and doping conversations• Why negativity dominates social media• FKT updates and upcoming attempts• Will Peterson's Appalachian Trail• The future of trail running cultureSupport our Sponsors: Janji (code: Freeoutside): https://snp.link/a0bfb726CS Coffee: CSinstant.coffeeGarage Grown Gear: https://snp.link/db1ba8abSubscribe to Substack: http://freeoutside.substack.comSupport this content on patreon: HTTP://patreon.com/freeoutsideBuy my book "Free Outside" on Amazon: https://amzn.to/39LpoSFEmail me to buy a signed copy of my book, "Free Outside" at jeff@freeoutside.comWatch the movie about setting the record on the Colorado Trail: https://tubitv.com/movies/100019916/free-outsideWebsite: www.Freeoutside.comInstagram: thefreeoutsidefacebook: www.facebook.com/freeoutside#Trailrunning #Runningnews #Outdoors #Outdooradventure
This is a free preview of a paid episode. To hear more, visit www.theflyingfrisby.comThere's a lot more to AI than software. AI requires electricity, transformers, substations, cooling systems, data centres and more. That all means copper. Lots and lots of copper.Right on cue, the copper price hit fresh highs last week at $6.68/lb, before pulling back. So today I am going to take a long overdue look at copper. Was last week's action just a spike that will soon fade away? Or was it part of something much bigger? TLDR - the second one.Let's start with a 50-year chart to give you some historical context.Copper peaked in the great inflationary blow-off of 1980, before spending the next twenty years doing essentially nothing. The 1980s and 1990s were an age of globalisation, disinflation and cheap commodities. Who cared about hard assets or mining? Then came the rise of China and the supercycle of the 2000s. China urbanised, industrialised and turned itself into a superpower. Copper exploded higher, peaking in 2011. That boom then gave way to a long hangover. The 2010s were dominated by tech stocks. Mining died a death. To survive mining companies cut capex, reduced exploration and focused on balance sheet repair rather than growth. That decade of underinvestment laid the foundations of the shortages being revealed today.Meanwhile, while investors were busy buying software companies and meme stocks, the world quietly decided it wanted to electrify everything.The really striking thing about the chart is the speed of the rallies when they come. Then the amount of time copper spends going nowhere.Now here's the ten-year chart, with the one-year moving average in red and the 55-day moving average in blue. To my eye, copper appears to have formed a major bottom in 2020 during the Covid panic. The violent correction in 2022 increasingly looks like an early-cycle shakeout.Technically, the chart is undeniably bullish. Copper is trading above both moving averages, both of which are rising strongly. Momentum remains positive.That said, in the short term, the metal does look extended. Sentiment has become hyper bullish. Every investment bank now seems to have a copper supercycle note. Type “copper” into X and see what comes up: we are going to the moon on a copper superjet (powered by electricity natch).Now here's the three-year chart, to which I've added the 50- and 200-day moving averages and the RSI. The trend is your friend, and it is up.Historically, copper tends to be seasonally weaker over the summer months, and this is a spiky chart within its uptrend. I think we see some range-trading and consolidation over the summer months, which will provide something of a buying opportunity. But the charts are only half the story.The more interesting question is why copper may be entering an entirely new structural era.
People say you should read the Bible, but have you tried? Seriously. Ever sat down, opened it, started reading, and felt completely lost in under two paragraphs? What is this even talking about and what are you supposed to do with it? Meanwhile, everyone online is quoting it to and at each other, to support whatever personal agenda they have, trying to prove they're right and everyone else is wrong. Maybe you're better off just staying out of it. Can't the Bible just be used to back up whatever it is you already think, or does it have an ultimate point? And if so, what is it? If you're new with us, let us know how we can be praying for you, we invite you to fill out an online Connect Card by visiting https://southhillschurch.churchcenter.com/people/forms/91550—If you are looking for what is next for you, we invite you to fill out an online “Next Steps” card by visiting https://southhillschurch.churchcenter.com/people/forms/672517To give with us select the Give tab on the Church Center App or visit https://southhills.org/giving/ and select the Corona Fund or Corona BOW Fund—Visit our Linktree to find out more about everything mentioned in today's message or follow along with the message slides:https://linktr.ee/SouthHillsCorona —To RSVP for On-Campus Events select the Events tab on the Church Center App or visit https://southhills.org/corona/ TL:DR - Adam Smith “Left Unread” 5.17.26
All the insight and fun of GameFace in 1/3 of the time! It's GameFace TLDR! GameFace is back with special guest Stevens Charles from LS Cream! Star Fox for Switch 2 controversy, reviews of Saros, Mixtape, Invincible VS, and Aphelion, the Switch 2 price increase, and much more!
Join Alison and TL;DR as we look at feminism, the only answer to the chaos of our times!
All the fun and insight of GameFace in 1/3 of the time! Assassin's Creed Black Flag Resynced, Mass Effect-like The Expanse, the new Steam controller, a pick-your-own-games Game Pass plan, Nintendo is sued over Trump's tariffs, Tomodachi Life, Vampire Crawlers, and much more!
GameFace in 1/3 of the time with improved production values! It's GameFace TLDR! We review Pragmata, Mouse: P.I. for Hire, and Replaced! Plus, Game Pass loses Call of Duty and gets a price decrease, new details on The Last of Us Part III and the next God of War, Metro 2039, FromSoftware's upcoming films, and much more!
This episode begins with reactions to Justin Bieber's weekend 2 Coachella performance. TLDR: we are not well. Then, an update on the Alix Earle/Alex Cooper situation, and an honorable mention to the Kim/Lewis soft launch and Dylan Sprouse. Links:https://x.com/judrewstin/status/2044625616608088284?s=42https://www.instagram.com/p/DXVI72_AXKc/?igsh=MTFrODVjY2ZiOWN0dQ==https://www.tiktok.com/t/ZP8gCR7BM/https://www.tiktok.com/t/ZP8gCJdoe/https://www.tiktok.com/t/ZP8gCNsyX/https://www.instagram.com/p/DXVLCx1CeM5/?igsh=MWswYXBtamQxY255YQ==ShopMy: https://shopmy.us/shop/commentsbycelebsCodes: SKIMS.com - after you place your order, be sure to let them know we sent you! Select "podcast" in the survey and be sure to select our show in the dropdown menu that followsShopify - Sign up for your one-dollar-per-month trial today at Shopify.com/commentsWatch the new season of Vanderpump Villa on Hulu and Hulu on Disney+ for bundle subscribersLearn more at Starbucks.com/partnersSee Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.
Listen If You Want to…. Couchella vs Coachella A catchy song that turned out to be AI BS Our questions about Artemis Why Testaments is a must watch And a TLDR need to know. See omnystudio.com/listener for privacy information.
Listen If You Want to…. Couchella vs Coachella A catchy song that turned out to be AI BS Our questions about Artemis Why Testaments is a must watch And a TLDR need to know. See omnystudio.com/listener for privacy information.
Listen If You Want to…. Couchella vs Coachella A catchy song that turned out to be AI BS Our questions about Artemis Why Testaments is a must watch And a TLDR need to know. See omnystudio.com/listener for privacy information.
© 2020-2026 Ivy Podcast Discovery LLC
