POPULARITY
Categories
703 episodes with Pip
244 episodes with Pip
278 episodes with Pip
92 episodes with Pip
74 episodes with Pip
57 episodes with Pip
66 episodes with Pip
49 episodes with Pip
48 episodes with Pip
71 episodes with Pip
47 episodes with Pip
34 episodes with Pip
43 episodes with Pip
37 episodes with Pip
37 episodes with Pip
103 episodes with Pip
22 episodes with Pip
18 episodes with Pip
22 episodes with Pip
15 episodes with Pip
17 episodes with Pip
58 episodes with Pip
30 episodes with Pip
44 episodes with Pip
22 episodes with Pip
33 episodes with Pip
22 episodes with Pip
22 episodes with Pip
13 episodes with Pip
17 episodes with Pip
17 episodes with Pip
38 episodes with Pip
23 episodes with Pip
20 episodes with Pip
8 episodes with Pip
21 episodes with Pip
6 episodes with Pip
14 episodes with Pip
13 episodes with Pip
14 episodes with Pip
19 episodes with Pip
14 episodes with Pip
8 episodes with Pip
5 episodes with Pip
11 episodes with Pip
18 episodes with Pip
16 episodes with Pip
10 episodes with Pip
10 episodes with Pip
7 episodes with Pip
10 episodes with Pip
8 episodes with Pip
9 episodes with Pip
13 episodes with Pip
17 episodes with Pip
16 episodes with Pip
6 episodes with Pip
11 episodes with Pip
30 episodes with Pip
8 episodes with Pip
![[Fanfic]Fanfic](https://ivyfm.s3.amazonaws.com/i320/633554.jpg)
20 episodes with Pip
5 episodes with Pip
4 episodes with Pip
6 episodes with Pip
4 episodes with Pip
5 episodes with Pip
5 episodes with Pip
7 episodes with Pip
11 episodes with Pip
5 episodes with Pip
5 episodes with Pip
14 episodes with Pip
8 episodes with Pip
4 episodes with Pip
9 episodes with Pip
6 episodes with Pip
8 episodes with Pip
108 episodes with Pip
7 episodes with Pip
3 episodes with Pip
4 episodes with Pip
20 episodes with Pip
6 episodes with Pip
3 episodes with Pip
7 episodes with Pip
12 episodes with Pip
11 episodes with Pip
6 episodes with Pip
19 episodes with Pip
5 episodes with Pip
4 episodes with Pip
5 episodes with Pip
4 episodes with Pip
18 episodes with Pip
4 episodes with Pip
![[CLIC] Podcast California Lodging Investment Conference](https://ivyfm.s3.amazonaws.com/i320/886410.jpg)
9 episodes with Pip
12 episodes with Pip
4 episodes with Pip
4 episodes with Pip
3 episodes with Pip
8 episodes with Pip
4 episodes with Pip
9 episodes with Pip
10 episodes with Pip
8 episodes with Pip
9 episodes with Pip
5 episodes with Pip
6 episodes with Pip
4 episodes with Pip
4 episodes with Pip
5 episodes with Pip
3 episodes with Pip
8 episodes with Pip
3 episodes with Pip
3 episodes with Pip
2 episodes with Pip
4 episodes with Pip
14 episodes with Pip
5 episodes with Pip
2 episodes with Pip
3 episodes with Pip
9 episodes with Pip
7 episodes with Pip
4 episodes with Pip
2 episodes with Pip
4 episodes with Pip
10 episodes with Pip
5 episodes with Pip
5 episodes with Pip
5 episodes with Pip
2 episodes with Pip
3 episodes with Pip
4 episodes with Pip
3 episodes with Pip
6 episodes with Pip
3 episodes with Pip
3 episodes with Pip
9 episodes with Pip
3 episodes with Pip
3 episodes with Pip
2 episodes with Pip
2 episodes with Pip
8 episodes with Pip
3 episodes with Pip
4 episodes with Pip
3 episodes with Pip
4 episodes with Pip
3 episodes with Pip
3 episodes with Pip
3 episodes with Pip
2 episodes with Pip
6 episodes with Pip
3 episodes with Pip
It's almost never the work. Great engineers get passed over for promotion because they're all results and no reputation. Steal 200 ways to get known for the work you're already doing → https://toolkit.oasisofcourage.com/reach Built on a whiteboard. Watch it → https://youtu.be/pqVQgAwoz1k Your work does not speak for itself. It never has, and it never will. Reputation = Results × Reach. Results are the impact you create. Reach is the number of people who can explain your value when you're not in the room. It multiplies, which is what makes it brutal. Nine parts results and one part reach gives you nine. Five and five gives you twenty-five. And world-class results times no reach still equals nothing. Curtis was a senior software engineer at NASA Ames, already performing at a high level and afraid of another PIP. He didn't get better at his job. He got better at giving his managers clear information about what he was contributing. Best review of his career, new machine learning work, an award for a proposal. The results were already there. In this episode: why my work speaks for itself is the most expensive belief in engineering, the equation and why it multiplies instead of adds, more vs better vs different, what reach actually means, which handful of people matter more than everyone else combined, and how to talk about your contribution without it feeling like self-promotion. You built the results. Now go build the reach. New here? Follow the show. Ready to build the whole plan → https://calendly.com/oaco-client-success-team/career-clarity-call Zach White is the founder of Oasis of Courage. Engineering degrees from Purdue and Michigan, former engineering leader at Whirlpool, 300+ engineering leaders coached since 2019. You shouldn't have to sacrifice your life to reach your potential at work. Also in Season 2: How Engineering Managers Get Promoted Without Bragging · How to Get Promoted from Engineering Manager to Director
emocleW, emocleW, emocleW to the Distraction Pieces Podcast with Scroobius Pip!This is your bonus FRIDAY REWIND episode! Today, we catch up with KSI, originally episode 395 from 2021-07-14.Original writeup below:Pip is joined by a Youtube veteran, and definitely someone with an admirable work ethic. Inspired by the likes of Childish Gambino and early adopters, KSI has carved out his own unique space in the world of social media and by the sounds of it, not even a global pandemic can stop this train running… Hear the story of his path from Watford, and how it led to where he currently finds himself including his boxing hustle, balance, fear of losing, working harder than anyone, how Youtube accepts no defeats and how to ascend by way of constant grind.PIP'S PATREON PAGE if you're of a supporting natureEVERYTHING KSIPIP TWITCH • (music stuff)PIP INSTAGRAMSPEECH DEVELOPMENT WEBSTOREPIP TWITTERPIP IMDB Hosted on Acast. See acast.com/privacy for more information.
Taylor Johnson presents a mixtape of his personal selection of tracks from BBC Introducing, with music from Venus Grrrls, Dylan Bradley and the Heat, Phoebe Green, The Marches, Bea Malka, Iona Luke, Helena Gao, Isabella Richardson, The Wild Infinite, Two Blinks, Jo From School, Heron Red, Pip, Hannah‑Morgan and a new Track of the Week from Birdseed.Produced by BBC Audio for BBC Radio 6 Music.
Welcome, welcome, welcome to the Distraction Pieces Podcast with Scroobius Pip!This week is the digital debut of Pip's live spoken word rendition of the DISTRACTION PIECES album in full!For total clarity, the order of events was the solo album 'Distraction Pieces', and then this podcast. So it makes total sense that in this whole podcast ouroboros and infinity loop we have in front of us, the podcast itself is the perfect platform for the very first digital unveiling of the spoken word version! While Pip was filming in Canada in maximum pandemic, time marched on until boom, it was time for the 10 year anniversary of his solo album. With the initial album on deck as a proper reissue, it turned out that with all that isolation and solo time that was in fact legally mandated, it was the ideal moment to add a little special something to the reissue situation. What better moment to leap back in time to the days of rocking up to a venue with a notebook and not much else, to perform spoken word and get your name out there where the game was a good deal of word of mouth and not all socials. Pip packed up his mic and jetted to Stanley Park, up on the tippy top of Vancouver, recorded the whole album acapella, and this formed disc 2 of the vinyl reissue. It was never digitally released, no Bandcamps, no streamers, nothing, and so if you didn't cop the wax you won't have heard this. So it really is a very special thing indeed, and represents a really important era in the Pip career path. Enjoy, and as this isn't a regular sounding episode (ie. not an interview), maybe set aside some time to take it in properly and without - indeed - distractions.Recorded by Pip • mix/edit/master by Buddy PeacePIP'S PATREON PAGE if you're of a supporting natureSPEECH DEVELOPMENT WEBSTOREPIP TWITCH • (music stuff)PIP INSTAGRAMPIP TWITTERPIP PATREONPIP IMDB Hosted on Acast. See acast.com/privacy for more information.
Double Tap Double Tap - Ep 476 August 24, 2026 Presented by This episode of Double Tap is brought to you by: Foxtrot Mike (Code: WLSISLIFE) Medical Gear Outfitters (Code: WLSISLIFE) Bowers Group (Code: WLS) Flatline Fiber Co (Code: WLS15) Second Call Defense Giveaways!! GAW Text Dear WLS or Reviews +1 743 500 2171 Public Show Titles Dear WLS Question from Jakey Poo from Indiana Dear WLS Attention fustercluck, Jakey Poo here, I have a classic rifle I'd like to hang on my wall somewhere, but I don't want it to sit up there and deteriorate. What should I do to it, to make it last? I do want it to remain functional, even after a cleaning, not necessarily straight off the wall, and it has wooden parts. Not sure if that makes any difference in regards to oils etc. Thanks lady-boys. Question from Jeremy from Nebraska Jeremy from Nebraska. I haven't been listening long enough, but I'm curious what Shawn's problem is with Taurus. I know they're budget guns but their recent stuff seems to be pretty reliable. Show me on the doll where they hurt you.? Question from typicalpnwguy from Oregon Hey cult daddy's, I've noticed that Primary Arms hasn't been mentioned and is no longer the title sponsor. Why that be yo?Ps… I love you, say it back -typicalpnwguy WINNER Question from K.Y. Horseman from Oregon Dear WLS K.Y. Horseman SBR question Going to buy a Henry Supreme, and deck it out with Midwest furniture. I want to SBR it, but I don't know anyone around me that I trust to do it. Was going to send it to Jermey and let him do it, but don't know how to do the NFA paperwork since I'm sending it out of state to SBR it. Do I form 1 it before I send it to Jeremy? Can he ship it back to me after the work is done? Help me Thanks Question from Anonymous Coward from Texas Lucky BrainlessI have a question on how you identify the caliber of a barrel. I recently found several ARs I had put up pre-covid. The issue is I don't remember which are 556 and which are 300 blackout. How can I verify so I don't load them with the wrong ammo? I would rather not have to take them apart, or blow them up. Gun Industry News Shootingnewsweekly Vortex Venom Enclosed Micro Green Dot (VEN-MGD3-E) Vortex has expanded its Venom enclosed red dot line by adding a green dot variant, the VEN-MGD3-E. The optic features a 3 MOA green dot on a DeltaPoint Pro footprint, enclosed 6061 aluminum housing, and is designed for improved visibility in various lighting conditions or for users with astigmatism. It offers a large viewing window, aspherical lens, motion activation, and long battery life. The Gist: Available starting ~September 2024 Impact: MSRP $289.99 (street price ~$200) Bottom Line: 1x magnification, 3 MOA green dot, 20,000-hour CR2032 battery life, 1.84″ length, 1.75 oz, 1 MOA/click adjustment, motion-activated illumination with 10-min auto-off, top-mount brightness buttons, unlimited eye relief, enclosed design on DeltaPoint Pro footprint Bearingarms Pentagon Investigation Finds No Mechanical Issues with Sig Sauer M18 Pistol Following the death of an airman in an incident initially reported as an uncommanded discharge of an M18 pistol (the military version of the Sig Sauer P320), the Pentagon reviewed the Modular Handgun System program. Officials determined that all alleged uncommanded discharges were traced to the trigger being pulled, with no mechanically caused accidental discharges ever recorded. Firearms mishaps represent less than 0.006% of all MHS pistols issued; the Pentagon attributes safety concerns to public speculation and online misinformation. The Gist: The Gist: Pentagon investigation concluded no mechanical failures in M18; all incidents resulted from trigger pull, including the fatal airman case initially misrepresented (two airmen pleaded guilty to false statements). Impact: Market Impact: Affects U.S. military adoption and perception of Sig Sauer M18/P320 platform; continued confidence in the Modular Handgun System program with no indicated changes to procurement. Bottom Line: The Bottom Line: Extremely low mishap rate (
emocleW, emocleW, emocleW to the Distraction Pieces Podcast with Scroobius Pip!This is your bonus FRIDAY REWIND episode! Today, we catch up with Niamh Algar, originally episode 315 from 2020-03-04.Original writeup below:You can tell even like 5 minutes in that Pip was proper excited about this one… And naturally he was right to be, as Niamh is fantastic if you didn't already know! You may have seen her in the beautiful and powerful ‘The Virtues' alongside Distraction Pieces spiritual godfather Stephen Graham, and if you did you will adore this chat as Niamh goes IN on the whole process. As always, any Shane Meadows talk is fascinating and this time is no exception, as we get the pure goodness and intrigue from the process of making such a show and the deep personal digging involved in Niamh's transformation into Dinah. Not only that, there's further confirmation of that ‘being the lead in your own story' point that comes up here and there on the podcast, and some good chat about her role in Desiree Akhavan's ‘The Bisexual', working with Guy Ritchie and OF COURSE a good amount about her upcoming film ‘Calm With Horses', alongside Barry Keoghan and Cosmo Jarvis. Full on gold. Enjoy.PIP'S PATREON PAGE if you're of a supporting natureCALM WITH HORSESTHE VIRTUESTHE BISEXUALIMDBPIP TWITCH • (music stuff)PIP INSTAGRAMSPEECH DEVELOPMENT WEBSTOREPIP TWITTERPIP IMDBPOD BIBLE Hosted on Acast. See acast.com/privacy for more information.
Right, buckle up, trawlers... Trump's crypto company has been rubber-stamped as an actual bank (by his own government no less!) and Marina and Jemma wave a cheery bye Felicia to his press secretary Caroline Levitt (left to perish on the decoy plane while Trump and a "very very very close aide" swanned off on the catering flight). They boggle at the Tates' Temu lawyer, a shouting salami in sunglasses, vowing to take their extradition fight "to Trump" - plus the 30-odd bellends doing solidarity press-ups outside a Miami jail. Then it's back home to Blighty, where Reform reckon they'll save £50bn by whipping PIP off up to three million people and gutting the health bit of Universal Credit, which the BBC broadcast in full, of course. The ladies get stuck into why it's cruel, unworkable and really a mental health story, discuss their own personal experiences of welfare and then it's on to Netanyahu branding us the "Islamic Republic of Britain," and an Underrated of the Week for the actual Pope. Amen!Join The Trawl's Facebook BRAND NEW Facebook Group: https://www.facebook.com/profile.php?id=61590931382660Get your tickets for The Trawl Live from https://thetrawl.tix.to/ticketsThank you for sharing and please do follow us @MarinaPurkiss @jemmaforte @TheTrawlPodcast Patreonhttps://patreon.com/TheTrawlPodcast Youtubehttps://www.youtube.com/@TheTrawl Twitterhttps://twitter.com/TheTrawlPodcastIf you've even mildly enjoyed The Trawl, you'll love the unfiltered, no-holds-barred extras from Jemma & Marina over on Patreon, including:• Exclusive episodes of The Trawl Goss – where Jemma and Marina spill backstage gossip, dive into their personal lives, and often forget the mic is on• Early access to The Trawl Meets…• Glorious ad-free episodesPlus, there's a bell-free community of over 3,300 legends sparking brilliant chat.And it's your way to support the pod which the ladies pour their hearts, souls (and occasional anxiety) into. All for your listening pleasure and reassurance that through this geopolitical s**tstorm… you're not alone.Come join the fun:Patreon: https://www.patreon.com/TheTrawlPodcast?utm_campaign=creatorshare_creator Hosted on Acast. See acast.com/privacy for more information.
Welcome, welcome, welcome to the Distraction Pieces Podcast with Scroobius Pip!This week Pip is joined by Hardcore Listing homie and long term real life homie CHRIS GLASSON, and podcast producer / DJ / beatmaker BUDDY PEACE for the finalé of a 3 part series of celebration episodes!The third side of the pyramid is finally here, as we close in on the three part celebration featuring, indeed, three parts of the puzzle. A lovely way to bring this thing home and tie up loose ends, hang up phones, open new cans of worms (and then re-seal them?), and effectively conclude this three part extended hangout. Thanks for checking in and celebrating 20 years of Pip!PIP'S PATREON PAGE if you're of a supporting natureHARCORE LISTINGBUDDYSPEECH DEVELOPMENT WEBSTOREPIP TWITCH • (music stuff)PIP INSTAGRAMPIP TWITTERPIP PATREONPIP IMDB Hosted on Acast. See acast.com/privacy for more information.
Pip Northeast is the breakout star of Netflix’s My Brilliant Career, currently one of the Top 10 shows in the world. But forget everything you’ve been told about getting ahead at work. After chatting about taking on one of Australia’s most iconic stories - and the woman behind it all, Miles Franklin, we put the career advice we’ve all been fed to the test. Should you really “lean in”? Do you actually need a mentor? Can you be in love and have a career? Pip gives us her verdict on the best, worst and most outdated career advice, with insights into her own experiences. My Brilliant Career is out now on Netflix. Big Small Talk Instagram Sarah-Jane Adams InstagramSee omnystudio.com/listener for privacy information.
Philippa and Katie are back together for the week in Ambridge — Sunday 16th to Tuesday 18th August — and there is a lot to process.Monty's story moves to a heartbreaking conclusion as Lynda and Robert say goodbye and sign the documents. Katie cannot cope and Philippa has questions about the rehoming process, the DNA fingerprints, and whether there is more to come from that evidence. Meanwhile Harrison and Fallon's relationship continues to unsettle, a possibly flirty exchange between Harrison and Lily gets noted, and Mick is becoming CSI Ambridge.Khalil is struggling and Keira has been quietly keeping his secret — until Susan unknowingly spills the beans. Now Azra knows, and neither host is convinced she is going to handle it well. There are real concerns about Khalil bottling things up, particularly given his history.Eddie is filling in for George at Meadow Farm and is cutting every corner available to him. George, remarkably, was right. Ruth still has not booked the genetic test and both hosts are furious about it. The Cruck Barn is taking shape with an exciting programme of rural crafts, jewellery making and creative writing — but can Ambridge actually afford it? And nobody has acknowledged the heatwave or drought even once, which is bothering everyone.Plus Lynda's DNA fingerprints, Rosie and Pip moving into the bungalow, who is now paying rent on Rickyard Cottage, and a key that would not work properly.Star of the Week, Twitter of the Week and predictions including Lynda being unmasked as a mafia gang leader.Topics covered: Monty | Lynda Snell | Robert Snell | Khalil | Keira | Azra | Eddie Grundy | George Grundy | Ruth Archer | Harrison Burns | Fallon Rogers | Lily Pargetter | Cruck Barn | Elizabeth Pargetter | Susan Carter | The Archers August 2026 Hosted on Acast. See acast.com/privacy for more information.
In this solo episode, Bradley Hamner opens a series on developing and coaching team members with a case for performance improvement plans. Most small business owners treat a PIP as a cover-your-back document and a signal that someone is already on the way out. Bradley argues that stigma costs owners good people and real money.He walks through four reasons owners skip PIPs, including corporate red tape stigma, close personal relationships with staff, and the absence of any HR infrastructure. He then lays out four reasons to use them: forcing concrete clarity on expectations, protecting the business legally and financially, relieving the invisible strain that unaddressed underperformance puts on A players, and keeping people worth keeping. He shares a client who built automatic PIP triggers into the culture and retains more than 80 percent of the people who go on one.This conversation moves beyond documentation and into how owners actually develop people. What does good performance look like in concrete numbers? Who on your team is quietly absorbing someone else's workload? And what does it cost to replace a person you could have coached? If you have avoided a performance conversation because the paperwork felt like corporate overhead, this episode is for you.Visit https://workshop.blueprintos.com to register for the upcoming Above The Business workshop.Thanks to our sponsorsCoach P ConsultingCoach P found great success as an insurance agent and agency owner, leading a large and stable team of top-performing professionals. Today, he shares the systems, delegation strategies, and specialization methods he developed along the way. Gain access to weekly training calls and mentoring at:https://coachpconsulting.comBe sure to mention you heard about it on the Above The Business Podcast.Autopilot RecruitingAutopilot Recruiting helps small business owners solve staffing challenges by taking the stress out of hiring. Their dedicated recruiters work on your behalf every business day. They optimize your applicant tracking system, post job listings, and source candidates through social media and local communities.With their continuous recruiting approach, you can save time, reduce hiring costs, and receive pre-screened candidates without paying hiring fees or commissions.https://www.autopilotrecruiting.comMention Above The Business Podcast when you reach out.Direct ClicksDirect Clicks specializes in digital marketing solutions designed for business owners who want measurable results. Their team supports companies through paid advertising, SEO, and strategic marketing systems that help generate consistent leads.Exclusive offer for listeners:https://directclicksinc.com/abovethebusinessGet a free marketing campaign audit where their team reviews your website, SEO, content, social media, and paid advertising, then provides actionable recommendations. If you partner with them, all setup fees will be waived.About Above The BusinessAbove The Business is hosted by Bradley Hamner, founder of BlueprintOS, and focuses on helping small business owners transition from Rainmaker to Architect by building systems, teams, and operations that scale without their constant involvement.
Akram's excited for what he might enter in the Flower & Produce Show, and invites Khalil to come to the allotment to help him choose. Khalil declines; he and Keira will be gaming instead. Azra and Akram agree Khalil isn't himself, probably still affected by the situation with Lynda's dog, Monty. They speculate that he might be feeling guilty about his rant at Brookfield last week. Azra resolves to go over there and apologise on Khalil's behalf. Later she and Ruth chat about Pip and Stella's wedding before Azra gently broaches the subject of the attack offers her apology for Khalil's actions. Ruth tells her not to worry, brushing the incident with Khalil aside. She admits she's still feeling shocked over the attack on the sheep. Azra sympathises, and they agree it's a horrible situation for everyone. Meanwhile Khalil complains to Keira that his family is breathing down his neck, whilst admitting he's having flashbacks to the images of the sheep. He loved Monty like a friend and can't get his head round what he did. Keira suggests Azra might be able to help, but Khalil's adamant his family mustn't know he's seen the pictures. They already worry too much about him. Later Azra and Akram agree it's a relief how understanding Ruth and David were about Khalil. There's an awkward moment for Keira when they quiz her about what might be bothering Khalil, but she successfully diverts them. They hope they're worrying about nothing.
emocleW, emocleW, emocleW to the Distraction Pieces Podcast with Scroobius Pip!This is your bonus FRIDAY REWIND episode! Today, we catch up with Steven Knight, originally episode 256 from 2019-03-06.Original writeup below:Chances are that if you are a Distraction Pieces Podcast listener (god bless you), you will be fully versed on Steven's work. Pip references his work with a great and contagious fondness and for very good reason - he makes some incredibly fine film and television! This is a perfect opportunity to hear from the man in that director seat behind the curtain on a huge amount of topics and angles including the rule-based world of film making, his work on Locke, Peaky Blinders, Taboo and all the rest in his insane body of work, his new film Serenity and how he lucked out on getting his first choice cast picks, playing with reality, how we are all the lead roles in our own realities, the BBC leaving him alone in creating, the far and wide fans of Peaky, how one gets a cast together and that old unfinished story of Florence Pugh trying to smuggle she and Pip into Peaky too! Let's see if the inception works, shall we? Great one here folks, enjoy!PIP'S PATREON PAGE if you're of a supporting natureSERENITYIMDBSTEVEN on ROTTEN TOMATOESSTEVEN on DEN OF GEEKPIP TWITCH • (music stuff)PIP INSTAGRAMSPEECH DEVELOPMENT WEBSTOREPIP TWITTERPIP IMDBPOD BIBLE Hosted on Acast. See acast.com/privacy for more information.
Anthropics Investoren erwarten für Oktober einen Börsengang bei zwei Billionen Dollar, was der größte IPO aller Zeiten wäre. Pip erklärt, warum der Termin geschickt gewählt ist. Für OpenAI sieht er die Lage umgekehrt. Dort kommt der zweite Vertriebschef in einem Jahr. Zusammen mit Cerebras hat OpenAI dafür eine Variante gebaut, die vierzehnmal schneller antwortet. Danach vier Modellstarts in einer Woche, bei denen ausgerechnet DeepSeek die Preise um bis zu das Zwölffache erhöht, und Elon Musk sein neues Grok für objektiv das beste Modell hält. Bei den Finanzierungsrunden geht es um Databricks, Lovable, Legora und Cognition, dazu um die Frage, ob man das Geld gerade nehmen und liegen lassen sollte. Silver Lake holt Workday von der Börse. In der Schmuddelecke erlaubt die Trump-Regierung privaten Firmen offensive Cyberangriffe und beruft sich dabei auf Kaperbriefe aus der Verfassung. Unterstütze unseren Podcast und entdecke die Angebote unserer Werbepartner auf doppelgaenger.io/werbung. Vielen Dank! Philipp Glöckler und Philipp Klöckner sprechen heute über: (00:00:00) Aus der Community (00:02:05) OpenAI wechselt den Vertriebschef (00:12:20) Ultrafast mit Cerebras (00:17:16) Anthropic-IPO (00:29:10) Anthropic kauft Decart (00:30:34) Braucht man ein KI-Device? (00:32:36) Gemini 3.7 Flash (00:34:10) DeepSeek V4-Pro (00:34:47) Grok 4.6 (00:37:11) SpaceX (00:38:49) Databricks und Snowflake (00:42:40) Workday geht von der Börse (00:44:11) Lovable (00:49:26) Legora (00:52:28) Cognition (00:55:18) Mistral (00:57:22) Kaperbriefe (01:02:05) Truth API (01:03:28) Chronext (01:06:55) Apple zahlt Verlage Shownotes OpenAI holt den zweiten Vertriebschef in einem Jahr - bloomberg.com GPT-5.6 Sol läuft mit Cerebras bis zu 14-mal schneller - 9to5mac.com Anthropic peilt einen Börsengang bei 2 Billionen Dollar an - ft.com Anthropic verhandelt über Decart für 6 Mrd. - bloomberg.com Google stellt Gemini 3.7 Flash vor - blog.google DeepSeek bringt V4-Pro und erhöht die Preise um bis zu das Zwölffache - theinformation.com Grok 4.6 startet zuerst in Cursor - gizmodo.com SpaceX-Leerverkäufern gehen die Kugeln aus - cnbc.com Databricks sammelt 5 Mrd. bei 190 Mrd. Bewertung ein - cnbc.com Silver Lake verhandelt über eine Übernahme von Workday - reuters.com Lovable verdoppelt die Bewertung auf 13,3 Mrd. - trendingtopics.eu Legora verhandelt bei mindestens 10 Mrd. - ft.com Cognition verhandelt bei 40 Mrd. - bloomberg.com Mistral will bis 2030 ein Gigawatt in Europa bauen - aibusiness.com Trump lässt private Firmen offensive Cyberangriffe fahren - bloomberg.com Kaperbriefe stehen in der Verfassung - xcancel.com KI-Agenten greifen Taiwans Regierungssysteme an - ft.com Presseverbände klagen gegen Trumps Truth API - ft.com Chronext-Kunden warten auf Zahlungen und Lieferungen - wiwo.de Apple verhandelt mit Verlagen über Nachrichten für Siri - techcrunch.com
Is your performance improvement plan designed to improve performance—or is it simply documenting an exit you already decided to make? Too many leaders use a PIP to compensate for months of vague expectations, delayed feedback, inconsistent accountability, and conversations they refused to have. The paperwork may be complete, but the leadership is missing. In Episode 143 of the Leadership Sandbox, Tammy J. Bond exposes the five signs a PIP has become corporate theater and explains what actually changes workplace behavior: clear standards, honest dialogue, consistent communication, documented expectations, and real follow-through. She also challenges leaders to recognize the ripple effect of every PIP: the entire team is watching to see whether leadership uses accountability to help people win—or merely checks HR's boxes on the way to termination. Topics explored in this episode: [00:31] When the PIP becomes theater [02:00] Avoiding discomfort is not caring for the employee [03:40] The entire team is watching what the leader does next [05:45] How vague expectations create performance problems [08:55] Stop using paperwork as a predetermined exit strategy [9:20] Five signs your PIP is a theatrical production [14:10] People don't want to fail at work [16:15] Dialogue—not documentation—is the behavior-change strategy [17:21] Lead the human being, not only the human doing Explore COMMAND™ at https://www.bondgroupenterprises.com/command-leadership.
Welcome, welcome, welcome to the Distraction Pieces Podcast with Scroobius Pip!This week Pip is joined by Hardcore Listing homie and long term real life homie CHRIS GLASSON, and podcast producer / DJ / beatmaker BUDDY PEACE for the second in a 3 part series of celebration episodes!The continuation in the three part saga, on some Lord Of The Rings / Back To The Future / Godfather type ishh but right there in the middle. A lot more ground to cover here including the later stages of the music making, progressing into podcasting via radio, the London clubnight, and how the podcast itself spawned many other podcasts to follow... So much to get through - ENJOY!PIP'S PATREON PAGE if you're of a supporting natureHARCORE LISTINGBUDDYSPEECH DEVELOPMENT WEBSTOREPIP TWITCH • (music stuff)PIP INSTAGRAMPIP TWITTERPIP PATREONPIP IMDB Hosted on Acast. See acast.com/privacy for more information.
Troy Jackson is joined by Rev. Traci Blackmon to discuss election protection, what it is and why it matters. They reflect on the importance of voting access and what a healthy, faithful culture of civic participation would look like in our communities.Each August, we focus on civic engagement as a CCDA Family. Learn more about engaging with your local government at ccda.org/civic.Learn more about The Faithful Witness Campaign at faithfulwitness.us.The Rev. Traci Blackmon is the former associate general minister of justice and local church ministries for the United Church of Christ, a position she held for eight years. She earned a bachelor of science in nursing from Birmingham-Southern College and a master of divinity degree from Eden Theological Seminary.Blackmon's work focuses on communal resistance to systemic injustice through the redemptive power of love. Her response in Ferguson, Missouri, to the killing of Michael Brown resulted in national and international recognition.She was appointed to the Ferguson Commission by Missouri Gov. Jay Nixon and to the President's Advisory Council on Faith-Based and Neighborhood Partnerships by President Barack Obama. She currently serves on the board of the Samuel DeWitt Proctor Conference and was named among 15 “Faith Leaders to Watch” in 2020 by the Center for American Progress.Learn more about Rev. Blackmon and her work at faithoutloud.org.Community organizer, pastor, leader and writer, Troy Jackson, has been with UNDIVIDED since its founding. He is a graduate of Princeton Theological Seminary and earned his PH.D in U.S. history from the University of Kentucky.Troy's book Becoming King: Martin Luther King, Jr. and the Making of a National Leader (The University Press of Kentucky, 2008) explores the critical role the grassroots Montgomery Movement played in the development of Dr. King. His other publications include his work as an editor on The Papers of Martin Luther King, Jr. Volume VI: Advocate of the Social Gospel (September 1948-March 1963) (University of California Press, 2007). With their children grown, Troy and his wife moved to a cabin outside of Cincinnati, OH, with their two dogs, Pip and Pico.Learn more about Troy and UNDIVIDED at undivided.us.Learn more about CCDA and how you can get involved at ccda.org. Connect with CCDA on Instagram, Twitter, Facebook, and LinkedIn. Follow CCDA on YouTube.
ark Zuckerberg bezahlt einen Profikämpfer für ein Instagram Video und veröffentlicht am selben Tag ein Manifest darüber, warum die Zukunft allen gehört. Pip nennt zwei Lackmustests, an denen sich zeigen wird, ob das ernst gemeint ist. Anthropic macht den Einführungspreis von Sonnet 5 dauerhaft. OpenAI kauft eigenen Mitarbeitern Anteile für sieben Milliarden Dollar ab, und zwar mit eigenem Geld statt über externe Investoren. Bei SpaceX fehlen für die angekündigten zehn Gigawatt rund vierhundert Milliarden. In China bricht Kimi aus der Testumgebung aus, ByteDance will das erste Modell mit zehn Billionen Parametern bauen, und 97 Prozent aller Humanoiden kommen inzwischen von dort. Dann fünf Jahre Frank-Thelen-Fonds. Unterstütze unseren Podcast und entdecke die Angebote unserer Werbepartner auf doppelgaenger.io/werbung. Vielen Dank! Philipp Glöckler und Philipp Klöckner sprechen heute über: (00:00:00) Zuckerberg im Cage (00:04:24) The Future is for Everyone (00:08:46) Metas Geschäftsmodell (00:13:59) Sonnet-5-Preis (00:15:25) OpenAI kauft zurück (00:22:28) Bubble Discussion (00:27:01) SpaceX-Finanzierung (00:31:55) Kimi bricht aus (00:33:12) ByteDances Riesenmodell (00:35:54) Chinas Kapitalmarkt (00:37:33) Humanoide aus China (00:39:16) Korea und Taiwan (00:40:42) Shein-IPO (00:41:18) Fünf Jahre 10xDNA (00:45:15) Spotify skippt Werbung (00:56:45) Zuckerbergs Yacht (00:59:26) Vantage Tripod (01:02:23) Amazons Gaskraftwerk (01:04:07) Wildberries (01:05:59) OpenClaw bucht Yoga Shownotes Zuckerbergs Essay: The Future is for Everyone - about.fb.com Zuckerberg legt seine KI-Vision auf 6.500 Wörtern dar - wsj.com Meta öffnet Muse Glimmer und legt 1 Mrd. für Gemeinden auf - ft.com Anthropic macht den Sonnet-5-Einführungspreis dauerhaft - xcancel.com OpenAI kauft Mitarbeiteranteile für 7 Mrd. zurück - bloomberg.com SpaceX nach den Zahlen: Capex, Lock-up und Leerverkäufer - ft.com Kimi K3 bricht aus der Testumgebung aus - techcrunch.com ByteDance trainiert ein Modell mit bis zu 10 Billionen Parametern - ft.com China öffnet 28 Billionen Kapitalmarkt für den Chip-Wettlauf - bloomberg.com China liefert 97 Prozent aller Humanoiden - bloomberg.com Südkorea und Taiwan überholen Japan bei den Exporten - asia.nikkei.com Shein peilt 30 bis 40 Mrd. Dollar für den Hongkong-IPO an - qz.com Fünf Jahre 10xDNA, gerechnet im Subreddit Finanzen - reddit.com Fondsvergleich 10xDNA gegen den Index - onvista.de Spotifys neuer Skip-Button und das Podcast-Geschäft - semafor.com Muddy Waters über Zuckerbergs Yacht - xcancel.com Googles KI kannte einen Namen aus einem privaten Dokument - techspot.com Die Namen für den Trabant-SUV, die Pip live generieren ließ - chatgpt.com Amazons Rechenzentrum in Texas mit eigenem Gaskraftwerk - nytimes.com Wildberries unter ukrainischem Drohnenbeschuss - wsj.com KI-Assistent hackt die Buchungsseite eines Fitnessstudios - abc.net.au
emocleW, emocleW, emocleW to the Distraction Pieces Podcast with Scroobius Pip!This is your bonus FRIDAY REWIND episode! Today, we catch up with Souad Mekhennet, originally episode 169 from 2017-09-20.Original writeup below:You may have heard Pip hyping this one up for a minute, and quite rightly so - ladies and gentlemen, please welcome journalist, author, writer, Souad Mekhennet! A huge episode here, as Pip sits down with Souad for a nice good while to really get into her own story... And what a deep, heavy and truly daring life she has led, from beginning to present day - hear her recall some chills-inducing tales and truly unique meetings, from getting in deep into situations with Isis commanders, her interview with 'Jihadi John', and her career covering all manner of terrorist activities over the years, and SO much more, made all the more fascinating due to her being Muslim and female, which positions her uniquely for her line of writing and interviewing. It's also fascinating hearing her thoughts on what creates a perfect situation for men to become radicalised, and how it can mainly stem from a broken family life. This is an absolutely gripping conversation, to which this writeup would never do justice, and it's a wonderful commentary companion to her book 'I Was Told To Come Alone', which you should (and surely will) snap up at your nearest convenience.PIP'S PATREON PAGE if you're of a supporting natureINSTAGRAMWIKII WAS TOLD TO COME ALONETHE ETERNAL NAZIPIP TWITCH • (music stuff)PIP INSTAGRAMSPEECH DEVELOPMENT WEBSTOREPIP TWITTERPIP IMDBPOD BIBLE Hosted on Acast. See acast.com/privacy for more information.
After nearly two years off the live mic, GMoney reunites with Patriots in Progress (Pip) for a wide-ranging, unhurried conversation that treats the current Bitcoin chaos as almost the calmest thing in the room. The two old comrades compare notes on surviving fifth-generation warfare, toxic maximalism, and the exhausting theater of geopolitics, arguing that stepping back is the only way to stay sane while the noose slowly tightens around the enemy. Expect a heavy dose of Pip's deep dive into Ecclesiastes, spiritual warfare, and why he thinks Bible prophecy is looking more plausible by the year, all woven together with the usual frog-adjacent banter. Along the way you get thoughts on Israel, tokenization, X Chain, and Satoshi's cryptographic roots in wartime code-breaking. Part reunion, part sermon, part signal-versus-noise pep talk, this one is a slower burn for listeners who care more about first principles than the daily shiny object.
Welcome, welcome, welcome to the Distraction Pieces Podcast with Scroobius Pip!This week Pip is joined by Hardcore Listing homie and long term real life homie CHRIS GLASSON, and podcast producer / DJ / beatmaker BUDDY PEACE for the first in a 3 part series of celebration episodes!To be exact, a celebration of 20 years of Scroobius Pip, dating back to the first musical output and initial starting point in the entire journey. This goes back to those very first days, the days of Myspace and the four track limit (makes you feel nostalgic doesn't it...), the street art world, making music on a four track recorder and the lo-fi home studio setups, Björk and Sage Francis as early influences - and speaking of influences - the spoken word scene and pals made around that, early touring including the performances outside gig queues, booze as a prop, and all sorts of fun lil' Easter eggs and such. The cliff-hanger is a goody too... Enjoy!PIP'S PATREON PAGE if you're of a supporting natureHARCORE LISTINGBUDDYSPEECH DEVELOPMENT WEBSTOREPIP TWITCH • (music stuff)PIP INSTAGRAMPIP TWITTERPIP PATREONPIP IMDB Hosted on Acast. See acast.com/privacy for more information.
Amazon zahlt OpenAI die vollen 50 Milliarden aus, obwohl keine der vereinbarten Bedingungen eingetreten ist. Google hat für Anthropic eine 200-Milliarden-Konstruktion aus Broadcom, Apollo, Blackstone, Morgan Stanley und ehemaligen Krypto-Minern gebaut. Pip erklärt dabei von Grund auf, wie Off-Balance-Sheet-Finanzierung funktioniert und warum niemand diese Chips in der eigenen Bilanz haben will. Daraus entwickelt er eine These: AI 2.0 kommt erst noch. Aus China kommen die passenden Belege, mit Alibabas neuem Modell und einem DeepSeek, das je Aufgabe ein Hundertstel der westlichen Konkurrenz kostet. Dazu Palantirs Zahlen und ein Rechenexempel über die Margen zwischen Chip und Kunde. Zum Schluss schauen wir in die SpaceX Earnings. Unterstütze unseren Podcast und entdecke die Angebote unserer Werbepartner auf doppelgaenger.io/werbung. Vielen Dank! Philipp Glöckler und Philipp Klöckner sprechen heute über: (00:00:00) SpaceX vor den Zahlen (00:13:06) Amazon investiert in OpenAI (00:17:25) Eine Milliarde Nutzer (00:19:51) Off-Balance-Sheet erklärt (00:23:23) Googles 200-Milliarden-Maschine (00:41:32) AI 2.0 (00:43:57) Alibaba Qwen3.8-Max (00:44:53) DeepSeek V4-Flash (00:46:41) Chinas Militär destilliert (00:48:49) Amazon bei drei Billionen (00:49:30) Palantir (01:00:33) Snap (01:01:07) xAI gegen Minnesota (01:03:05) Telegram im App Store (01:04:07) Google zieht Satellitenbilder zurück (01:05:25) Liechtenstein gehackt (01:08:47) Airtable an Bending Spoons (01:09:26) SpaceX-Earnings Shownotes Amazon schließt 50-Mrd.-Investment in OpenAI ab - ft.com OpenAI überschreitet eine Milliarde Nutzer - wsj.com Googles 200-Mrd.-Finanzierungsmaschine für Anthropic - ft.com Off-Balance-Sheet-Finanzierung erklärt - corporatefinanceinstitute.com Alibaba veröffentlicht Qwen3.8-Max - bloomberg.com DeepSeeks V4-Flash ist das günstigste bekannte Modell - reuters.com Chinesische Militärforscher trainieren mit US-Modellen - reuters.com Amazon überschreitet drei Billionen Dollar Marktwert - bloomberg.com Palantir-Quartalszahlen, Umsatz plus 93 Prozent - cnbc.com Palantir-Aktie steigt nach turbogeladenem Wachstum - marketwatch.com Snap stellt AR-Brille für 2.195 Dollar in Aussicht - bloomberg.com Snap-Aktie springt über 10 Prozent - cnbc.com Richter lässt Minnesotas Nudify-Verbot in Kraft - nbcnews.com Telegram nach kurzem Rauswurf zurück im App Store - reuters.com Google zieht KI-Satellitenbilder zurück - npr.org Hacker stehlen 31.000 Datensätze aus Liechtensteins Geldwäscheregister - zeit.de
1. Ante es escándalo delexjuez y renunciante secretario de Corrección y Rehabilitación, FranciscoQuinones, la gobernadora designó nueva secretaria. Quinones se fue en medio depesquisas por tener sobre seis querellas de hostigamiento e imputaciones detener relación directa con líder sindical que representa a los empleados deCorrección.2. Gobernadora anticipaplan de interrupciones del servicio de agua potable3. Comunidades afectadaspor falta de agua se manifestarán en San Juan4. Al descubierto: la víarápida que sacó a Esencia del escrutinio público 5. Senadores del PIP pidena Gobernadora que explique de dónde Esencia sacará agua6. Construcción de unarotonda en El Morro y otras construcciones en San Juan generan preocupaciónentre ciudadanos que temen no se proteja el patrimonio histórico de la ciudad7. El expresidente peruanoAlejandro Toledo pide un indulto de Keiko Fujimori: “No quiero morir en lacárcel”. Veintiséis años después de liderar la resistencia contra AlbertoFujimori, el exmandatario, preso por corrupción, depende de la hija delautócrata para recuperar la libertad8. Muertos por sismos enVenezuela se elevaron a 6,125, tras 10 días sin datos oficiales9. El líder de Ceuta acusaa Marruecos de la «atrocidad» del cruce masivo de la frontera.Este es un programa independiente y sindicalizado. Esto significa que este programa se produce de manera independiente, pero se transmite de manera sindicalizada, o sea, por las emisoras y cadenas de radio que son más fuertes en sus respectivas regiones. También se transmite por sus plataformas digitales, aplicaciones para dispositivos móviles y redes sociales. Estas emisoras de radio son:1. Cadena WIAC - WYAC 930 AM Cabo Rojo- Mayagüez2. Cadena WIAC – WISA 1390 AM Isabela3. Cadena WIAC – WIAC 740 AM Área norte y zona metropolitana4. X61 – 610 AM en Patillas5. X61 – 94.3 FM Patillas y todo el sureste6. WPAB 550 AM - Ponce7. ECO 93.1 FM – En todo Puerto Rico8. WLRP 1460 AM Radio Raíces La voz del Pepino en San Sebastián9. WOQI 1020 AM – Radio Casa Pueblo desde Adjuntas 10. Mundo Latino PR.com, la emisora web de música tropical y comentario Una vez sale del aire, el programa queda grabado y está disponible en las plataformas de podcasts tales como Spotify, Soundcloud, Apple Podcasts, Google Podcasts y otras plataformas https://anchor.fm/sandrarodriguezcotto También nos pueden seguir en:REDES SOCIALES: Facebook, X (Twitter), Instagram, Threads, LinkedIn, Tumblr, TikTok BLOG: En Blanco y Negro con Sandra http://enblancoynegromedia.blogspot.com SUSCRIPCIÓN: Substack, plataforma de suscripción de prensa independientehttps://substack.com/@sandrarodriguezcotto OTROS MEDIOS DIGITALES: ¡Ey! Boricua, Revista Seguros. Revista Crónicas y otrosEstas son algunas de las noticias que tenemos hoy En Blanco y Negro con Sandra.
Watch the full episode on YouTube:We first covered Baseten last year when DeepSeek mania was at peak hype. Now they have raised a monster $13B round and become one of the new cohort of AI Infra decacorns that are (with Nvidia, Intel, and the semis complex) chief beneficiaries of the Inference Inflection. We return to Baseten at the peak of the 2026 edition of Open Weights debate. Ali has published a viral breakdown of Kimi K3:And since you last saw him, Philip has spoken at AI Engineer and written the definitive book on Inference Engineering spotted all over SF:Three years ago, inference engineering barely existed as a category.Today, it is one of the most critical disciplines in AI. Inference engineering inherently tackles a different question than standard model training: “How do you turn those weights from training into a product that is fast, reliable, and affordable at scale?” Focusing on these creates an entirely new optimization problem.In one recent GLM-5.2 experiment, quantizing more of the model actually preserved its benchmark quality while increasing throughput by 20%, because the errors introduced in different layers could cancel each other out.Inference is no longer just the final step after training. It is becoming its own engineering discipline, with its own research problems, infrastructure, and increasingly specialized roles.In this episode, Baseten's Philip Kiely and Ali Taha join swyx and Vibhu to explain what actually happens after a new open model is released and what it takes to turn “we generated a token” into a fast, reliable, production-ready API.We go deep on cache-aware routing, disaggregated prefill and decode, quantization, speculative decoding, KV-cache movement, model parallelism, GPU kernels, and the race to make frontier models up to 10× faster. Philip and Ali explain why inference optimizations can still produce gains of 20%, 100%, or even 200%; how quantization errors can cancel one another out; why identical weights can behave differently across clusters; and how Baseten grafted a Kimi vision encoder onto GLM-5.2 without changing the underlying language model.The conversation then expands beyond LLMs into NVIDIA Dynamo, mega kernels, Rubin, AI-specific chips, local inference, video generation, diffusion versus autoregressive models, and the enormous compute barrier to generating coherent long-form video. Finally, we explore the convergence of training and inference, continual learning through persistent KV cache, and the emerging loop where models help optimize the infrastructure that runs them.We discuss:* What happens when a 200,000-token request enters an inference system* Cache-aware routing and reusing previously computed KV cache* Why prefill and decode are increasingly handled by different GPUs* When dedicated deployments become cheaper and more reliable than shared APIs* How speculative decoding uses a smaller model to accelerate a larger one* Tool calling, structured outputs, and what LLMs actually do* What it takes to support a new open model on day zero* Grafting Kimi's vision encoder onto GLM-5.2* Retrofitting inefficient model layers with components from other architectures* Why models sometimes collapse into repeating the same token* How hardware, kernels, and race conditions create nondeterministic failures* Preserving model fidelity while making inference faster* How quantization errors can cancel each other out* Why inference optimizations still deliver gains of 20%, 100%, and 200%* How optimized serving can make a model up to 10× faster* NVIDIA Dynamo, KV-aware routing, and distributed model serving* Speculative decoding the speculative decoder* Why local AI is about making models less dumb while data-center AI is about making them less slow* Tensor, expert, and pipeline parallelism across GPUs* Hardware-aware model design, auto-tuning, and the case against mega kernels* Rubin and why inference is becoming a systems problem* Whether modern GPUs are evolving into programmable AI ASICs* Why enormous models like Kimi K3 require GB300-class hardware* Why open-source video generation still trails Veo, Kling, and other closed models* The quadratic attention bottleneck behind long-form AI video* Autoregressive video, real-time generation, and compounding quality drift* Why future video systems may combine autoregressive and diffusion architectures* Training for inference and inference for training* Continuous post-training, deployment, evaluation, and improvement loops* How GLM-5.2 helped optimize the kernels serving GLM-5.2 itself* Why faster networking could unlock dramatically faster decoding* Continual learning, KV-cache compaction, and persistent model memoryShow Notes* How to build a day-0 API for Kimi K3* 22580: From GPT2 to Kimi3, ExplainedPhilip Kiely* LinkedIn: https://www.linkedin.com/in/philipkiely* X: https://x.com/philipkiely* Inference Engineering: https://www.baseten.co/inference-engineering/Ali Taha* LinkedIn: https://www.linkedin.com/in/aliestaha/* X: https://x.com/waterloointernTimestamps00:00:00 Introduction and the 200K-Token Prompt00:03:18 Dedicated Deployments, Speculative Decoding, and Tool Calling00:11:26 Launching Production-Ready Open Models00:19:06 Model Retrofits, Failure Modes, and Nondeterminism00:28:22 Quantization and Canceling Errors00:32:15 The Race to 10× Faster Inference00:40:48 Dynamo, Speculation, and Local vs. Data-Center AI00:50:18 Model Parallelism, Auto-Tuning, and Mega Kernels01:00:55 Rubin, GPUs vs. ASICs, and Custom AI Chips01:10:03 Giant Models and the Limits of GPU Memory01:12:42 AI Video, Quadratic Attention, and Autoregressive Generation01:21:47 Audio, Images, and Diffusion Models01:27:32 Training, Self-Optimizing Models, and Continual Learning01:40:06 Closing ThoughtsTranscriptIntroduction: Baseten, Waterloo Intern, and Inference EngineeringSwyx [00:00:00]: Okay, we're here in the studio with Philip, old friend from Inference Engineering, the book, as well as Baseten and everything that you've done, you and I have done before, as well as Ali. Welcome.Ali [00:00:15]: Pleasure to meet you.Swyx [00:00:15]: Waterloo intern.Ali [00:00:16]: Waterloo intern, always.Swyx [00:00:17]: When did you get “Waterloo intern” as a handle?Ali [00:00:19]: As a handle? Oh.Ali [00:00:20]: I think the rebranding happened mid-March. When I saw it was open, I was like, “I have to take it. Up for grabs.”Philip [00:00:26]: The problem is that Ali is really good at his job and is not gonna be an intern much longer.Philip [00:00:30]: So we have to figure out who's gonna get the handle.Ali [00:00:33]: Well, I'll pass the torch over to the next intern.Swyx [00:00:34]: Oh, okay. It can be, like, you just pass it to another Waterloo grad.Ali [00:00:37]: To another Waterloo intern. No, bruh.Philip [00:00:39]: Yeah.Ali [00:00:39]: Intern.Swyx [00:00:40]: Intern, yeah.Ali [00:00:40]: And no.Philip [00:00:41]: You gotta get an intern from Waterloo.Ali [00:00:42]: Yeah, I've gotta get an intern from Waterloo.Swyx [00:00:44]: Right.Ali [00:00:44]: But they have to follow the path.Swyx [00:00:45]: Oh, it could, but it could come from Baseten, so it's like whoever Baseten gets from Waterloo.Ali [00:00:48]: Right.Swyx [00:00:49]: Has the title of Waterloo.Ali [00:00:50]: It stays in the ecosystem.Philip [00:00:51]: Exactly.Ali [00:00:52]: Halfway through the internship, you either get it or you're out.Philip [00:00:55]: You should also do, like, a big graduation ceremony where you change the handle.Ali [00:00:59]: Just say it.Philip [00:00:59]: For everybody.Swyx [00:01:00]: You guys are good at ceremonies, clearly. We had a nice launch of the book, very successful. But before we get into all that, I wanna start off with a fun question for you. Okay, you're an expert inference engineer. What happens when I send a long query, say two hundred thousand tokens into Baseten's inference? What's the process of query through GPU model routing, balancing, all that? What is all the stuff that we don't think about?Long Context Requests, KV Cache, and Cache-Aware RoutingPhilip [00:01:26]: With a long query specifically, the first thing that I'm gonna ask is, “Have you sent me this query before, or at least part of it?” and I really hope you have, because it's gonna be a lot easier for me and a lot cheaper for you. So the first thing that we're gonna look at is some cache-aware routing, where we're going to see, we probably have a number of instances, a number of replicas up serving whatever model you're hitting. We want to send this one to something with, number one, available prefill workers, and number two, ideally some cached input already there so that we can skip prefill on at least part of these two hundred thousand tokens. If you're doing two hundred thousand tokens, it's probably coding or a multi-turn agent or something where you would expect to have that cached. If you don't, we're gonna have to send it to a prefill worker. We've at least on certain models disaggregated prefill and decode, so you're going to have one set of GPUs that's solely going to process the input, create the KV cache, and get you your first token, and then that's going to be passed over to a separate set of GPUs, which is going to run decode. We're going to iteratively make those tokens. We're probably going to have some speculator model in front of that. I'm going to assume that you're doing coding, and because of that, our speculator model, which assumes you're doing coding, is gonna have a high draft token acceptance rate. If I'm wrong and you're asking me to summarize every Harry Potter book, it's gonna be slower. And then we stream that output to you and account for it, charge you, a couple of pennies and say, “Hey, would you like to send another one?”Swyx [00:03:04]: Except Baseten doesn't charge by pennies.Philip [00:03:07]: Well, yeah, we charge. I'm assuming that we're talking about the public model APIs. If you are setting up a dedicated deployment, then yeah, it's not pennies.Public APIs vs. Dedicated DeploymentsSwyx [00:03:18]: Yeah, one of the key differentiators when I was talking with Baseten initially was that people who want very high volume just need to rent by the box, ‘cause then it's up to you to figure out how to saturate the box.Ali [00:03:31]: And more often than not, it's, like, way cheaper if you're pushing, like, millions of tokens per hour, if you just pay per hour instead of pay per token.Philip [00:03:37]: Yeah, they do. I think that we've increasingly seen a lot of demand for the pay per token APIs, just because everyone wants to try open models, and then once they find a use case that's really sticky, then they move over to dedicated.Swyx [00:03:51]: Is there a best practice on when it's time to swap over?Philip [00:03:54]: Couple reasons. Yeah, reliability, that's a big one, right?Ali [00:03:57]: Like, if they have a very specific use case, they want you to train something specifically for them, like they want their own spec dec, for instance, for their own traffic.Swyx [00:04:04]: Spec dec is speculative decoding.Speculative Decoding and Custom SpeculatorsAli [00:04:05]: Speculative decoding, yeah.Swyx [00:04:07]: You have to explain.Ali [00:04:07]: Sorry. Like, speculative decoding is like, if you have a huge model, right? And so the model is going to be generating one token at a time every single turn, every single forward pass. So we attach, like, this little, like, parasite, like this layer that goes on top of the model, and this model just has to predict. It does three very fast autoregressive forward passes, and it will predict, like, three certain tokens, and then you do one forward stage over the entire original model in order to see if those predictions were correct or not, and then you accept them or you reject them. Now, this draft model is traffic specific, so if you, like, Philip said, if you're summarizing Harry Potter books, I can train exclusively that draft model on Harry Potter books, and I can guarantee you that I'm gonna accept the three tokens every single time. And so with that case, I increase your decode speed. I wouldn't be able to provide this to you if you're a shared endpointSwyx [00:04:53]: YeahAli [00:04:53]: ‘cause I have no idea if you're doing Harry Potter, if you're doing coding, if you're doing English. We don't know. Also, there was a thing in the book that mentioned that if they really cared about a specific threshold, chapter four, I think. Do you remember that?Philip [00:05:06]: Yeah. The things that you can do is you can set a specific, like, batch sizing, a specific, like, parallelism strategy if you're trying to optimize for, like, throughput versus latency. You can. Maybe a NVFP4 quant doesn't pass your benchmarks and you wanna run a model at higher precision, you could do that. There's just a bunch of reasons why you might wanna have your own endpoint and the biggest one, of course, just being, like, you don't have to deal with someone else doing a hundred million tokens of benchmarking traffic at the endpoint when you happen to be trying to serve your users.Swyx [00:05:40]: Yeah. I think one thing that is. That is a classic journey. Like, it's people is asking the, what happens when you type Google into the browser. Tool calling, is that just, you're generating JSON or is there more complication beyond that?Tool Calling, JSON, and Structured OutputsAli [00:05:58]: Certain customers that we have, they have their own post-trained models, and so they demand a tool calling that's not just, like parse a file or go find the weather. It's something that's very specific and you have to do post-training on this. And if the post-training on the model is not good or if the quantization after the post-training to get the inference to be fast, the model will struggle reading the JSON file and reading the tool calling. But it doesn't require its own like sandbox. It's not like it's going to use that tool calling to like escape a sandbox or like it doesn't have to be contained. It can just be a normal dedicated deployment. The challenge with tool calling more and more seems to be that the companies want certain tool calling which is a very sensitive thing to train. And because you're dealing with all of the JSON outputs, if it doesn't like close the end of the request in a very certain manner, you end up with a model that did the tool calling and like the thinking and so as a result of that, it didn't see the result and just hallucinated the result as it decoded. That seems to be the most challenging thing with tool calling, not really the sandboxes model.Philip [00:06:56]: Yeah, that's a challenge on the training side and then on the inference side, there's work that you can do to scope the possible output. So we published this at this point close to two years ago, the solution to this problem which is you make a state machine and you use that to constrain the output to a specific format. So this is the structured output problem. If you remember backSwyx [00:07:27]: Yeah, the specific grammar is,Philip [00:07:29]: Yeah, exactlySwyx [00:07:30]: GML had this thing.Philip [00:07:31]: Yeah. So it's like the old-school “make sure this is only JSON”, return only JSON orSwyx [00:07:38]: YeahPhilip [00:07:38]: Grandma's gonna die type of prompts.Swyx [00:07:39]: Is it BNF grammar? At some point OpenAI had released a thing that was like, yeah, if you want to constrain your output, write BNF grammar, back as NOR.Philip [00:07:47]: In our inference system, it's just a specified output format. And you get the guarantee that your output's gonna be structured along that format. And so applying that to tool calls can like help cut down on. You can still call the wrong tool or call no tool. It doesn't solve the certainty problem but it at least solves the output structuring problemSwyx [00:08:10]: YeahPhilip [00:08:10]: Within tool calls.Swyx [00:08:12]: And MCP is just another form of tool, right.Philip [00:08:14]: Yeah, exactly.Swyx [00:08:15]: As far as there's no special thing there.Philip [00:08:16]: The thing I'm always like explaining to people is the LLM is not capable of doing anything. It's only capable of making suggestions of what to do and then if those suggestions are formatted in a certain way and applied to a system that knows what to do with them, then an action occurs.Swyx [00:08:32]: Yeah. Part of the fun stuff is, this is solved outside of tool calling too. Like in an agent loop if the output is not correct or you're right, like reasoning, tool calling was done in the reasoning trace, just be like, “Oh, I don't know what to do. Let me just try again.” And it might get there after a few tries. And on your point of training, sometimes this is harder in smaller models, so you don't have the same exact quality outputAli [00:08:56]: Right.Swyx [00:08:57]: When you just swap from a big model, right?Ali [00:08:59]: Yeah. I will say that, before, I think we need to go back to inference engineering proper.Ali [00:09:04]: But, I had expected that something would replace JSON because it's hard to stream JSON ‘cause JSON must be complete and you must have open and close brackets and everything. So it's hard to parse something or validate something while it's being streamed. So people invented all sorts of things that are like, I forget the name of some of these alternatives, but it's something like TOML, something like YAML. But JSON seems to be dominant still.Philip [00:09:30]: The JSON outputs aren't that long, right? Like you could have a long-- ‘cause tool calls also contain the arguments in them and perhaps for a certain tool you might pass like a very long argument. But my impression of the median tool call is that it's a relatively small number of tokens, right? So I would expect that speculators are generally fairly good at something as formatted as JSON. And so you would have like a pretty fast decode step there and that the streaming wouldn't be as valuable, but maybe I'm wrong about that.Ali [00:10:02]: I think you're also bounded by the software or that the model is gonna integrate with if the software is built with JSON for the tool calls or if the company that you'- if your customer says that this is how our software works and our tools are interfaced with JSON, you can ask them to like, change their software and say like, “Yeah, this is gonna be better for the model.” but like with the right training shouldn't be that much of a difference. Also more profitable if it outputs more tokens probably.Swyx [00:10:25]: Depends on your business model.Swyx [00:10:27]: It really depends. But I will say that, as a writer with like experience a lot with generated output, I do try to move from text to JSON text which is very long JSON, right? Like there's paragraphs in every field because I'm trying to structure it, right?Philip [00:10:44]: Right.Swyx [00:10:44]: I want you to first make factual statements, then make opinions then make bullet point summaries, have dates, have entity references have your sources for references, all these things. Anyway, so these are things that like I think people who really experiment with structural output have to really care about. But, let's, let's recurse up the stack a little bit. Before we started recording, you mentioned something really cool, which is that there's a lot of engineering that-- inference engineering that goes on when a new model provider releases a new model, right? So let's call it GLM-5.2, Kimi K3. I had previously assumed, especially if it's like, well, GLM 5 to 5.1 to GLM-5.2, like that you've supported them before. Is it that much work?What It Takes to Support a New Open ModelAli [00:11:26]: It's a lot of work.Swyx [00:11:28]: Yeah. Okay. So like, a lot of people, all you guys, right whenever a new model launch like, people rush to say like, “Oh, Hugging Face supports this, Fireworks supports this, Spacetime supports this,” and I'm like, “Yeah, of course we support it.” But what goes into that? What goes intoPhilip [00:11:40]: I think it's more than just support it too, right? It benefits the consumer a lot. Like I think it was with Kimi K2.5 or GLM-5.2 the latest, there was an inference war, right? X provider is at 90 tokens a second. The next day we're at 150. The nextSwyx [00:11:55]: I kinda kicked that off with the GLM-5.2.Swyx [00:11:58]: I wrote a Twitter article about. It got like half a million views,Ali [00:12:02]: Based on being numberSwyx [00:12:03]: YeahAli [00:12:04]: Or it's for something else.Swyx [00:12:05]: Yeah. Which,Ali [00:12:06]: Oh my GodSwyx [00:12:07]: Which then got everyone really excited about, hey, how can we, bend tracks a little bit further and,Philip [00:12:14]: There's a difference between support the model, as in I can make a token out of this model, and support a model, as in I have a production-ready API from this model.Philip [00:12:26]: Getting to the point of I can make a token out of this model is not that hard because generally the, open source inference engines, vLLM, SGLang of the world oftentimes even receive weights ahead of time, maintainers do, or the people making the model merge PRs to ensure support. So you generally can, just get it working on the standard open source stack without too much pain in most cases. The challenge is, every inference company is gonna have own proprietary stack. Some open source components, some in-house stuff. And for any arbitrary model, there's going to be some new stuff. Sometimes you get lucky, like K, two five to two six was, like, pretty similar.Quantization, Speculators, and Production ReadinessAli [00:13:16]: Yeah. It was pure continued post-trainingPhilip [00:13:18]: YeahAli [00:13:18]: If I remember correctly.Philip [00:13:19]: Even in those cases, there's still stuff you have to do. You have to redo the quantization work. You're taking the model from. Generally, these models are not released in NVFP4, and we want them to be in NVFP4 for maximum Blackwell compatibility. So we have to perform that quantization, and, calibrate the quantization to make sure that we're not causing any regression in the model's intelligence. And then we also have to train the speculator, as we've talked about. Generally, we have. We have ZDR, zero data retention on our model APIs, so we don't know exactly the traffic that people are sending us, but we know what's popular. We know that coding use cases are popular. We know that agents, agentic use cases are popular. So we can get public data sets that are representative of that traffic and train general speculators. Now, with speculators today, you need to train the speculator using the base model itself because you're getting hidden states out of the model from running inference on these specific prompts, and that is the training data you use to create the speculator. So there's that process which you need the real model weights for. And then there's of course just the process of, standing up all the infrastructure behind it, loading all this stuff, testing it. And then when there's a new model with a newer architecture, I think that, like, the DeepSeek models tend to be the most challenging as they have, like, the most novel architectural stuff going on, model after model. But every new model has something. Kimi K2 had. Oh, sorry, GLM-5.2 hadAli [00:14:53]: Sparse attention.Philip [00:14:54]: Yeah,Ali [00:14:54]: YeahPhilip [00:14:54]: the DSA.Ali [00:14:55]: Right. Which is brought from DeepSeek.Philip [00:14:57]: Yeah. AndAli [00:14:59]: So you can copy-paste then?Philip [00:15:01]: It kindAli [00:15:01]: I don't know how this works.Philip [00:15:02]: So, like we had to, like, build support for that into our runtime. And you're right, like it is really interesting the way that all of these open source labs borrow from each other. For example, like GLM-5.2 doesn't have vision. So something that, Haley, a guy on our team, if we could take a look at this, he, like, grafted the Kimi vision encoder onto GLM-5.2.Retrofitting Vision into GLM-5.2Ali [00:15:27]: We'll be training the projector.Philip [00:15:28]: Exactly. So if you think about, like, the encoder, there's the encoder, which is the part that looks at the image and turns it into latent information, and then there's the projector which likeAli [00:15:38]: You can say latent space. It's okay.Philip [00:15:41]: And then there's the projector that maps it onto, the model itself, and then there's the model weights. You don't wanna mess with the model weights because you run a chance of making the model dumber at something else for the purpose of giving it vision. So instead, Haley started with just a projector, which is only a handful of millions of parameters.Ali [00:16:02]: That would be, yeah.Philip [00:16:02]: Yeah.Ali [00:16:03]: Can you show the training one?Ali [00:16:04]: Like the way it groksPhilip [00:16:05]: YeahAli [00:16:06]: Very interesting.Philip [00:16:06]: And maybeAli [00:16:07]: That right therePhilip [00:16:07]: Maybe Ali, you should take it from here. You've got a betterAli [00:16:10]: Ooh, double the sandPhilip [00:16:11]: Understanding of this than I do.Ali [00:16:11]: Yeah. You can see, like, he. The way he trained this is really cool. At the beginning, he was training it using just like, “Here's a picture of a mountain. Can you describe what's in this mountain?” And that caused it just like the first, learning walls. Like here you can see this all we're trying to teach it is to translate the encoded. Like it's already taken the encoder from Kimi K. It's taken the image. It'Philip [00:16:31]: Yeah. FrozenAli [00:16:31]: FrozenPhilip [00:16:32]: With adapter.Ali [00:16:32]: Exactly.Philip [00:16:33]: Yeah.Ali [00:16:33]: So the brain is frozen and the eyes are frozen. It's just we're tryingPhilip [00:16:37]: AlignAli [00:16:38]: Interconnect between the eye and the brain, right? So the projector. And so you take the tokens and then he's like, “Oh, can you describe what's in this image?” And he's like, “Oh, it's a mountain,” or it's a person or it's a human, whatever the case is. But that didn't cause complete understanding. So he changed it such that every image was associated with a data set of questions. Like, does this image have a white male? Does this image have birds in the top corner? Does this image have a scientist in it? All of that stuff. And it would have to answer questions correctly. And using not just training on describing an image, but being able to answer question, another question, answer over time. Like you can see the grokking, which is like genuinely insane, that retrofitting vision into a large LLM can learn to that extent. And even for images that it doesn't perform well on, for instance, if you ask it a picture of like Stephen Hawking, “Who is this?” Maybe it doesn't get it, but it will say something like, “This is Albert Einstein.” Like it still understandsPhilip [00:17:25]: Close enoughAli [00:17:26]: That this is a scientist who is a man who has, some significant achievements, all that stuff. So that's like really cool.Philip [00:17:32]: Yeah. So, we've covered Hao Tian before, who the author of the LLaVA paper that did this, a while ago. And I think that's very foundational work for anyone who hasn't done vision work before.Ali [00:17:41]: Same with the CLIP and MetaCLIP, where you go from just captioning to building out questionsPhilip [00:17:47]: RightAli [00:17:47]: Off the image and how much better you can get performance.Philip [00:17:50]: Right. Right. Right. Yeah. But what's, what's so exciting about this is if you look at a model like this. Now, this is a little bit more of a research project. It's not. It got to 56% on MMLU Pro, I think. So not quite frontier. But if you're running this model, you haven't suffered any loss on your GLM-5.2 quality. If you don't have an image, it'll just behave exactly the way it used to. And ultimatelyAli [00:18:14]: Which in the inference code you literally do not include the other part, right?Philip [00:18:18]: Yeah. You would just skip the encoder if you don't have an image input.Ali [00:18:22]: Okay.Philip [00:18:22]: Just confirming.Philip [00:18:23]: YeahAli [00:18:23]: Does it affect a lot on the overall inference side? Like you're not adding much, you're adding a very small vision encoder. These are typically likePhilip [00:18:30]: They're super fineAli [00:18:31]: Less than a billion parameters, right?Philip [00:18:32]: Yeah. It's, - There's a little bit less standardization among vision encodersSwyx [00:18:37]: YeahPhilip [00:18:37]: So the support matrix can be a little bit, sparser. But overall, yeah, it's a pretty, it's a pretty minor component of the overall system. And ultimately what you get out of the system is all of a sudden you have Kimi Vision, GLM weights, and DeepSeek attention all in one model.Open Source Model Grafting and Franken-MergesPhilip [00:18:56]: And that's, I think, a lot of the power and beauty of open source, is that you can take all of these different components and combine them together into a system that's better than anyoneSwyx [00:19:05]: YeahPhilip [00:19:05]: Can be individually.Swyx [00:19:06]: People used to say that you would also do Franken-merges where you would take likePhilip [00:19:10]: YeahSwyx [00:19:10]: Layers from each model.Swyx [00:19:11]: Does anyone do that anymore?Ali [00:19:13]: Well, to your point previously when you were mentioning like, the work that goes into supporting a model when it first comes out, like GLM-5.2 or MiniMax M3 or whatever the case is. Sometimes you do have to like, you do have to switch out some things. Like, for instance, the MiniMax M3 head uses full attention, and with full attention you end up with this like insane bottleneck in spec dec ‘cause you're doing auto-regressive token generation for three tokens, and you're doing this like N squared over all of the tokens that are in your sequence. Your KV cache is like very large because it's not sparse, it's not top K. So we find it better to like, okay, we're gonna replace this, we're gonna replace this layer with a layer from another model that's using like GQA, for instance. And then just with the right training, you can get it to have the same acceptance rate. So it is very possible to retrofit layers from other models and very much needed. If a layer is like inefficient, the training just becomes the challenge, like how do you ensure that you train it properly? Which again to your earlier point is like the mesh between training and inference. As in like you need very good training in order to do fast inference. That's like, I feel like more and more becoming true.Swyx [00:20:21]: Yeah. Anything else on the support side when you say like get it to fully production ready?Loop Detection, Race Conditions, and Non-DeterminismPhilip [00:20:26]: Yeah. I think that there's also a question of just, we can test a model to a pretty extensive degree, but we're trying to get it out quickly and then you see a bunch of other people test it and you get interesting results. There was an issue with, GLM briefly where we had some like mode collapses where it would just output the same token over and over again for certain prompts on certain temperatures. Like once you expose an endpoint to the real world, there's going to be, so many more varieties of things given to it that you're able to, discover and patch things. So it's not just a, day zero process, it's then like for the first week, for the first month, if a model remains popular, like how do you both fix bugs and then continue to push the envelope on performance?Ali [00:21:21]: What do you mean you don't want your model outputting S?Swyx [00:21:24]: Is there loop detection on that stuff, by the way? It still happens like quite a lot, which is surprising.Ali [00:21:30]: We have like we, in our endpoint, like if a model was to output the same token like four plus times, we just cut the generation. We say like, “Oh, sorry, this-- Like try again,” or like we will reprocess the request. ‘Cause we know then, like if it, like if, yeah, it's four times the same token, it's probably collapsed.Swyx [00:21:45]: Yeah. Is there a way to opt out in case I really want that?Ali [00:21:48]: You want that?Ali [00:21:50]: I think there's a way that we have to handle it. I'm not exactly certain, but I feel like in certain models, like when they output something like you can imagine, like a table for instance, and so they want, they wanna draw like 12 dashes and 12 dashes. Yeah, I think there's a way for that to happen. I think we only do it on certain tokens. Like we exclude certain special characters.Swyx [00:22:07]: Yeah.Ali [00:22:07]: So we only do it on like certain like S is the most common almost. GLM-5.2Swyx [00:22:11]: OhAli [00:22:11]: And I think it was DSV 4 as well. Like you'd just have like looping issues where like you literallySwyx [00:22:17]: ItAli [00:22:17]: Just have like S.Swyx [00:22:18]: Yeah. Is there a special, something special about S? No, just randomlyAli [00:22:21]: It just seems to be the one token involved.Swyx [00:22:23]: Yeah. And it'Philip [00:22:24]: Is thereSwyx [00:22:24]: And it's only temperature 0Ali [00:22:27]: NoSwyx [00:22:27]: Even at other temperaturesAli [00:22:27]: Even at like 0.9 or whatever, it will still, it will still collapse.Swyx [00:22:30]: That's weird, right?Ali [00:22:30]: It's, it is an inference problem to be honest, like a software problem. Like oftentimes, the image you run will-- like NVIDIA will release an image for instance, and if we will upstream the changes from their latest TensorRT-LLM image into our stack, we'll find that it fixes it. Or oftentimes this will only happen in an inference engine that you're using like SGLang. But if you were to switch to vLLM, that isn't the case. So it seems to be like an extremely like deterministic software issue and not really a model issue. It's not like a weights problem. Like I'- we'll say like, “Oh, it's a problem with the quant. We did PTQ wrong,” right? But that isn't, that doesn't make sense because the same weights used with a different inference engine does not repeat the problem. And sometimes it's, the kernels that are being used in the backend have like these very subtle sometimes race conditions, where if you were to use this model hosted on one cluster, you will never get this problem.Swyx [00:23:19]: Oh my God.Ali [00:23:19]: But if you host it on a different cluster, you will. And the reason is the KV cache transfer from a node to node in that one cluster is using a slower interconnect than the node to node in another cluster. So that exposes the race, whereas in another cluster it doesn't. So then you end up just like, okay, this model is not gonna be hosted on this cluster. We're gonna host it on, another cluster because that cluster exposed that problem. But then it ends up with like, okay, is it the software? Is it the model weights or is it the hardware?Swyx [00:23:42]: There is a thing about this with temperature 0 still not being deterministic, right?Ali [00:23:46]: Right.Swyx [00:23:46]: Mostly because of hardware. Even at temperature 0 same model, you won't always get the same output.Swyx [00:23:52]: Even-- But I'm surprised by the race condition one because, I thought PyTorch was a graph that like guarantees that you at least, execute things in the right order.Ali [00:24:02]: Well, yeah, true. Like I'm not, I'm not saying that there is. Like well, you have things like PTL optimizations where like you can start a kernel before the end of the previous kernel, and that's like ‘cause you want to do that because there'sSwyx [00:24:12]: It's like pipeliningAli [00:24:12]: Expense. Exactly.Swyx [00:24:13]: Yeah.Ali [00:24:13]: But it'- But you don't do it cleanly. Like you overlap a little bit of the execution. No, it is very possible that the kernel itself, like that one block that is supposed to be running in this instance of time, that kernel itself has a race condition. For instance, like a missing barrier. Like often if you're designing a kernel and you want it to make it to be very fast, if you don't test it extensively, you'll, you'll have certain threads access data points from registers before they've been written to by other threadsSwyx [00:24:36]: YeahAli [00:24:36]: For example, because like your barrier is wrong or your synchronization was wrong. But yeah, like the testing itself is very difficult in those like, andSwyx [00:24:42]: And there's no like borrow checkerAli [00:24:45]: What does that mean?Swyx [00:24:46]: Like Rust. Like the. If you're trying to have like memory safety It sounds like a comparable problem.Ali [00:24:52]: Well, yes, but you're working in CUDA, right, NVIDIA GPUs. Like- You just need a higher level language like modular Maybe that's what modular is supposed to do. I don't know.Quantization Quality and Vendor FidelityVibhu [00:25:00]: How do you see keeping quality of the model? So you talked about all these steps of, okay, you gotta do quantization, train your own speculative decoderAli [00:25:07]: RightVibhu [00:25:07]: Run on different hardware. Looking at other model providers, okay, you kicked off a inference speed race on the consumer end. What goes into keeping quality the same across them, right? Sure, you can run benchmarksAli [00:25:22]: YeahVibhu [00:25:22]: But, like, how do you determine how much quantization are there standards? What goes intoPhilip [00:25:27]: There's a few things on quality. Most inference optimizations are lossless. KV caching, for example. You are just recomputing or preventing recomputing the same values. Speculation, of course, if a draft token is wrong, it gets rejected. The main lossy optimization is quantization. And that really comes down to, number one, data format, number two, which parts of the model you choose to quantize, which layers, and number three, like doing a lot of calibration on the quantized weights, to ensure that you're preserving all the outliers. There's other tricks that you can do, though. A big one is long context, ‘cause one thing you asked at, right at the beginning is, “Oh, what's gonna happen if I send a 200,000 token request in?” So with a long input sequence, you need to, store a lot more information. You need to process a lot more tokens. And so even if a model has a context of a certain length, you might, as an inference provider, choose to build an API with a shorter context length, and of course a full length one as well. Because if someone doesn't need the full million token context, for example, you can get them better performance. I don't know if that's exactly like quality of the model. The way that I think about quality is to what degree are we faithfully serving the original model? If you think of a golden implementation of a model that performs exactly the way the model is designed to perform, I think of quality as how close are we getting to that, 100% fidelity of the model.Philip [00:27:13]: You can also, of course, think about quality from the training side and how do you push yourself past 100%. But when I think about purely inference optimizations, it's getting faster while staying as close to that 100% fidelity mark as possible. And certainly our standard internally is that, like you should not be able to tell the difference between our API and a, official API. I think Kimi in particular does a good job of vendor benchmarking hereAli [00:27:41]: YesPhilip [00:27:41]: Where they haveAli [00:27:42]: They released an actual vendor benchmark.Philip [00:27:43]: Exactly, yeah.Ali [00:27:44]: ‘Cause they accused, some people, Amazon? There was some provider that was not doing very well on Kimi's benchmark.Philip [00:27:50]: Yeah.Philip [00:27:51]: So, with Reflect we probablyVibhu [00:27:52]: This was a long time ago, right?Philip [00:27:54]: No.Ali [00:27:54]: Yeah, like threeVibhu [00:27:55]: They alsoAli [00:27:55]: Four, five months agoVibhu [00:27:57]: This also happened with, I don't remember which model, but they pulled out quite a few, and then they started a whole chart about this. It might have beenPhilip [00:28:03]: Kimi Vendor Verifier.Ali [00:28:04]: Yeah.Philip [00:28:05]: Yeah.Ali [00:28:05]: Yeah, ‘cause you, ‘cause you'd be pissed, right? Like if you'Philip [00:28:07]: Yeah.Ali [00:28:07]: If like if I'm a consumer and I'm using like Amazon's endpoint for instance, and I've used Kimi and I'm like, “Oh my God, like this is bad,” I'm not gonna say, “Oh, Amazon quantized the model in a bad way.” I'm gonna say, “Oh, Kimi sucks.” Right?Philip [00:28:17]: Yeah.Ali [00:28:17]: So it seems like that makes sense.Philip [00:28:19]: Yeah, they care. They care.Vibhu [00:28:21]: Justifiably.Ali [00:28:21]: Yeah, justifiably.Vibhu [00:28:22]: This is probably a stupid question, but just checking, has anything improved from main quantization?Philip [00:28:28]: Yeah.Vibhu [00:28:28]: Like, is quantization always strictly worse?Ali [00:28:30]: Well technicallyVibhu [00:28:32]: NoAli [00:28:32]: It's a lossy. QuantizationPhilip [00:28:33]: YeahAli [00:28:33]: Is a lossy, it's a lossy implementation.Philip [00:28:36]: Speed improvesVibhu [00:28:36]: Speed improves.Ali [00:28:37]: It the number, likeVibhu [00:28:38]: No, I' always look for inverse scaling laws.Philip [00:28:40]: Yeah.Ali [00:28:40]: Yeah.Vibhu [00:28:40]: This is something I learned from Noam Brown, where like things that normally act in one direction sometimes do.Philip [00:28:45]: Well, technically when you run a benchmark, because these models are deterministic, sometimes your,Ali [00:28:52]: YeahPhilip [00:28:52]: NVFP4 quant is like, two basis points higher than yourAli [00:28:56]: No, it's noise. It's noise.Philip [00:28:57]: Yeah, exactly. I'm like, yeah, it's, it's within. That's why I always say within margin of error.Philip [00:29:01]: And I stopped saying that because everyone assumes that what is, well, within some margin of error, we're barely inside of that to the worst, so we're saying. But yeah, sometimes it's just like, gives you a higher output score. But like Ali said, that's noise. To my knowledge, you're not necessarily making the results better. You're just trying to, again, like keep your fidelity as close to 100% to the original model.Layer Selection, KL Divergence, and Better QuantizationAli [00:29:27]: There is, to your point, research that we did on MP. I don't know if you are able to pullPhilip [00:29:31]: YeahAli [00:29:32]: A tweet we did. One of our research interns, Joshua, I think it's a tweet on how we have 20% better quantized GLM-5.2 than NVIDIA. Essentially what we found throughout like this month research is, okay, quantization is a lossy. It's. You're compressing the data from, occupying 16 bits to occupying, four bits, for instance. And so you're losing some information, and you're trying to minimize that. And so when I say that I'm gonna quantize the model, my job becomes how do I find the layers that I can quantize, and how to find the layers to not. For instance, with image models, I don't quantize modulation layers, and I don't quantize out projections because those two are. Like out projection is what you see as the user. Modulation is what the model sees or understands. Right, exactly. And so to his paper, do you have the. It doesn't have the. Yeah. It's a long paper. I don't know if I can findVibhu [00:30:25]: If there's a part to search or it's probably in the thread.Ali [00:30:28]: It's probably in the thread.Vibhu [00:30:29]: Yeah.Ali [00:30:29]: But the long and the short is it is very possible that quantizing more of the model makes the results. Like if I have a model that I quantize layers one, five, and 10, and another model where I only quantize layers one and It is possible that the model in which I quantized more information is going to perform better because the quantization errors have canceled out. And so what Joshua showed in his mathematical proof where he had like a verifier in, is that you can predict which layers are going to have quantization errors that will cancel out with each other, and you choose to quantize those layers. And so the result of doing this mathematical quantization is you end up with a model that's 20% more quantized than another provider, so you get 20% more throughput of it because there's more layers than running an NVFP4, and your quality is better than that other quant because the layers that you chose to quantize have their errors cancel out, like one layer skewed to the right one layer skewed to the left, one layer skewed to the right. Your final logits distribution is more similar to the original distribution of the model, so you have better fidelity. And so the way we proved this was with KL divergence. So instead of just scoring on the benchmarks, we scored the KL divergence between the logit distribution of the quantized model and the logit distribution of the original full precision model, and we showed that with this technique we get. If your probability distribution on the logits which token it wants to select is more of the same as the original model, you're probably gonna end up staying true to the original model. So yeah, so it seems like previously before this, it seemed like the industry was, well, the more you quantize, the worse it's gonna be, ‘cause the more loss you introduce. That's not exactly, not necessarily true. So yeah, doesn't improve it, but can cancel out.Philip [00:31:57]: I think it might be this, but reminds me a good bit about pruning where you can prune off certain layers.Philip [00:32:03]: But very interesting. Didn't know this was a whole paper you guys put out.Ali [00:32:06]: It's. Fun fact, it was originally 72 pages, this paper, and then we decidedPhilip [00:32:11]: WowAli [00:32:11]: We can't tell. We couldn't release it. So it's now 45.Swyx [00:32:15]: Still 39 pages, so very substantive. We talked about evals and all these things and, like what's possible in terms of speedup? Like it's like probably like the numberInference Speedups and BenchmarkingSwyx [00:32:25]: Thing that people do wanna care about, and it's something that you wrote about in your post. Like official API is 70 tokens per second, and you push it up to 90. Is that like a normal thing?Philip [00:32:36]: So what's cool about working in inference, the reason that I think inference is going to be a useful place to do engineering for a long time, is that if you look at highly optimized domains like, say, finance, if you're in finance, you measure how much better you got in basis points. It's like, “Oh, I got five basis points better, like twentieth of 1% better,” that's huge news because everything is so optimized. When we publish optimizations, it's 20%, it's 100% it's 200%. So there's still probably like a lot further to go, honestly. Like you'll, you'll know that inference is pretty much solved when researchers start publishing about how they got 1% faster at something.Swyx [00:33:19]: Which by the way, because I am from the finance background, in the ‘70s, that was the margin at the time. When you did quantitative finance research, you would findAli [00:33:27]: And like 20%, tens of percent.Swyx [00:33:29]: That's. Yes.Philip [00:33:29]: Yeah.Swyx [00:33:30]: And now it'Philip [00:33:31]: Tiny fractionsSwyx [00:33:32]: For those people interested, look up Andrew Lo's paper. He had a really interesting illustration of quant, stat arb, distribution, narrowing down from like those kinds of 20% differences in the ‘70s, down to nothing today, which is very cool.Philip [00:33:48]: Exactly, and we're at the beginning of the same type of thing. Now benchmarking is hard. I think anyone will tell you that, and benchmarking provider speeds is hard because there's so many variables that go into it. What hardware are you using? How much load do you have on the system? What's the exact nature of the prompts and input and output sequence lengths? All that stuff. But overall, when you start stacking these improvements, you're looking at multiples. You can look at it. The most common form, of course, is TPS, tokens per second, which is bad naming by us in the industry, ‘cause there's two tokens per second. There's tokens per second, the throughput number, and the latency number.Ali [00:34:31]: TTMT, yeah.Philip [00:34:32]: Like total tokens per second out of the, out of the GPU as a throughput number. Most people only care about tokens per second as the latency number, which we should call ITL, intertoken latency, but we don't.Philip [00:34:44]: Anyway, so you can imagine a standard API without many optimizations for a 1 trillion parameter model operating somewhere in the 30 to 50 tokens per second range for reasonable traffic profile. And we generally see the goal of, pushing to 10X that. But, not necessarily day zero, but by stacking enough optimizations, if you have, say like four optimizations, each of which doubles performance. Or sorry, three optimizations, each of which doubles performance, then you stack that up, that's an 8X gain. That's the order of magnitude that we're working with in this space. We're trying to make things substantially faster, not just go from like 70 to 90.Swyx [00:35:38]: Are you saying you've. You have done that?Philip [00:35:40]: So let's say you have as a reasonable baseline, 30 or 40 tokens per second. You can achieve 10X that. So like on GLM-5.2, if you run it unquantized, perhaps on H100s even, and you're just using an off-the-shelf inference engine with no particular optimizations, no speculator, nothing extra around like KV routing, no disaggregation, you're, you're probably, yeah, looking at that like 30 to 40. You think that's like a reasonable baseline?Swyx [00:36:12]: Right. Right.Philip [00:36:12]: To get to something like 10X, there's a lot of trade-offs that you're making. If we're running at more like a 300, 400 tokens per second range, you are using the best hardware possible. You have a optimized speculator. You have done all of your quantization work. You are Seeing a pretty high cache hit rate. You are running with a reasonably small batch size and a parallelism configuration that is tuned for latency versus throughput, but it is possible. So the spreads that you see if you, like, go on artificial analysis or you go on OpenRouter and you look at, the worst provider to the best provider, oftentimes can hit that range. 10X is of course very aggressive. It's oftentimes maybe more of a four to six times improvement. But that's the performance that makes us really excited, is when we can get these huge gains, not just go from 70 to 90 tokens.Stacking Optimizations: NVFP4, Speculation, and DisaggregationAli [00:37:19]: It's also, like, hardware dependent. Like, ifPhilip [00:37:20]: YeahAli [00:37:20]: If you have a thing where you're serving it on just, like, a node of H100s and then you throw, like, you shard the model across, like, four nodes of B200s. Like, you can definitely increase the speed with just throwing more hardware at it. Like, normalizing for the same exact hardware and the same number of GPUs.Philip [00:37:35]: Yeah. Then you're looking at, like, a two to 4X improvementAli [00:37:38]: Right. RightPhilip [00:37:38]: Depending on the inference optimizations. So yeah, it's. Some of it's, what's the call, and some of it's who's the driver.Vibhu [00:37:46]: If you break down the two to 4X, say the example is run GLM-5.2Ali [00:37:51]: YeahVibhu [00:37:51]: On B200sAli [00:37:53]: YeahVibhu [00:37:53]: Single node, right? What's, like, the cost trade-off for effort to get, like, the last bit of juice out versus what should people just think of, right?Ali [00:38:01]: Spectre quantization. Yeah.Vibhu [00:38:03]: Spectre quantization.Ali [00:38:04]: That's, that's, that's like 95%. LikeVibhu [00:38:06]: And how far does that get you? And how easy is that for the average person to do? So say right I wanna throw the weights of GLM-5.2 on a node of B200s, how easy is it to find speculative decoder- decoder model or already quantized model? How much work goes into it?Philip [00:38:23]: If you're doing it up front, it's quite a lot of work. If you're doing it today, there's going to be people who have published things that you can just, you can just grab some NVFP4 weights. You can grab a speculator. Yeah, if we're thinking about, like, what are the 2Xs we're stacking, going from, BF16 to NVFP4 is, it's not quite a 2X, right? It's like. I think it's about, like, 30 to 40%, from 16 to 8, and then another 30 to 40% multiplied from, 8 to 4. So that doesn't quite get you a 2X, but, like, roughly a 2X. Speculator, roughly a 2X. Disagg on top of that if you're able to get enough hardware and put enough traffic through it, another roughly a 2X. And then you add in some, double-digit percent increase from having just a better runtime with, the latest kernels and stuff behind it. And that's how it stacks up.Ali [00:39:21]: YeahPhilip [00:39:21]: So building each of those, like, building the, quantized weights is, for someone who really knows what they're doing, hours to days of work. Building the speculator, again, like, hours to days of work. And the, disagg setup, hours to days. Well okay, but like once you haveAli [00:39:39]: Once set up. Once set up. YeahPhilip [00:39:40]: Yeah, getting disagg working for the first time, I'm saying, of course, is very difficult.Philip [00:39:44]: The marginal implementationAli [00:39:48]: Like, if you're just grabbing, like if you are a person, like just a normal consumer who has access to, like, a node of B200s and you're wondering, “How can I just host it myself?” You don't need to quantize the model yourself. There's always gonna be, like, an open source quantized checkpoint. NVIDIA's gonna push one out if no one else does. You. Usually, the providers will have their own spec dec that they've trained as well. You don't need to train your own spec dec. You can just use that as well.Philip [00:40:09]: Yeah. Like, GLM-5.2 has its own MTP.Ali [00:40:13]: Right. Right.Vibhu [00:40:14]: What's multi token prediction?Philip [00:40:15]: Yes.Ali [00:40:16]: I'm justVibhu [00:40:16]: Can you explain that?Ali [00:40:16]: I'm just an expert.Ali [00:40:18]: I can do it for you in case I get it wrong?Vibhu [00:40:20]: No.Vibhu [00:40:21]: Yeah, you should correct if we're wrong, but their multi-token prediction can be used for self-speculative decoding.Ali [00:40:27]: I'm not sure. I'm not gonna correct that.Vibhu [00:40:28]: Okay. I'm semi-confident in thatAli [00:40:30]: Okay. YeahVibhu [00:40:30]: But someone can check. But it's useful to paint the story of, okay, not just the average person, but say a company wants to switch from serverless inference I wanna throw this up on. I wanna rent some GPUs, throw it up. These are the steps you take to do significantly faster than just put it behind vLLM.Ali [00:40:48]: Right.Vibhu [00:40:49]: I was waiting for a mention of Dynamo.Vibhu [00:40:51]: I feel like, that's supposed to be the baseline that you measure against.Dynamo, KV Routing, and Disaggregation ToolkitsPhilip [00:40:55]: I would think of Dynamo as less of a box system and more of a toolkit for building with. So when we talk about doing aware routing, when we talk about doing KV offloading, when we talk about doing, PD disaggregation, Dynamo fundamentally is. By the way, Dynamo is an open source library from NVIDIA.Ali [00:41:17]: We've done a pod with KylePhilip [00:41:18]: OkayAli [00:41:19]: Kyle Cranin.Philip [00:41:19]: Cool. So then your listeners know then that it supports all the different inference frameworks. And it is multi hardware, which is interesting.Ali [00:41:28]: But it's just a router, it's not like an optimizer layer.Philip [00:41:30]: Yeah. All it does, like, what Dynamo is good at, it is a library for moving information around your cluster, around your hardware. So if you have, KV cache on one place and you need it to be somewhere else, Dynamo coordinates NIXL for you to move that around.Philip [00:41:49]: That doesn't mean that, like, out of the box, you just say, “Pip install Dynamo,” and then you get, like, a massive performance speed up. It's more of a developer toolkit.Ali [00:42:01]: Yeah. I would have said it would. It comes with a set of defaults that you can then swap out.Philip [00:42:06]: It does. If the industry at large, I think, was, like, rolling out all of these deployments, standard, then I think it would be, like, a credible baseline. But, we've got to, we've got to benchmark against, like, what we're seeing in the wild.Speculative Decoding Methods: Medusa, EAGLE, n-Gram, and Spec-SpecVibhu [00:42:23]: I did wanna talk a little bit more about PD disagg, because that is probably, like, number three after quantized and speculative decoding. In your book though, I was just gonna pull out the book.Philip [00:42:31]: Yeah.Vibhu [00:42:32]: Like section 522 on Medusa, 523 on EAGLEPhilip [00:42:35]: YeahVibhu [00:42:36]: 524 on gram.Philip [00:42:37]: It's 55, would be disaggregationAli [00:42:42]: Yeah. Well, no, I just wanted to dwell a little bitPhilip [00:42:44]: YeahAli [00:42:44]: The other. Like, so what do you choose to include? What do you choose to not to include? Because there was all these other techniques.Philip [00:42:51]: Yeah.Ali [00:42:51]: Are these still relevant? Because I think they came out, like, a year and a half ago maybe.Vibhu [00:42:55]: Medusa is quite old.Philip [00:42:56]: Yeah, Medusa's old.Ali [00:42:58]: It was old.Vibhu [00:42:58]: But is it in the book as a good, here'sPhilip [00:43:01]: BaselineVibhu [00:43:01]: Baseline vanilla understand it?Philip [00:43:02]: Like you should know this.Vibhu [00:43:03]: Like I read the paper, I'm like, “ it makes so much sense.”Philip [00:43:05]: Yeah.Philip [00:43:05]: So with the book, I had a couple goals. One was to give people just a working vocabulary for the space as a whole, and the other was to give them some intuition about how each of these techniques works. As I mentioned in my AI Engineer talk, which is the first public addendum to this, the speculation space has moved much faster than everything else. So yeah, even at the time that I wrote the book Medusa, I very much included as a way for people to understand how the space evolved rather than what the most modern technique is. And now of course, there's DFlash, dSpark. There's, there's newer techniques even than EAGLE, although EAGLE is still very commonly used.Ali [00:43:51]: SpecSpecta.Philip [00:43:52]: Yes. Speculative decoding.Vibhu [00:43:54]: What canAli [00:43:56]: Oh, it's a paper by Tri Dao and it's like, it's doing speculative decodingVibhu [00:44:00]: HuhAli [00:44:01]: For the speculative decoder.Philip [00:44:02]: Oh, in spec- oh my God.Ali [00:44:02]: It's literally just an another. It's like, yeah, that's the most simple way to explain it, and it seems like he got trivial speed ups there. But it seems that the complexity with training, it's almost like in our mind at least, it's almost as complex as training GANs. Like it's like a very delicate balance and oftentimes you, it's just but yeah, it's literally speculative decoding on speculative decoding.Vibhu [00:44:21]: Speculative.Ali [00:44:22]: Yeah. We saw this paper.Vibhu [00:44:24]: It's interesting, right?Ali [00:44:24]: Yeah.Vibhu [00:44:24]: I wouldn't even expect it to be very particular to train, I wouldAli [00:44:29]: Right.Vibhu [00:44:29]: The naive part of me is like, okay, train speculative decoder.Ali [00:44:32]: But like, and it makes sense, like the whole idea of speculative decoding is you. It's like, it's like almost like the iPhone auto predict version but for a normal model, right? Like you're just, you're just, generating three tokens and you're like, okay, I'll do prefill on them. And so you save those three turns for your original model. Now your speculative decoder is doing three turns of auto regression, so why not just have an even smaller model?Ali [00:44:53]: The other question there is what are the size of speculators? So say forPhilip [00:44:58]: Right. It's like a billion parameters.Ali [00:45:01]: Like for MiniMax, it's. Yeah. It's like one layer. It's like one 60th of the original model usually.Philip [00:45:06]: Yeah. I think we should do a paper when we get back to the office.Philip [00:45:10]: SpeculativeAli [00:45:11]: SpeculativePhilip [00:45:11]: Decoding.Ali [00:45:13]: No, it's, it does seem like how, when do you stop? But then it also seems like if you're able to train spec-spec decode for instance, right? Like if you're able to have a small model that is accurately predicts what the intermediate speculator is gonna predict, that is able to predict what the original target model's gonna predict, then why not just use that smallest model directly, right?Vibhu [00:45:34]: Yeah. This isAli [00:45:35]: Like it seems likeVibhu [00:45:35]: Adjacent to the routing problem.Ali [00:45:36]: Right.Vibhu [00:45:36]: Yeah.Ali [00:45:36]: Right.Philip [00:45:37]: The thing with speculators is one of the practical constraints on using them is that you do have to run a small model on the same hardware that you're running the big model on. There is a orchestration and resource competition problem inherent in that, and that is one of the constraints on speculation in general, is that draft tokens cost resources to create and cost software complexity to manage. And so if you have like infinitely recursive speculators, you add in quite a bit of that complexity on the actual implementation within the inference engine as well, not just in the training process.Vibhu [00:46:17]: I was gonna say, I would wonder if you could do similar, like distillation and pruning of, it's the same thing, it's just a model. Can we not just distill a lot of the weights, quantize the speculator, out of my domain? The question that also comes up is, this is all for big server workloads, right? How much of this applies to, say I have this MacBook, I wanna run Gemma really efficiently. Similar problems, not the same?Local AI vs. Data Center InferencePhilip [00:46:45]: Pretty different. I talked to Selo, about this on his podcast a couple weeks ago. The difference between inference engineering for the data center and for production workloads versus inference engineering for local AI, is that we start with fundamentally like different constraints and different goals. With local AI, it's how do I fit this model onto my hardware and then make it less dumb? And with data center influence, it's how do I load this model and then make it less slow? And we care about less dumb, and they care about less slow. But the local AI inference engineering ecosystem, I think has a lot for us to learn from in the data center space. They are experts in various forms of quantization, including dynamic quantization that we just don't touch, in the pruning, in the distillation, in the, layer removal. There'Ali [00:47:42]: Layer removal matters less.Philip [00:47:43]: Yeah. There'Ali [00:47:44]: No one loves pruning really.Philip [00:47:45]: Yeah. Well, but the, but they doVibhu [00:47:46]: Which is surprising, right? But that's, that's a whole different thingPhilip [00:47:48]: Just to fit something on the laptop.Ali [00:47:50]: Right.Philip [00:47:50]: So yeah, it's a, it's an interesting, it's an interesting space. Not necessarily that like their techniques make sense for us to do in the data center, because we have different resources and different goals, but more that the process as well as the openness of that field is something to, admire.Ali [00:48:12]: Yeah. Like to your point, like, certain optimizations that would. Like for instance, Turbo Quantum Sharper, like it made such huge hype on that and we did like a whole deep dive on Twitter and like said, what is it? How does it work? Why is it good or not? And it took off and it was implemented on local devices because your memory bandwidth is so slow on like a MacBook, for instance. But try putting the same thing on like an NVIDIA GPU on a B200 Turbo quant would not be. Like, it would not be used. Like, NVIDIA - Like, NVIDIA made it clear that this is not a good optimization, and we've seen it firsthand where the overhead of doing dequantization, quantization of, in the kernel itself with turbo quant kernel, each end is much slower than the time that you save from doing the bandwidth. ‘Cause on the B200s, you have like 3.5 terabytes per second. You don't need decrease the storage that much. You don't need to do, FP4 KV cache. You don't need to use a requant. There's, there's, there's better optimizations to be made. But on Edge devices, it's extremely important, it's extremely useful. So, seems to be, like, different optimizations there, but then they're all uniquely combined with like all you wanna quantize the model, you wanna do speculative decoding, like certain common prefixes with bothPhilip [00:49:18]: Principles.Ali [00:49:19]: Yeah, exactly. Exactly. Exactly.Philip [00:49:20]: They also do a lot of work on, model parallelism, especially over, heterogeneous topology, where you have, some sparks and they are wired together with, Ethernet, DGX sparks.Ali [00:49:35]: Yeah, this is the Exo Labs guys.Philip [00:49:36]: Yeah. You have, a nu
Der Hedgefonds des 25-Jährigen Leopold Aschenbrenner, war vierfach gehebelt auf die KI-Rally gesetzt, lag im ersten Halbjahr 450 Prozent im Plus und musste dann innerhalb von Stunden fast alles verkaufen. Ken Griffins Citadel hat die Reste eingesammelt. Pip erklärt, wie Margin Calls funktionieren, warum so ein Blocktrade für den Käufer beinahe risikofreies Geld ist und welche drei Erklärungen es für Aschenbrenners Aufstieg gibt. Danach senkt OpenAI die Preise um bis zu 80 Prozent, was zu der Frage führt, ob es je eine Softwarekategorie gab, die so schnell billiger wurde. Es folgt die große Earnings-Runde mit Apple, Microsoft, Meta, Amazon, Reddit und Robinhood, und die Beobachtung, dass zwei Konzerne für denselben Capex völlig unterschiedlich behandelt werden. In der Schmuddelecke will Josh Kushner Anteile an der Weltmeisterschaft kaufen, und Google Earth lässt jeden ein Atomkraftwerk in den Iran setzen. Unterstütze unseren Podcast und entdecke die Angebote unserer Werbepartner auf doppelgaenger.io/werbung. Vielen Dank! Philipp Glöckler und Philipp Klöckner sprechen heute über: (00:00:00) Aschenbrenner und Citadel (00:20:11) OpenAI senkt Preise (00:30:00) Anthropic-Modelle hacken (00:32:55) OpenAI-Umsatz (00:36:00) Tesla und SpaceX (00:39:35) Apple (00:43:58) Microsoft (00:46:18) Meta (00:54:23) Amazon (01:08:25) Reddit (01:09:24) Robinhood (01:12:50) FIFA-Ultimatum (01:15:12) Gefälschte Satellitenbilder (01:17:43) LinkedIn-Slop-Button (01:23:46) Pentagon gegen Anthropic (01:26:25) Durow Shownotes Situational Awareness sucht Kapital nach KI-Ausverkauf - ft.com Citadel kauft Aschenbrenners Aktienportfolio - ft.com OpenAI senkt GPT-5.6-Preise um bis zu 80 Prozent - axios.com Anthropics Modelle hackten drei Firmen im Test - wsj.com Juli-Umsatz uebertrifft das ganze zweite Quartal - cnbc.com Tesla erwaegt Verkauf des China-Geschaefts - wsj.com Apple-Quartalszahlen im Liveticker - cnbc.com Apple bremst wegen Engpaessen in der Lieferkette - ft.com Microsoft-Quartalszahlen, Azure knackt 100 Milliarden - cnbc.com Groesster Kurssprung der Firmengeschichte - finance.yahoo.com Meta-Aktie faellt nach Zuckerbergs Agenten-Vision - ft.com Amazon erhoeht KI-Investitionen auf 220 Milliarden - ft.com Amazon-Quartalszahlen, AWS waechst 37 Prozent - cnbc.com Big Tech investiert mehr als eine Billion in KI - ft.com Reddit-Quartalszahlen, Umsatz plus 61 Prozent - cnbc.com Robinhood mit Rekordumsatz durch Volatilitaet - marketwatch.com Infantino setzt FIFA-Verbaenden eine Frist von 53 Tagen - telegraph.co.uk Wie man ein Atomkraftwerk in den Iran faelscht - digitaldigging.org LinkedIn fuehrt einen Melde-Button fuer KI-Schrott ein - 404media.co Richterin zerlegt den Pentagon-Fall gegen Anthropic - axios.com Durows Reaktion auf den russischen Haftbefehl - xcancel.com
#corrupción #partidospoliticos #puertorico El PPD acusa al PIP de inmoral por los donativos en efectivo y le sale el tiro por la culata. | Evidencian que el alcalde de San Juan y la presidenta de su Legislatura Municipal, Gloria Escudero, mienten. | Entérate del telón de fondo de Leo Díaz al grito de Jenniffer González corrupta y Thomas Rivera Schatz el paladín de la corrupción. ¡Conéctate, comenta y comparte! #periodismoindependiente #periodismodigital #periodismoinvestigativo Síguenos en nuestras redes sociales: tiktok.com: https://x.com/Bonita_Radio Facebook: / bonitaradio Instagram: / bonitaradio X: https://x.com/Bonita_Radio
1. Juan Dalmau se las canta al PPD y su política de barricada cuando acusaal PIP de inversionismo político. 2. “Parque si, cemento no”, era la consigna en firme oposición a propuestoestacionamiento aledaño al Escambrón. Las intenciones detrás de la OrdenanzaMunicipal 55 para construir una estructura con capacidad para 950 vehículos hangenerado resistencia de múltiples sectores.3. Educación asegura que recortes de más de $1,200 millones no afectaráninicio del año escolar4. Expolicía cumplirá 75 años en prisión por incesto y otros actos encontra una menor de 11 años5. Puerto Rico reportó 496 feminicidios en siete años6. Inestable el sistema eléctrico: fallas y recortes impactan la generación7. Gobernadora anticipa nuevas medidas para enfrentar la sequía8. Tras más de una década de carrera artística en Estados Unidos, elartista puertorriqueño Ángel Iván Rivera Morales regresó a exponer en PuertoRico con Trópicosmos, una muestra individual inaugurada en I AM Art Gallery, enSanturce.9. Haitianos pierden su protección en EE. UU. mientras el Gobierno alistasu deportación masiva10. Ordenan la captura deEvo Morales por protestas en Bolivia11. Estados Unidosbombardea Irán en represalia por los ataques contra las fuerzasestadounidenses. Este es un programa independiente y sindicalizado. Esto significa que este programa se produce de manera independiente, pero se transmite de manera sindicalizada, o sea, por las emisoras y cadenas de radio que son más fuertes en sus respectivas regiones. También se transmite por sus plataformas digitales, aplicaciones para dispositivos móviles y redes sociales. Estas emisoras de radio son:1. Cadena WIAC - WYAC 930 AM Cabo Rojo- Mayagüez2. Cadena WIAC – WISA 1390 AM Isabela3. Cadena WIAC – WIAC 740 AM Área norte y zona metropolitana4. X61 – 610 AM en Patillas5. X61 – 94.3 FM Patillas y todo el sureste6. WPAB 550 AM - Ponce7. ECO 93.1 FM – En todo Puerto Rico8. WLRP 1460 AM Radio Raíces La voz del Pepino en San Sebastián9. WOQI 1020 AM – Radio Casa Pueblo desde Adjuntas 10. Mundo Latino PR.com, la emisora web de música tropical y comentario Una vez sale del aire, el programa queda grabado y está disponible en las plataformas de podcasts tales como Spotify, Soundcloud, Apple Podcasts, Google Podcasts y otras plataformas https://anchor.fm/sandrarodriguezcotto También nos pueden seguir en:REDES SOCIALES: Facebook, X (Twitter), Instagram, Threads, LinkedIn, Tumblr, TikTok BLOG: En Blanco y Negro con Sandra http://enblancoynegromedia.blogspot.com SUSCRIPCIÓN: Substack, plataforma de suscripción de prensa independientehttps://substack.com/@sandrarodriguezcotto OTROS MEDIOS DIGITALES: ¡Ey! Boricua, Revista Seguros. Revista Crónicas y otrosEstas son algunas de las noticias que tenemos hoy En Blanco y Negro con Sandra.
#corrupción #partidospoliticos #puertorico El PPD acusa al PIP de inmoral por los donativos en efectivo y le sale el tiro por la culata. | Evidencian que el alcalde de San Juan y la presidenta de su Legislatura Municipal, Gloria Escudero, mienten. | Entérate del telón de fondo de Leo Díaz al grito de Jenniffer González corrupta y Thomas Rivera Schatz el paladín de la corrupción. ¡Conéctate, comenta y comparte! #periodismoindependiente #periodismodigital #periodismoinvestigativo Síguenos en nuestras redes sociales: tiktok.com: https://x.com/Bonita_Radio Facebook: / bonitaradio Instagram: / bonitaradio X: https://x.com/Bonita_Radio
Welcome, welcome, welcome to the Distraction Pieces Podcast with Scroobius Pip!This week Pip is joined by Mystery Jets OG BLAINE HARRISON!Mystery Jets are a total indie band sureshot, who are currently in the 20th anniversary year of their debut 'Making Dens' debut. They've been total crowdpleasers since the first days, rocking any venue they've graced (including Banquet Records to name one, who get flowers in perpetuity). Back then it was a VERY different time, and a near-unrecognizable media terrain in which many bands were assigned a persona, or vibe, or aura by the music press. The Jets were not immune to this and were often saddled with this and that from various publications, but thankfully forged ahead, and always retained pure love from their fanbase. Pip catches up with original member Blaine, who is such an easy breezy chat subject and has his own questions here and there too (Jets and Pip share 20th anniversary celebrations this year!). It was a really vibrant era back in 2006, but it's so valuable to see a successful band who are still rocking it with no loss of passion or ambition. A really fascinating episode which, as always, will appeal whether you're a Jets-head or you've never heard of 'em. ENJOY!PIP'S PATREON PAGE if you're of a supporting natureONLINE / TOUR DATESSTOREATTITUDE IS EVERYTHING (accessibility charity)FLEA JOINTBOY WHO RAN AWAYEEL PIE ISLANDSPEECH DEVELOPMENT WEBSTOREPIP TWITCH • (music stuff)PIP INSTAGRAMPIP TWITTERPIP PATREONPIP IMDB Hosted on Acast. See acast.com/privacy for more information.
Yeah it's long…she was a big deal!Part two sees us follow Rosa to Berlin from Zurich, with her doctorate in hand and a fig leaf marriage for a German passport. The partnership with Leo Jogiches has become long distance (rocky) and she's flourishing out on her own, developing long lasting and important friendships with the likes of Clara Zetkin and Karl Kautsky. Rosa Luxemburg's dreams for the working class are big and bold and beautiful, recognising the decimation of human dignity that industrialised capitalism and militarism have unleashed. Rosa meets a devastating and violent end in January 1919, but her legacy and writings are vindicated with every fresh global crisis. Meanwhile the SPD party of Germany won't stop growing, and there will be big bureaucratic changes and opportunities for the politically ambitious.…also an oddly prescient rant about the demise of the Starmer premiership of the UK at the time of recording, early summer 2026, but look sorry not sorry, been carrying this around for a while. ++++++
Pip is back for another mystery in Holly Jackson's Good Girl, Bad Blood (2020) AKA A Good Girl's Guide to Murder book two.There's plenty to admire about this entry, which finds Pip growing up and discovering the world isn't fair in a pretty devastating way.Plus: hating Ant & Lauren; admiring the dark ending; and Max Hasting's trial reminding us of real life Canadian scandals involving the World Junior hockey players and Jian Ghomeshi. Ick.Wanna connect with the show? Follow us on Instagram and BlueSky @HKHSPod or use the hashtag #HKHSPod:> Brenna: @brennacgray> Joe: @bstolemyremote (Instagram) or @joelipsett (BlueSky)Have a mail bag question? Email us at hkhspod@gmail.com Theme music: Rewind Kid "Rhythm Revolution" Hosted on Acast. See acast.com/privacy for more information.
Jensen Huang setzt seinen allerersten Post auf X ab, und zwar für einen Brief zur Verteidigung offener Modellgewichte, den binnen eines Tages 50 Unternehmen unterschreiben. Zwei fehlen: Amazon und Anthropic. Dario Amodei legt daraufhin die eigene Position nach, und Pip nimmt seine drei Forderungen auseinander. Danach geht es um Claude Opus 5, um die neue Sicherheits-Allianz nach dem Hugging-Face-Hack und um die Frage, warum ausgerechnet ein chinesisches Modell die Aufräumarbeiten übernehmen musste. Der größte Brocken ist Nvidias Bürgschaft über 250 Milliarden für ein Rechenzentrum mit 10 Gigawatt, dazu 5 Milliarden für Ilya Sutskevers Labor und eine neue Runde bei Anduril. Pip rechnet nach, wie viel Schulden inzwischen außerhalb der Bilanzen stecken und ob der Vergleich mit 2008 trägt. Dann Sheins Zahlen vor dem Hongkong-Börsengang, Chinas eigene Belichtungsmaschinen, Metas Deal in Louisiana und ein Vorschlag zum Wahlrecht, der aus dem 19. Jahrhundert stammt. Unterstütze unseren Podcast und entdecke die Angebote unserer Werbepartner auf doppelgaenger.io/werbung. Vielen Dank! Philipp Glöckler und Philipp Klöckner sprechen heute über: (00:00:00) Huangs Open-Weights-Brief (00:13:27) Anthropics Gegenposition (00:28:38) Claude Opus 5 (00:31:42) Open Secure AI Alliance (00:36:41) Nvidia bürgt für OpenAI (00:50:24) Nvidia investiert in SSI (00:55:10) Anduril (00:59:04) Shein (01:18:53) CXMT (01:21:40) Chinas DUV-Maschinen (01:25:01) Meta in Louisiana (01:27:22) Google-EU-Strafe (01:28:17) Wahlrecht nach Steuerlast (01:36:17) Europas TBPN Shownotes Tech-Allianz verteidigt Open-Weights-Modelle - theinformation.com Huangs Brief verdoppelt sich auf 50 Unterzeichner - forbes.com Unsere Position zu Open-Weights-Modellen - anthropic.com Claude Opus 5 liefert mehr Leistung für weniger Geld - the-decoder.de Nvidia gründet Open Secure AI Alliance nach Hugging-Face-Hack - reuters.com Nvidia bürgt mit 250 Mrd. für OpenAI-Rechenzentrum - wsj.com Nvidia investiert 5 Mrd. in Sutskevers Safe Superintelligence - wsj.com Anduril verhandelt über 100-Mrd.-Bewertung - reuters.com Shein rutscht vor dem Hongkong-IPO in die Verlustzone - ft.com CXMT legt am ersten Handelstag 466 Prozent zu - ft.com China startet Serienfertigung eigener DUV-Belichter - theinformation.com Wie Metas Rechenzentrum in Louisiana verhandelt wurde - nytimes.com Google begrüßt die US-Einmischung bei der EU-Strafe - news.bloomberglaw.com Shopify-Chef befürwortet Wahlrecht nach Steuerlast - fortune.com Europa bekommt seine eigene TBPN - techcrunch.com
In 2014, Collin Henderson was young, in shape, and had blood pressure in the 150s. Anxiety. Depression. Completely stuck. And no idea why. What broke it open wasn't a book or a seminar. Someone handed him a framework for his mind, the same way athletes get a framework for their body. Proactive. Repeatable. Trained. That framework became the foundation of everything he teaches today. Collin Henderson is the founder of Master Your Mindset, a mental conditioning coach to Nike, Microsoft, Amazon, and Salesforce, and one of the most practical mindset teachers George has ever had on the show. This conversation is a masterclass in what it actually looks like to train your mind, not just hope it cooperates. What You'll Learn In This Episode: Why mindset is conditioning, not motivation and the difference between the training room and the weight room The 6,000 vs. 70,000 thought problem and why the untrained mind always loses The "What x3" coaching framework to get unstuck fast Why identity is the bullseye and how your words are either coding you to win or fail Collin's 4-minute HA Method daily mental workout The bison vs. cattle framework for navigating hard seasons with people by your side The Oz Method for influence, sales, and human connection What "victory goes to the vulnerable" actually looks like in practice Key Takeaways: ✔️If the mind is untrained, the thought will win. If the mind is trained, the thinker will win. ✔️80% of human thoughts are negative. 95% are recurring. You're not stuck because of circumstances, you're stuck because of cycles. ✔️You never outperform your self-image. Identity is the bullseye. Words are your wand. ✔️Prehab, not rehab. The mind gym is proactive work, not something you do after you're broken. ✔️Gratitude lowers cortisol by nearly 30%. It's not soft. It's science. ✔️The 4-minute HA Method: breathe (1 min), I Have, gratitude, I Am, self-talk, I See, visualization. Borrow the future to build the present. ✔️Fear loves isolation. Nobody should worry or win alone. Do you have bison people? ✔️Asking for help is not weakness. Hiding and trying to do it yourself is. ✔️Victory goes to the vulnerable, because the coward and the warrior both feel fear. The difference is what they do with it. Timestamps & Highlights: [00:03] — 2014: high blood pressure, anxiety, depression, and the framework that broke it open [01:27] — Welcome: George's Yoda for mindset has entered the building [04:20] — Why knowing mindset matters and doing something about it are two completely different things [07:00] — The training room vs. weight room: what mental conditioning actually is [09:16] — 6,000 conscious vs. 70,000 subconscious thoughts — and why the untrained mind loses [13:07] — The "What x3" framework to get clarity when you're stuck [15:55] — Identity is the bullseye: how your words are literally coding your brain to fail or win [18:21] — Living by design vs. default — and your right to be average [28:00] — Designing places for your brain to go before pressure arrives [32:38] — It begins before it begins: the high-performance pre-game [35:01] — Victory goes to the vulnerable — and what the coward and warrior have in common [38:42] — The 4-minute HA Method: breathe, I have, I am, I see [44:00] — Gratitude: the research, the results, and why it's not optional [47:00] — Michael Phelps, decisive self-talk, and the number one driver of confident action [56:23] — The Oz Method book: connect with curiosity, clarify with story, collaborate on solutions [1:03:48] — The three Oz characters and the Scott story: from PIP to president's club in 11 months Connect with Collin Henderson He is the founder of Master Your Mindset, LLC, a two-sport Division I athlete, former top-ranked medical sales professional, and mental conditioning coach whose clients include Nike, Microsoft, Salesforce, Amazon, Novartis, #1 NBA draft picks, and Heisman finalists. He's authored seven books and hosts the Master Your Mindset podcast. His new book, The Oz Method, releases June 20th. Website: thecollinhenderson.com Instagram: instagram.com/collinhenderson YouTube: https://www.youtube.com/@collinhenderson6749 Quiet Mind Course: thecollinhenderson.com/quietmindcourse The Oz Method: Available for pre-order on Amazon Your Challenge This Week: Pick one thing from this episode. One. Give it 90 days. Then connect with Collin, follow him on Instagram, try the Quiet Mind Course, or grab The Oz Method on Amazon. Follow George: @itsgeorgebryant Work with George:The Alliance — Community for entrepreneurs training their minds alongside their business. 1:1 Coaching — Limited spots. Apply at mindofgeorge.com/coaching-consulting/ Live Retreats — In-person experiences where mindset and strategy meet. Follow for dates.
The Two KRSTN's are back with a lot of yapping about life and BTS! Join us while Pip catches up Kristen on everything that's been going on with Bangtan! We miss you all a lot!!!!
Alphabet liefert Zahlen, die staunen lassen. Trotzdem fällt der Kurs. Davor geht es um LinkedIn und das Slopometer, das misst, wie menschlich deine Posts noch sind. OpenAI verdoppelt seine Agenten-Nutzer im Wochentakt, während drei OpenAI-Modelle bei einem Sicherheitstest aus der Sandbox ausbrechen und ausgerechnet ein chinesisches Modell den Schaden eindämmen muss. Der US-Kongress zieht prompt einen Kill-Switch aus der Schublade. Stripe will OpenRouter für zehn Milliarden übernehmen, Moonshot raist schon die nächste Runde. Bei den Earnings: Tesla wächst kräftig, verdient aber weniger, SAP kämpft mit der Cloud-Transformation, und Pip formuliert eine steile Übernahme-Wette. In der Schmuddelecke lobbyieren Anwälte gegen Robotaxis und Zocker wetten bei Polymarket auf Waldbrände. Zum Schluss noch eine Milliardenstrafe aus Brüssel. Unterstütze unseren Podcast und entdecke die Angebote unserer Werbepartner auf doppelgaenger.io/werbung. Vielen Dank! Philipp Glöckler und Philipp Klöckner sprechen heute über: (00:00:00) Slopometer (00:11:16) OpenAI: 10 Mio. Nutzer (00:16:35) KI-Kill-Switch & Hack (00:22:24) Stripe kauft OpenRouter (00:34:49) Moonshot (00:37:18) Tesla (00:52:37) Alphabet (01:08:11) Reddit (01:11:00) SAP (01:19:54) Trump gegen Forschung (01:21:15) Anwälte gegen Waymo (01:22:26) Polymarket Waldbrände (01:25:08) Zuckerberg Optimismus (01:26:46) Meta StoryKit (01:28:02) Musk-Interview (01:29:58) EU-Strafe für Google (01:31:40) Reiche Startup-Strategie Shownotes OpenAI-Agenten erreichen 10 Mio. Nutzer - bloomberg.com US-Kongress plant KI-Kill-Switch nach OpenAI-Hack - politico.com Modell bricht bei Hugging-Face-Test aus - openai.com Stripe verhandelt Uebernahme von OpenRouter - wsj.com Moonshot zielt auf 50-Mrd.-Bewertung - xcancel.com Reality bites: Tesla und die Musk-Glaeubigen - wsj.com Pips Analyse der Google-Earnings - linkedin.com Spielt Reddit ein gefaehrliches Spiel mit Google? - barrons.com SAP: Cloud waechst, Gewinnprognose gesenkt - wsj.com Trump leitet Forschungsgelder zu KI um - nytimes.com Anwaltslobby gegen autonome Autos - marginalrevolution.com Wetten auf Waldbraende bei Polymarket - derstandard.at Zuckerbergs KI-Optimismus-Kampagne - axios.com Meta testet KI-Kinderbuch-App StoryKit - 9to5mac.com Musk-Interview beim Economist - instagram.com Substack markiert KI-generierte Newsletter - techcrunch.com EU verhaengt 890-Mio.-Strafe gegen Google - ft.com Kritik an Reiches Startup-Strategie - businessinsider.de
Welcome, welcome, welcome to the Distraction Pieces Podcast with Scroobius Pip!This week Pip is joined by longtime podcast and life friends CHRIS & STU & KUNT for a roundtable episode!Quick catch-up if you're new to the show (trigger warning for repeated "kunt" usage): Chris & Stu host the Hardcore Listing podcast, Stu does Off The Beat & Track and a ton of others, and Kunt is the frontman of the band Kunt & The Gang. Stu and Kunt also do the Acceptable In The 80's podcast. Pip, Chris & Stu form together to do the 'DrunkCast' episodes of Distraction Pieces, where they drink and generate like 5+ whole hour long episodes. So there we go!Now we're all on the same page, here is the hypersquad teamup (minus booze) to chat about the hardest kids in school. Whether you remember them from back in the days (whatever era), or you're familiar with them right now if you're in school at the moment, this roundtable chat will ring bells with you for sure. For those who've led a charmed tough-kid-free life and are over 40, think Gripper Stebson from Grange Hill. There are definitely worse than that but we're using shorthand here! Names have been shortened or indeed redacted, but the memories are mostly clear and chances are that you'll start to recall the kids from your school as the episode unfurls. Or maybe you were the hard kid yourself. Which leads us to the existential side-quest - does a hard kid know they're a hard kid? One to consider. Enjoy! PS: the catchup at the start is mainly due to those unfamiliar with the entire cast here - cos you'll be hearing the word Kunt over about 80-90 minutes more than you might have before. As alwyas with these ones, go steady if you're playing it out loud in a situation with younger fam-folk! PIP'S PATREON PAGE if you're of a supporting natureHARDCORE LISTINGACCEPTABLE IN THE 80sOFF THE BEAT & TRACKSPEECH DEVELOPMENT WEBSTOREPIP TWITCH • (music stuff)PIP INSTAGRAMPIP TWITTERPIP PATREONPIP IMDB Hosted on Acast. See acast.com/privacy for more information.
Bonnie Harlan is the Founder of Prevent Pet Suffocation, Inc., a non-profit dedicated to spreading public awareness of the suffocation dangers pets face from chip bags and other food packaging. Bonnie created Prevent Pet Suffocation shortly after her four-year-old rescue dog, Blue, suffocated in a Frito Lay Cheetos chip bag in 2011. Prevent Pet Suffocation has grown to an international following reaching people in countries all over the world. Bonnie has written numerous articles on pet suffocation and appears regularly on television interviews, radio interviews, webinars, and podcasts. A media piece Bonnie collaborated on with WUSA9-CBS won an Emmy in 2020 for excellence in television. She is available for speaking engagements, interviews, and writing articles for magazines, newspapers, and online publications. Bonnie typically hears from 3 to 4 devastated dog owners a week who have lost their pet to suffocation and had never heard of it before. Bonnie has a Master of Liberal Arts from Houston Baptist University and a Bachelor of Arts from the University of Arizona. She resides in Texas with her husband and her English Springer Spaniel, Pip.Legal Disclaimer: This podcast is intended for educational purposes only and does not constitute advice or professional services by either the host nor any of the guests. WEBSITE: www.preventpetsuffocation.com PET SUFFOCATION SURVEY: https://www.surveymonkey.com/r/petsuffocationsurveyPETITION TO FRITO LAY: https://www.change.org/p/frito-lay-add-pet-suffocation-warning-labels-to-your-chip-bags#INFOGRAPHICS: https://preventpetsuffocation.com/pet-suffocation/Thank you for listening to the Enlightened Pet Behavior Podcast. I hope that you and your beloved pets have found valuable insights for a more harmonious life together. Please remember that this podcast provides educational information only and is not a substitute for professional veterinary or behavioral advice. If you need personalized support, please don't hesitate to contact me to explore how we can work together to achieve your pet behavior goals. You can reach me at www.enlightenedpetbehavior.com or via email at susan@enlightenedpetbehavior.com. Special thanks to Mac Light for composing the podcast's music; you can find him at www.maclightsongwriter.comIf you find the show helpful and enjoyable, please consider showing your support! Subscribing, following, rating, reviewing, and sharing with friends takes just a moment but significantly boosts the show's visibility, helping more pet parents discover it. Thank you for your support!
Have you ever imagined how your life was supposed to go, only to discover that reality didn't match your expectations? Through Charles Dickens' Great Expectations, we explore how Pip's pursuit of wealth, status, and acceptance causes him to lose sight of who he is and the people who truly love him. But when everything falls apart, Pip encounters unexpected and undeserved grace. Like Joe's forgiveness toward Pip, the gospel reminds us that God moved toward us in love while we were still sinners. We don't have to chase our worth, earn God's love, or become someone else. Through Jesus, grace is freely given—and it changes everything.
emocleW, emocleW, emocleW to the Distraction Pieces Podcast with Scroobius Pip!This is your bonus FRIDAY REWIND episode! Today, we catch up with Isy Suttie, originally episode 97 from 2016-05-10.Original writeup below:A slammer of a show lies ahead of you ready for your listening enjoyment, as we welcome one of the UK's finest acting/writing/musical talents, the multi-threat herself, ISY SUTTIE! A wonderful and whimsical (license to use that word has been officially granted by Isy herself, you'll see) chat with Pip and Isy as we begin on the subject of hot beverages in some depth and take off from there! And from Tea Corner we do indeed take off into some awesome places, hearing from Isy on her wide ranging skills as an author/actor/musician, her podcast and radio show and her own early versions of these as a young child, Edinburgh Fringe experiences and the recent juggle of all this while being a mother! SO much more besides... Powerhouse business right here - get inspired and enjoy this wonderful chat.PIP'S PATREON PAGE if you're of a supporting natureIMDBINSTAGRAMTEDx on dangers of smartphonesPIP TWITCH • (music stuff)PIP INSTAGRAMSPEECH DEVELOPMENT WEBSTOREPIP TWITTERPIP IMDBPOD BIBLE Hosted on Acast. See acast.com/privacy for more information.
Philippa is back with Lucy Speed, who plays Stella Pryor in The Archers, for part two — and this time the Facebook group gets their say.Lucy answers questions on everything from whether Stella would save Rosie or Cleo in a disaster (the answer is very clear), to whether a prenup would make life easier for everyone at Brookfield. She talks about not knowing Stella and Pip would get together when she first joined, what Stella's wedding outfit might look like, and whether the baby question could derail the wedding entirely.Also: Thor the miniature dachshund makes his podcast debut, Lucy reveals she played thirteen different characters in Radio 4's Don Quixote, why she has loved radio since childhood, and who Stella would choose as her Race Across the World partner — with a strong case made for Rex.Part one of this interview is available now if you missed it.Topics covered: Stella Pryor | Pip Archer | Lucy Speed | Thor | Rosie | Cleo | Wedding | Prenup | Race Across the World | Rex | Radio 4 | Don Quixote | The Archers 2026SUPPORT ALL ABOUT THE ARCHERS:You can BUY US A COFFEE here: buymeacoffee.com/allaboutthearchersYou can buy our MERCH here: https://www.redbubble.com/people/aboutthearchers/shopDo join our FACEBOOK Group: https://www.facebook.com/groups/1127587031446013/ Hosted on Acast. See acast.com/privacy for more information.
Welcome, welcome, welcome to the Distraction Pieces Podcast with Scroobius Pip!This week Pip is joined by frankly long overdue comic and podcaster ELIS JAMES!Elis is something of a missing piece in the Distraction Pieces puzzle, one of those guests whose appearance was only a matter of time (a long time, sure, but inevitable). It's a pleasure to have him, and as you'd expect/demand, so much ground gets covered. Silence and not stressing it, a Humble Pie album of all things, podcasting with John Robins, the nuance of performance environments and the beauty of old venues, the weird and not awesome world of early morning comedy gigs, performing standup in Welsh, opportunity and being bold, the long and surprising road to the Albert Hall, football, technology adjusted ambition, the Cymru Connection (and the New York Times discovering it), and other favourite podcasts. Oh but so much besides. Lovely stuff, enjoy!PIP'S PATREON PAGE if you're of a supporting natureONLINEMANY ELIS & JOHN LINKSEP 428 Elis & John podcastHUMBLE PIEELIS IN NEW YORK TIMESSPEECH DEVELOPMENT WEBSTOREPIP TWITCH • (music stuff)PIP INSTAGRAMPIP TWITTERPIP PATREONPIP IMDB Hosted on Acast. See acast.com/privacy for more information.
That's right! Part 1! And you'd better enjoy this one, because it's the most fun we're gonna have on this topic! Imagine, you're a young woman, charming, witty, a political genius with an unbridled passion for life and your fellow human beings. You make it to Switzerland, one of the only places women can study and meet the love/biggest political influence /worst pain in the arse of your life. He's rich and good looking. Coalescing here are countless political refugees from multiple different countries, struggling to throw off the yolk of autocratic oppression, and you can switch between multiple different languages to engage with them. Something unstoppable seems to be swelling throughout Europe, an irrefutable appetite for massive change. And you have the intellectual acumen, unstoppable energy and resounding humanity that this cause will need to propel it into the twentieth century. God…can't we just stay here?++++++
This episode jumps back in into my interview with John Fasal, Charlie Campagna and Eric Potter. 3 of the world's top field recordists, working in films and games. They discuss the never ending list of new gear out there and how they approach what gear they use, plus their favourite non traditional types of microphones and when they use them. We also announce the details on how you can take part in this year's Tonebenders Listener Field Recording Stories. Go to https://tonebenderspodcast.com/field-recording-story-submissions/ to get all the details. Go back and here part one of this talk in Eps 365. ______ SPONSORS: If summer has your next project heating up, Sound Ideas is making it easier to build up the library you'll need. During the Sound Ideas Mid Year Sale, you can save 50% on all proprietary items. From impacts and ambiences to vehicles, animals, machinery, production elements, and specialty effects, you can stock up on professional sounds before the next deadline hits. Get ahead of the edit with sounds ready for the scenes, transitions, textures, and surprises still to come. Shop the 50% off, Mid Year Sale now at https://sound-ideas.com/ ________ If you are interested in field recording, you should know about the O-Mini P48 and the brand new O-Mini PIP miniature omni-directional electret microphones. Each one is hand made by Chris Trevino, a practicing field recordist, and a really engaged member of the sound community. He puts a lot of work into making and testing each mic to ensure they live up to his high standards. They are ultra-sonic capable, which makes manipulating your recordings with them a lot of fun. They are also extremely affordable. At $150us for the P48 & $130 for the PIP, they offer a lot of value for a stereo matched pair. Find out more at https://www.chrisatrevino.com/store * Please note that the Plug-In Power O-Minis are currently our of stock and will be available again in August '26. ________ Make sure you are up to date with everything Tonebenders is doing, from upcoming events to the latest episodes by signing up for the once-a-month Tonebenders email newsletter: https://tonebenderspodcast.com/join-our-email-list/ Episode Notes: https://tonebenderspodcast.com/366-2026-sound-effects-field-recording-roundtable-pt-2/ Podcast Homepage: https://tonebenderspodcast.com/ This episode is hosted by Timothy Muirhead
In this episode I'm joined by Pip Henderson, Dr Will Vickery and Prof Shane Pill to discuss a paper which examines how sport coaches conceptualise success. Pip works in the Discipline of Trauma and Injury at Flinders University. She is a member of FHMRI's Health Equity Impact Program (HEIP). Pip's research practices are grounded in meaningful community engagement. Working directly with students, players, coaches, teachers, principals, and sport and community leaders across multiple projects has facilitated her deep understanding of the value and complexities of education, sport, and health across a variety of settings. Pip has worked with a range of local and national education, health and sporting organisations, including Oceania Hockey Federation, Tennis Australia, Volleyball Australia, the Department for Education, and the Australian Council for Health, Physical Education and Recreation (ACHPER). Will is a Senior Advisor at the Australian Sports Commission where he is responsible for the development and delivery of community coach education across Australia. Will has held positions at various academic institutions within Australia and the UK with a focus on sport coaching. He has supervised numerous undergraduate and postgraduate research projects across an array of disciplines (sport science, sports coaching, sport psychology). He has also held a number of cricket coaching and related roles ranging across community through to high-performance sport. Shane is one of Australia's leading coach educators and developers, and thought leaders in physical education. He is Professor of Physical Education and Sport at Flinders University. Shane's research is in the fields of sport coaching and physical education curriculum and pedagogy, and education leadership. He has authored over 300 peer reviewed and scholarly articles. Shane has been awarded Australian Council for Health, Physical Education and Recreation (ACHPER) Life Member & Fellow status for his distinguished service to the fields of sport, physical education and recreation over many years.
emocleW, emocleW, emocleW to the Distraction Pieces Podcast with Scroobius Pip!This is your bonus FRIDAY REWIND episode! Today, we catch up with Guz Khan, originally episode 185 from 2018-01-24.Original writeup below:Mad entertaining from start to end, as Pip sits down with Guz for a truly matey chat but which brings out the best of both! They were first acquainted back around a year ago on set on the upcoming ‘Walk Like A Panther', where they behaved like naughty schoolkids on set by all accounts, and this is a catchup / roundup / ahead look at what Guz is involved with - a huge amount it turns out, and a wide range of topics get handled, such as Guz being new to the whole celebrity life and how surreal it can be (down to the weirdness of having a team), the importance of vibes on set and getting on with the crew, how he was essentially raised by three strong women and how that shaped his personality and humour, the art of storytelling, outrage vs laughing it off, teaching and how moments of discussion and clarity are so important to catch and build on, getting the tone right in filming in his hometown and creating a friendly space to do so, and just a whole ton more… Oh, and his new show ‘Man Like Mobeen' on BBC3! Massive.PIP'S PATREON PAGE if you're of a supporting natureIMDBINSTAGRAMBAITPIP TWITCH • (music stuff)PIP INSTAGRAMSPEECH DEVELOPMENT WEBSTOREPIP TWITTERPIP IMDBPOD BIBLE Hosted on Acast. See acast.com/privacy for more information.
Today, Andy Burnham's Labour leadership bid has been backed by 322 Labour MPs.If no one else enters the contest, as expected, Burnham will be declared Labour leader next week before taking office as prime minister on 20 July. Burnham is currently just one short of securing 323 nominations, which is when it is mathematically impossible for a rival to reach the 81-MP threshold needed to run against him. Adam and Chris discuss.An interim review of Personal Independence Payments (Pip) in England and Wales had found it to be "not fit for purpose". Last year, the UK government asked Sir Stephen Timms to review whether Pip was "fair and fit for the future". His initial report is being published on Thursday ahead of the final recommendations due in the autumn.Adam and Alex discuss with Ben Chu Policy and Analysis Correspondent, BBC VerifyYou can now listen to Newscast on a smart speaker. If you want to listen, just say "Ask BBC Sounds to play Newscast”. It works on most smart speakers. You can join our Newscast online community here: https://bbc.in/newscastdiscordGet in touch with Newscast by emailing newscast@bbc.co.uk or send us a WhatsApp on +44 0330 123 9480.New episodes released every day. If you're in the UK, for more News and Current Affairs podcasts from the BBC, listen on BBC Sounds: https://bbc.in/4guXgXd Newscast brings you daily analysis of the latest political news stories from the BBC. The presenter was Adam Fleming. It was made by Jack Maclaren with Gabriel Purcell-Davis. The social producer was Joe Wilkinson. The technical producer was Jonathan Greer. The assistant editor is Chris Gray. The senior news editor is Sam Bonham.
Welcome, welcome, welcome to the Distraction Pieces Podcast with Scroobius Pip!This week Pip is joined live on stage at the Harry Styles-curated Meltdown Festival by incredible artist TOMO CAMPBELL!To quote London's Cob Gallery, "[Tomo's] abstracted depictions of 'traditional' subjects such as hunting, parades and explorers are, as he puts it, 'never quite solid or whole', yet they exude an extraordinarily rich sense of vision". Now that's just so you know at least a little about Tomo if you're not familiar! After the chat with Pip live on stage you will likely be very inspired to pick up on where the chat leaves off and have a look at his work online (or in person if you're nearby).The conversation itself is fabulous, and features a huge amount to leave you well acquainted with Tomo, including art and class, education, inspiration, the slow creep from Houndslow to London, the choice between football and art, mood iterations (that's your album title right there), arbitrary pricing, ballet coincidences, power in ambiguity and art in application to music. Lovely stuff all in all - BUT - the questions at the end were relatively unsalvageable, so while you won't be able to make them out that well, hopefully the responses will allow for context. Apologies you're left to pick up the slack on those but they weren't available in the recordings so this was the best that could be done. See how you get on there anyway! Enjoy!PIP'S PATREON PAGE if you're of a supporting natureCOB GALLERYINSTAGRAMORIGINAL FLYER FOR EVENTSPEECH DEVELOPMENT WEBSTOREPIP TWITCH • (music stuff)PIP INSTAGRAMPIP TWITTERPIP PATREONPIP IMDB Hosted on Acast. See acast.com/privacy for more information.
John Fasal, Charlie Campagna and Eric Potter are 3 of the world's top field recordists, working in films and games. We previously brought these three legends together in eps 315, and everyone decided it was time for a follow up chat. This time we focus mainly on the craft of recording weapons, including finding a good place to record them, developing your "voice" as a recordist and how the industry has changed in recent years. Big thanks to Formosa Hollywood for hosting us in their beautiful Stage 2. We also announce the details on how you can take part in this year's Tonebenders Listener Field Recording Stories. Go to https://tonebenderspodcast.com/field-recording-story-submissions/ to get all the details. Part 2 of this talk will come out in one week. ______ SPONSORS: If summer has your next project heating up, Sound Ideas is making it easier to build up the library you'll need. During the Sound Ideas Mid Year Sale, you can save 50% on all proprietary items. From impacts and ambiences to vehicles, animals, machinery, production elements, and specialty effects, you can stock up on professional sounds before the next deadline hits. Get ahead of the edit with sounds ready for the scenes, transitions, textures, and surprises still to come. Shop the 50% off, Mid Year Sale now at https://sound-ideas.com/ ________ If you are interested in field recording, you should know about the O-Mini P48 and the brand new O-Mini PIP miniature omni-directional electret microphones. Each one is hand made by Chris Trevino, a practicing field recordist, and a really engaged member of the sound community. He puts a lot of work into making and testing each mic to ensure they live up to his high standards. They are ultra-sonic capable, which makes manipulating your recordings with them a lot of fun. They are also extremely affordable. At $150us for the P48 & $130 for the PIP, they offer a lot of value for a stereo matched pair. Find out more at www.chrisatrevino.com/store * Please note that the Plug-In Power O-Minis are currently our of stock and will be available again in August '26. ________ Make sure you are up to date with everything Tonebenders is doing, from upcoming events to the latest episodes by signing up for the once-a-month Tonebenders email newsletter: https://tonebenderspodcast.com/join-our-email-list/ Episode Notes: https://tonebenderspodcast.com/365-2026-sound-effects-field-recording-roundtable-pt-1/ Podcast Homepage: https://tonebenderspodcast.com This episode is hosted by Timothy Muirhead
With Martyn Gibson's water company links, Neil asks a favour of him to help David and Chris with the Show's equestrian situation. Neil shocks Martyn by revealing that Adam has left Home Farm, quickly realising Martyn didn't know, as Martyn storms over to see Brian. Martyn wants assurance that Home Farm will honour their contract for the Borchester Land harvest. Ed reassures Martyn and Brian props Ed up as his deputy, leaving Martyn to make a pointed comment about paying Ed a manager's salary, and Ed tentatively asking Brian about this. Brian puts him off and suggests a drink. Neil's faux pas leads Susan to wonder about offering Adam work in the shop while Jim's away. They also discuss George and Amber falling out over the Gordons' posh cot, and Bert staying at Clive's. Neil thinks they need a city break, like Jim. Lottie invites Neil and Susan to join her and Chris in the busy Bull, also offering Neil a way to avoid Martyn. She then includes Brian and Ed. Privately, Martyn offers Chris some help with the Show, then finds Ed. Martyn wants Ed to effectively spy on Brian at Home Farm and report back to him. Ed's not keen, and Martyn offers a potential bribe in the form of a large tree surgery job. It appears Ed won't be bought though. At their table, Lottie invites everyone's best idea for Pip and Stella's ‘Lesbi-Hen' party. Susan wins the prize of a free dinner with her idea of a treasure hunt based on meaningful locations around the village.
Welcome, welcome, welcome to the Distraction Pieces Podcast with Scroobius Pip!This week Pip is joined by the very hilarious comic, writer and podcaster MARK WATSON!You can go ahead and add "long overdue guest" somewhere in the salutation there too, as Mark is someone who's been missing from the DPP alumni for way too long. Mark's been busy performing comedy all around the globe for a minute, and at some point you will have heard Pip proclaim his love for 'No More Jockeys' - the theme quiz with Tim Key, Alex Horne and Mark, so you'll likely know him from at least one point of entry. If not, fear not, as this is a lovely chat where no former knowledge is required! Pip and Mark go into several menu items including the perils of being 40+ (get your ailments bingo card prepped), peeing in odd places like a regular stray dog, getting lifts with parents as an adult, famous contacts in one's phone, riders at shows (ie backstage snax, beverages and more), surprise Mum spottings, squeezing material out of rugged situations when it might be better to just not have the situation, and his new podcast project (link below). Enjoy!PIP'S PATREON PAGE if you're of a supporting natureONLINEMURDER OF A FAMOUS BASTARDTHE INFINITE SHOWNO MORE JOCKEYSSPEECH DEVELOPMENT WEBSTOREPIP TWITCH • (music stuff)PIP INSTAGRAMPIP TWITTERPIP PATREONPIP IMDB Hosted on Acast. See acast.com/privacy for more information.
© 2020-2026 Ivy Podcast Discovery LLC
