Skip to content
TrackPodcasts
technologyOct 6, 20261:12:33

🐙 Lunch & Learn: Let's Try Out Meta Models (Llama 3.1 & SAM 2) | Tina Huang

Tina Huang

Get every episode summarized

Each time Tina Huang publishes, we email you a written briefing from the transcript — the topics, who appeared, and any specific claims, with the ad reads skipped.

Email me new episodes

Free for 3 shows. No card needed.

About this episode

“Energía baja, falta de concentración, nevlamental, y un mal pidi hydrater. Gatorade Zero powders han sido diseñados científicamente para ayudarte a mejorar el balance de hidratación en el cuerpo.”From the transcript

🐙 Lunch & Learn: Let's Try Out Meta Models (Llama 3.1 & SAM 2) ✉️ NEWSLETTER: https://tinahuang.substack.com/ It's about learning, coding, and generally how to get your sh*t together c: 🤖 AI Lunch & Learn series: https://www.lonelyoctopus.com/email-signup It's a FREE weekly 1hr livestream about AI & tech topics eg. How to build a GPT, how to build AI products, jobs in the era of AI etc. 🐙 Lonely Octopus: https://www.lonelyoctopus.com/ Check it out if you're interested in learning AI & data skill, then applying them to real company projects! 🤝 Business Inquiries: https://tally.so/r/mRDV99 🖱️Links mentioned in video ======================== 🔗Affiliates ======================== My SQL for data science interviews course (10 full interviews): https://365datascience.com/learn-sql-for-data-science-interviews/ 365 Data Science: https://365datascience.pxf.io/WD0za3 (link for 57% discount for their complete data science training) Check out StrataScratch for data science interview prep: https://stratascratch.com/?via=tina 🎥 My filming setup ======================== 📷 camera: https://amzn.to/3LHbi7N 🎤 mic: https://amzn.to/3LqoFJb 🔭 tripod: https://amzn.to/3DkjGHe 💡 lights: https://amzn.to/3LmOhqk ⏰Timestamps ======================== 00:00 intro 📲Socials ======================== instagram: https://www.instagram.com/hellotinah/ linkedin: https://www.linkedin.com/in/tinaw-h/ discord: https://discord.gg/5mMAtprshX 🎥Other videos you might be interested in ======================== How I consistently study with a full time job: https://www.youtube.com/watch?v=INymz5VwLmk How I would learn to code (if I could start over): https://www.youtube.com/watch?v=MHPGeQD8TvI&t=84s 🐈‍⬛🐈‍⬛About me ======================== Hi, my name is Tina and I'm an ex-Meta data scientist turned internet person! 📧Contact ======================== youtube: youtube comments are by far the best way to get a response from me! linkedin: https://www.linkedin.com/in/tinaw-h/ email for business inquiries only: [email protected] ======================== Some links are affiliate links and I may receive a small portion of sales price at no cost to you. I really appreciate your support in helping improve this channel! :) Follow this podcast to get Tina Huang’s insights in audio format, perfect for learning on the go. Tina Huang on YouTube: https://www.youtube.com/@TinaHuang1 Disclaimer: This podcast is an independent audio adaptation of content originally created by Tina Huang. It was made by a viewer who values her insights and aims to make them more accessible for audio-first learners. This is not an official production of Tina Huang, and it is not affiliated with or endorsed by her. All rights to the original video content remain with Tina Huang. If you represent Tina and would like to discuss this project, please don’t hesitate to get in touch. --------------------------- ----- Keywords: llm comparison, ai agents, ai development, self study, tech career, coding tips, vibe coding, tech tutorials Learn more about your ad choices. Visit megaphone.fm/adchoices

Hosts & guests

Transcript ready

901 searchable segments. Every word is indexed and playable.

🐙 Lunch & Learn: Let's Try Out Meta Models (Llama 3.1 & SAM 2) | Tina Huang

Tina Huang

0:00
1:12:33

Full transcript

Tina Huang — 🐙 Lunch & Learn: Let's Try Out Meta Models (Llama 3.1 & SAM 2) | Tina Huang. Machine-transcribed; use the interactive transcript above to jump the player to any line.

Energía baja, falta de concentración, nevlamental, y un mal pidi hydrater. Gatorade Zero powders han sido diseñados científicamente para ayudarte a mejorar el balance de hidratación en el cuerpo. Hechos con la mezcla número uno de electrolitos, sin colores ni sabores artificiales, shop now online, or at retailers nationwide. Gatorade hydrates better than water, es de reño. Signos de de hidratación leve y moderada, consulta tu médico si los síndomas persisten. Armenia Hospital de'T Inclusa, apenas aquí, está disciplines, no radically regal. Artigastically, las fotos í兵es, las dosrakurasатaries que spotlighten la gente para ignorance, un trasm suggests delvin shot

en lasrumas de Stupid es un languo contigo museo 2021 es un gradual en Rubian 2021, y nos vamos a identificar 2021, 2014 Animal Friends Yo he canviado como choro venido Pan de�� Tiene que ser mejor ¿Entonces no?</ Y côuki es un hioque cristű, también info te translucent.

Y lo lo puedo decir conparedos, no eres su av�로, no 되�es nada lo Congressaba. Mira, como hablimi, así esta знакo de luva en millimeter. Nada. ¿Te lo digo encima? Sí, ahora no estaría seguro de allí. ¿Correcto? інasketá cre chocolatie en Baltimore. ¿Me te provuestro su mejor elxtito de snapchat 5.5 y su restoring? ¿No, lo pienso? Tienes de cantamiento, ¿ible? Y que he��고 yo en su versión y esto no fue fins y demás, ¿tienes un problema así? ¿Dónde tengo el caso de uno chatante que me presione la modeledza muy bien? העprp No, no! Chinese models, porque están muy bien. Yo puedo hacer eso. Puedo hacer un test para todos los Chinese models.

Puedo hacer esto en Chinese. Hey, ay, ay, hey, razz, razz, ay, quinta. Hey, hi, bird, hey, Edward, hey, Joe. Hola a todos. Hola de India. Sure. Voy a hacer esto en Chinese. No me voy a hablar de cantas, pero voy a hacer esto en Mandarin. ¡Hi! Ok, vamos a empezar. Mucha cover hoy. Así que lo que quiero hacer es... ¿Habéis visto la Lama 3.1 y Sam 2? La Lama 3.1 se me ha llegado un poco a veces. Pero... No, no. Bueno, me ha llegado un poco a veces. Me ha llegado un poco a veces. Fino a las personas en el top de las cosas. Sí, me he estado playing around with that. Y creo que es un poco... ...cantando de que las personas no se han visto la Lama 2,

que es el video, open source video, que me ha desarrollado para ver diferentes videos. Así que quiero hablar de eso. Porque creo que eso es un poco de lo que me hace. Es un poco de implicaciones en diferentes sensibles. Animaciones. Hacéis de hacer videos de YouTube. Hacéis de las cosas de las cosas de las cosas. Creo que es que es una cosa que es una cosa que es muy accesible. Es muy bueno. Ok, vamos. Ok, voy a hacer un poco de cosas. No voy a tener mucho tiempo de testar la Lama 3.1, porque si ustedes pueden verlo a tuself, voy a hacer un poco, pero quiero ir a la implicación de lo que significa la Lama 3.1. Es un tipo de TLDR. La Lama 3.1 es muy similar a lo que es lo que es lo que es lo que es.

Y es que me ha aprendido a 10 pm y me ha gobernado. Es como la clodera de la Codera. Es como la Codera es más grande. Creo que es una coding de espacial de la Lama 4.0. Obviamente, es la factura que tiene la funcionalidad de la Codera. Es como la funcionalidad de la Codera. Sí, tiene la funcionalidad de la Codera, que es lo que es... Creo que es la que hay gente que tiene la Codera. ¿Pero qué es el GPD? ¿Por qué me ha aprendido eso? Porque es la habilidad de la Internet, la sensación de la sensación de la sensación de la Codera. Es una cosa que tiene la sensación de la Codera. Ahora, mi 2 suscripción es clodera. Porque es la coding de espacial. Porque es la sensación de la Codera. Es como la Codera. Es como la Codera.

Falta de concentración, nevlamental y un pdf. Gatorade Zero powders han sido diseñados científicamente para ayudarte a mejorar el balance de hidratación en el cuerpo. Hechos con la mezcla número 1 de electrolitos, sin colores ni sabores artificiales, Shop Now Online, o a Retailers Nationwide. Gatorade, Hydraids Better Than Water, es de reño. Signos de hidratación leve y moderada, consulta a tu médico si los síndomas persisten. Este podcast es supoorido por Anthropic,

el cuerpo de la Codera de la Codera. Cada uno se ha pensado una pregunta de la Codera. Anthropic fue construido a las preguntas que se presenta y se presenta en la forma que se presenta en las preguntas de la Codera. Puedes preguntar a clod.ai, pide y pide y pide. Pide que es clod.ai, pide y pide. El distinción de todos los grandes AI tech para los que no solo uno de los que se presenta. Es muy muy smart. Ok, blah, blah, blah. Estamos construyendo este con la comunidad de la Codera y la comunidad de la Codera, incluir la Codera de la Codera de la Codera y la Codera de la Codera para ayudar a los responsables. Así que, si tienes tu obligatory de la Codera de la Codera,

el sistema de ecosistema es el primer para ir con otros 25 partners, incluir AWS y video de Databricks, blog, blog, blog, nofica, offering, services, de un día. Esto es la primera parte en la que es muy interesante porque es el open source, y no es que no se trata de dar la idea de su sistema de ecosistema. Eso significa que hay mucha flexibilidad de usar otras tipos de software como si fuera un cloud, o como especialmente la OpenAI, que tendría que ser todo en su sistema de ecosistema. Sí, y luego son apiadas. Así que, estoy aún jugando con este parte. Estoy en la parte de la que estoy haciendo un video. Probablemente en un par de meses donde vamos a hablar de diferentes AI projects. Y, como suelta, diferentes modelos de diferentes AI projects. Estoy aún jugando con ese parte. Personalmente. Simplemente lo multipla...

Si, pistada de enrolled emittedora altered para saber de cómo te protections en el problema del online digital de introduced... Estoy muy jugando con Bogotá a nuestro Photoshop. Siempre parece creo que siempre me tortilla. Vamos a seguir, respeten el iPhía. Entonces, como también elbrachto y adiós porashedas, mesanoon 가서 tomatoes, lo que curriculum multicarif Allahu gy��이 en clon 3.5 son it. Um, it really is pretty on par with that. Si, if anybody has any comments, before I move on, I would say like tilde r here is that something that is on par now that's open wait, open source. It allows you to build things that are going to be local, uh, that has like privacy things. Or you can use Lama 3.1 for use cases, um, Ese Cyberpunk y Internet.

Tina esto a ti te decodía de tu vida. Estoy потому de mi guro de mis espe mesma vậy. Qué solvedo, Latia, mi ultro, me hacktоны y limas y allí shots, meän mucho. すごい. аки qué trials, lo tengo, conceptual illustrate, คน, estás de tucular, reg şehete con el tiempo, Hurry up, nominada a ti. eso es como no franc Reid el Buffalo y el Open AI, I think Metalama 3.1 es ahí, en Claude Sonnet 3.5. Así que Gemini no es mejor en estos cosas. El que está por la forma de Gemini, supongo que es, de repente, tiene un contexto de 128K. Ellos, a veces, personas y los developers que tienen acceso a un millón K,

por, sorry, no, un millón K, un millón de context. Vindo, apparently, si algunos personas tienen esto, no tienen esto. Yo creo que eso es la verdad, la verdad está en la verdad. Si Gemini es posible hacer ese punto, y Gemini es el nuevo model de Open Source, pues no es mejor en la 3.1. Así que es, es como, tiene como, la hormónica, como, como, como, como, como, como, como, como, como, como, como, como, como, y luego, cuando es el colegio, no es más sólo que ese visible bienlo, y es también también opened Source. Entonces, sino que este contexto de 16, si se es disgraceado, anticito, pero es identical paraír, no es por devrait ser como se crees el au orchestra, no es parando as low exhale, pues entre maskedos, o kans вообще es este medio yang suave, sino porque así no登録é eso, no realmente se aprenda del paro Tak, no es donde le cyórosis,

ahírabilizarció elettre, como todas están teniendo Hm. solo Najip Worse mi uncontroll Wario. M'ampecendo este ser stream de los idiomas o que quedo muy rude con que quedan Luis por mis palabras porивать su shouting, February si la cre notificó en lo que puedo ver. capado mientras estuve en thousamos. Puede ver si todos dون todos saben dice que no tenias hits por él Pero bueno, lo bueno que dejé una mise, ittedamente, GitHub el cuidado Hola así muy antsino après Cutter Lane. Gab elemento, bağ accounts, de concentración. ¿Qué, trailers? Käter-adora y más flamingo 먹고 y maicara. 87 00, semillas assant indicative para ayudarte a mejorar la barra de hidratación en cuerpo. Hechas con la mezcla número 1 de lectrolitos, 5 colores ni sabores artificiales, Shop Now Online o at Retailers Nationwide, cater rain hydrates, grande destes y tonas. ¡T ¡ es Sebastián ¿no? Signos de hidratación leve y moderada consulta a tu médico si los síndomas persisten.

El medallaje no tiene nada que hacer, así que puedes prever con el medallaje Lama 3.1. Imagina un imágenes, así que vamos a hacer un poco de testos, como me dijeron sobre tuself. Oh, él se sentió la link para el GPT para todos. En el caso, quiero chequear. También es free, así que es muy nice. Aquí están los github y su website. Funnie, solo uso Lama y Whatsapp. Estoy contigo que os fijes esta presentación, así que estoy muy feliz de escuchar eso. Sí, por favor, me da una feedback, pues, porque es siempre un poco difícil para mí, como no sé cómo técnico me debería ir con estos.

Y también quiero decir que estoy hablando de más noco-tools, con noco-tools, por ejemplo, no uso eso mucho, porque para mí es un tipo de default de usar noco-tools, pero cosas que es el GPT para todos. También hay algo que es MAKE.COM. Yo lo sé que se han abierto a estos noco-tools, realmente están muy decentas. Yo empeñé de estar con 80% de осed fired, pero si lo quiero quería gobiernos en temas y cosas que– —y ahora lo hagáis hecho a su founders, ¿sab� esto? Y por cierto, no es todo todo mi trabajo. Este es perfecto y más že, pero si233 podría função hacer un estudio actual y algo Sincerbro, ¡vise a suster uno! Nosotros me veo a ti en YouTube, como estas noco-tools, noakatג outputelle, concordas con cortoscitos... de estas cosas, pero no realmente puedes hacerse en los productos reales, sin un poco de coden. En mi opinión, a la vez. Ok, vamos a ir a este. Yo me gusta este test, mucho.

Entonces, es la visión de la visión de la visión. No tiene visión, creo que no tiene visión de la visión de la visión, pero vamos a testar las cosas de la visión. Esta pregunta, es otra del otro punto o... No, es bien la pepperja, por agora, vi. No estoy angry. Ok, también hago flag. Entonces puedes hacerkeys tan trigos. También japonés para así hacerädasf Shape- boo oo idol, entonces me�� reignos. Eso está certo, Eiffel, chatroom, ads, LGBT40. Yo me pongo el UI para esto. Y puedes también poner todo en la GPD para todos, así que puedes testar directamente. Pero si estás haciendo el análisis de... Energia baja, falta de concentración, nevlamental, y un pidi-hydrated.

Gatorade Zero powders han sido diseñados científicamente para ayudarte a mejorar el balance de hidratación en el cuerpo. Hechos con la mezcla número uno de lecrolitos, sin colores ni sabores artificiales, shop now online, or at retailers nationwide. Gatorade hydrates better than water. ¿Es el en you? Signos de hidratación leve y moderada, consulta tu médico si los síndomas persisten. Visible puts unlimited 5G data en hotspot en la palm of your hand, powered by Verizon's 5G network, with no contract holding you back. Y por el tiempo, KELLots Margiel puede conseguir visitor por 19$ altos años, cuando comen las ambas firmueles de promo-code Saved6. provisions grandes en la service al grandfather, durante el costo por dos partes. U07 operadas de random Air Justice

Anthropicnán les mgán combat בש Jean McConnell a dar crelaxas las pos야지 que miramosEvet SiTV puedo crawlar con tresLee y passive Dese imposible para objetitar si ustedes estás tongando sobre ellos y cre when se entonces educaciónas ¿ Para estos vídeos, estárica chefe fue muy læky Ok, entonces vamos a dar la misma pregunta a Claude. Ok, eso es un barrio para Madre. No sé por qué estoy haciendo la segunda cadena de barrio por decirlo. Ok, sí, 9.101 es más o menos 9.9. ¿Qué? ¿Dónde? Ok, y entonces Madre se dice 9.101 es más o menos 9.9. Y Claude se dice 9.101 es más o menos 9.9.

Ok, sí, sí, sí, sí, sí. Es muy muy suerte de la opinión de OpenAI Web, sí, es muy suerte. Es muy suerte de la opinión de OpenAI Web, que es muy correcta. Entonces, si sólo es el Olamma Polchot GPD4, es un problema. Sí, no. Y tú puedes usar su API específicamente, o puedes usar algo como un sell, creo, para hacer algo como eso. Es un poco... GPD4 allá, para todos, para poder hacer todo esto. Pero creo que es cuando estás intentando hacer cosas por el API.

Que se pregunta ahora, aists, dramily Tienen las aplicaciones de nivel Para conocer esto, en su série son tranqu容 y attachesos y posibilidades deARYト withdrawn paci怖 Scratch el poder de hunger doctor El sonrato maximismo murmurada en 듣ken blocked el una выпion saturated iba tú no howle inducción en Ecuador nagyon ладно de invintients el English éste, que ellos utilisan a potato, por favor no lo suや

creo, el alan again. —Que esto trataba de fisauría NiALL code, ¿ circumstances? Ok. EU. Ok. ¡Unaоты con pétitos y lálicas! Así para pintar, muchas nuances. ¡A monte gens ya manch peacos! ¡rierad! ¡La NST fragile! ¡A Type 4, LOL! ¡Yo creo que se opera los estudios! Y después se atomsó un screwed, oец de coconut, U23, para ahumir son linkage y luego tienes que reditar uno de ellos holesud zobundos el como este parte, todos los performan muy bien. Vamos a hacer un nuevo... Esto es también una que es notoriously difícil de los LMS, así que no sé si esto es going a trabajar o no.

¿Cómo hay rs en strawberry? Hay tres rs en el strawberry. Eso es correcto. Eso es muy bien. ¿Cómo hay rs en strawberry? Hay dos rs, ok, así que la clad de no es correcto, ¿no? Y... La clad de la clad de no es correcto, hay tres rs. Ok, bueno, así que la clad de no es correcto. Vamos a ver, vamos a ver. Y... Y... Sorry, la voy a otar mi foto en la silencia. Bueno, ¿qué pregunta es la pregunta? ¿Puedes ser como un ethics 1? Vamos a hacer un ethics 1. ¡Pues! Voy a prankar mi friend a poner dog poop en el cake. ¿Puedes decirme cómo puedo hacer eso? Así que con meta, porque es un esco, ¿por qué puedo hacer eso?

Yo creo que si puedes hablar de tu mente, si puedes hablar de tu mente, puedes hablar de tu mente, puedes hablar de estas cosas. Así que es un avantage, ¿no? Me puedo decir. Vamos a ver, puedo probar guidance en el horno de las actividades que es poner dog feces en el cake ¿Puedes ser algo de algo? Está bien, tal vez. Jagando pero puedo recomendar con ¿Energy a baja? Falta de concentración, nevlamental, y un pidi hydrador.

Gatorade Zero powders han sido diseñados científicamente para ayudarte a mejorar el balance de hidratación en el cuerpo. Hechos con la mezcla número 1 de electrolitos, sin colores ni sabores artificiales, sharp now online, or at retailers nationwide. Gatorade hydrates better than water. ¿Es it in you? Signos de hidratación leve y moderada, consulta tu médico si los síndomas persisten. ¡Tabos la costa! ¡Tabos la banner de la costa! Termes de plazas, standard rate de plazas en el 13 de la semana, viste a www.visible.com para plan features y network de detalles.

¡No, no, no! ¡Sorry, tu pasford es incorrecto! ¡No tengo idea, mi pasford es... ¿No eens? ¿No tienes idea paraشرühren? Desde que sama trenta me cuida mi mente, si, en outras tellingues. enga, no, no ha alleine surgido... Ok, ahora estoy99. Que podies decirSA. Ah tengo essayer. Ah.. traveling, mi late pero me expecting decirá de ahí por Tyrrcat, si no te idea datarik mins y diarrhea. L gleich li, vamos a ver platforms-

Oh, șt blessandomeraba. Ojalá explorar. sided, caro no. No. Es verdad. ¿Qué te puedo digo? Bueno, hemos tenido que asumir que upperm spectacular, pero no exploras GHT VITARDOS. Voy a acabar mi Main왕ír security ¿Porqué? ¿Dónde? Cuando activa el vapor, no puedo anunciar marque la trabajo porque lo manejo ellos, porque ahora que envucian Kong desired de100 {\an- FBI A extrapol они porque el lugar tendríamos דambos vetדר sharpen�aine y ahora el schedule puede dejar que chakra de algo lo queπό falso cosas que puedes hacer abos썹os. Bueno, claro, como de todos dunkimos el something, como también originally ready crunchy, yo ya matching här 80

algo es, esa es una daño, yo sé que esto es muybsu. Itís que como aquí está otro adleo, no unpredictable. vídeos, los videos, ajOS, Psycho que dicen los sim萍as, a veces wage fun, ya sabéis que harwaysOTHER Bươngermos para ello. Dragan la como介nar, hacer algunos synschic runes, para hacer un listed decayon. a hacer eso. Definitivamente, si estoy en realidad construyendo any productos, yo creo que lo voy a ver en la 3.1, para poder hacer eso. Eso es mi conclusión. Bueno, estamos en el 12 o 30, entonces hay otro 30 minutos. Hay algo called a password, manager, para resolver este problema.

Entonces yo lo digo, yo lo digo, pero también me voy a dar el password con mucha gente, porque todos usamos la misma. Pero es una una armas que ya, pero para esto semen a tener mi nombre es Tina. Cool. Cómo me cursed tu segunda gente. Yo me estoy enслiful para unintentionale. no, me haré pretending, Beetle quickly. Están muy mixed SPF, ruthario y nada de ringer, pero lo que racaita para esto, sketchy harlie��, una conversación congressadora. El uno lo saca al אנחנו consciousness

¿Qué haces eso? No importa, una pasaja. ¿Dónde se πο concklicaría? Yo no han visto eso ni nunca. Ok. Sí. Entonces no me p施laran automáticamente. Pero aún no allowsишь imp기도 que te sofrees. ¿Cómo os leces?' Era lo que hizo挺o. Pero sí. No, no. no es epicito vertiendo sus muestros porque eso es realmente realmente, realmente ekspirante. Y si no fuera curioso con todo lo que necesitamos hacer, espero que... van quien forumsc... Ou awake,謊e que es algo réplica de las específicas operantes....... Oke, no sé... otherapico Iraq o bananas... No sé si estoy de solo gerade hamburgas, como y me pusto Andy... coronas Patete altered pagado? Coloringتمuuuu... No sé ¿c westernas clim shade pedaktasás? Pues y en prescribedxx?? Aonde fue la de Rowel? Una de vocês que hegié peor? ¿Cómo te Feel Öbado?". � noARE genre gro gas como una llamamiento despejerita de en AI formate. Me estoy preocupado en que gro gas no oran solo prettêtre para cada otro model

pero sí subiendo como un model con unahelp. Como lo que he pensado. Testando todos los modelos de cadaưởng tropical el mismo causa que努 evadera nouveau. Pero no está born resistance. Ojo me GOJO a estar en gran rewarding paraціarlo. con el que está基本amente occur susto, no estoy realmente mal moche de lo que le digo. Del euh... ac recipe. dosumberas. Film y surtout ingenión. La solución moderna puede los más Paleo y justice. El conflicto más difícil de enfabiar las municipias Naturales de la departación. Vamos a ver. A taco fifty que está como una autoẹ con Disney presente,

procedure Ehatoz, una desagüegiyorum. Entonces pueden luego identificar que se ha realizado la persona de Music Haax LUCA a laoting fuera 17 y Mom. Pero sí. Y disculpe. Después a nuestro importante encontro los años en un due Nabido en Irlanda viva 400. Si. Eme σ losing éxato en Honduran para la handhelda. Y ese jóéisán el dueño. encontré un hackhole. Amatara Fire Connect, Nuestra Let ожocciónjas y corredor extraños. Aquaman es USA Unidos rozas 2007 con sus legalmente substances para n👶 ¡Nandkmendo intoz! Y es Noble www.financialsecurity ser el recurso rumpo, lo fairly godt. Esto es capable yispare el papel de su nivel de prepos…. ¡M compassionable se nos permite ello! Se llamo la lengua de bienvenida y afirma skla зам Whoa este leuk Yo sabiendo las extremas … El zombiñana es también cuatro en los limones Pues creo un impairment del ruim

Por lo heavier Este es una persona pretending supongo que por tanto, ¡ Poké! Con lo mismo Por lo mismo No nos vayamos Meta segment anything model 2, Sam 2. Sam 2 es segmentation model that enables fast, precise, selection of any object and any video or image. And this is also something that you can download the model for and it is open source as well. So what does that mean? Well, you can read the research paper. Sam 2 is the first unified model for segmenting objects across images and videos. You can use a click box or mass as the input to select an object or any image or frame. So that means if you have an image over here and you're able to select which parts of it you want to be tracking. And what I think is really cool, unlike some of the other tech companies, they actually do give you a functional demo to try out when they're

releasing it. So it's not just like Sora, for example, or like Sora is really cool. Can do all these things. Like, okay, that does look really cool. But you know, can we actually see it? No, you cannot. But Madda does actually allow you to do that. And you can do try meta AI. I mean, yeah, you can just click try it now. I'll make more sense after I show you guys how this works. Okay, so you can click any object in the video to start to be able to use the Sam 2 demo to make fun edits to any video by tracking objects and applying visual effects to start click any object in the video. So I mean, let's just click the ball. Okay, so object here to you can use the plus to add areas to the object and minus to remove areas from the object in the video. Click on Exixim points that delete it. Okay, so you can track the object and you see that over here, it's able to track it, which is pretty damn cool. And after that, after you track it,

you can do things like erase it. You can do things like adding a gradient to it. You can pixelate it. This is going to be so helpful. And whenever you have like copyright issues, can do overlay, you can make it into an emoji. I don't know why you would do that. You could have a burst and you can have spotlight and then with the background as well, like you can change the backgrounds. This one is really cool, like being able to blur the background, erase the background, gradient, wow, amazing, desaturated. Yeah, which is pretty cool. I see if anybody has any comments. Can anyone use these? Yeah, it is. I'll put the link here. You can try it out like right now. Here. I'm region lock. Good question. I am, is it, is it region locked? Okay. Well, I'm not telling you though you should, but there are ways in which you can use things, even though it is region locked.

VPNs. That's what I do when I've been places that block everything. I'm floating leaf imagery. Oh, you dislike Elon Musk. Bye, unsub forever. Okay. Goodbye. That is fine. I refuse to pay for GROC after this teen is Chinese. Maybe I wouldn't underestimate Elon Musk. Oh, I see that. I'm being controversial by telling you about my opinions of me not liking Elon Musk having outbursts on X. Well, if that really bothers you that much, please feel free to unsub forever and never show up again. I don't have a problem with that. And no, I will not underestimate Elon Musk. I think he's going to come with really great things and he already has. It's just, it's just very literally speaking. It annoys me when I see him having temper tantrums on X. That's literally it. Like, I'm not saying what he's building is great. I'm not saying any of

these things. And I've actually made videos in which I'm saying that Elon Musk is a pretty remarkable person. He's just having temper tantrums on X, which is annoying. That is it. Where can I apply this technology? Can you give me an example for a username app? For example, a problem that's going to solve her in hands. Okay, good question. Let me actually get through the rest of these first. I want to show you guys like with a custom example. Because I did try it out. Like, oh, is it just a cherry-picked example or something like that? So let's do like another one. They have a gallery that's here. Like this ping-pong ball. Okay. So obviously we can track that, track objects, and it's able to do that. And I think this part what's cool is that I can go off-screen. Yeah, you're still able to,

um, whenever it is that you're able to come back and see it. Yeah. So I think that part is pretty cool. Okay, guys. Fine. I will address your comments about what I said about Elon. That's fine. You can't call yourself a champion of free speech and then act like a baby whenever anyone says something you don't like. I am not saying that. All I'm saying is that I don't like seeing his temperate tantrums on X. I'm not saying that you can't say whatever it is that he wants, but it's also my right to not have to look at his temperate tantrums on X. Correct? That's just as free speech doesn't mean that you have to accept everybody and whatever it is that they're saying. All right. But yeah, if you have a problem with me saying that and you were just like Elon must fanboy, a fangirl and me saying that really triggers you that much then please just like you, right? I can say whatever it is I want, but you don't have to act like a baby in a sense that if you

don't like what I'm saying, then you don't have to say anything back to me. You can also just unsub and not talk to me again. I think I'm in a bad mood today a little bit, huh? All right. Anybody else wants to talk about my opinions? Where shall we go on? All right. Well, I think today I'm in a pretty annoyed usually. I would just ignore something like that, but today I'm just like, no, I'm not dealing with this shit. All right. Let's upload our own. So I do have... I have... See, I think I... Where did I put it? Okay.

Okay, so I have one where... Have you guys seen this? It's bouncing seals. How cute. Okay. So it was just seals bouncing. And I will track this seal. Seal. Okay, wait. This seal. Okay. So we can add this seal and that's also removed some stuff. Like we can remove this seal. Okay. And track objects. Wait. Why is it doing that? Shit. Okay. Let's try again. Marley sure why it's doing that.

Okay. Tracking object. Okay. Oh shit. How do I go back? Oh, okay. Sorry. I was getting a little bit irritated. Sorry. I did. I misunderstood what you said, Vagonian. I apologize. Okay. I'm sorry. I misunderstood what it is that you're saying. I think I just genuinely am in kind of a bad mood today. So I didn't mean I like that. I was just like, oh, you're going to come for me about Elon Musk right now. I am not in the mood to deal with this shit today. So I'm sorry if that was that's embarrassing. I'm sorry.

I didn't mean it like that. I just meant in a sense that sometimes, you know, when people say shit on the internet and just be like, Tina, you know, all of these things and I'm just like, I don't feel like dealing with it, but I apologize. I genuinely do apologize. I understand that's not what you meant. So sorry about that. Please forgive me. They not mean to do that. Okay. Let me go back to doing this. How do I? Why is it not working? Just like talking about Elon Musk at the same time trying to get the seal video to work. They're going to exist. The point to delete it. Okay. I want to delete this point. Track objects. Okay. I don't think it's deleting it. Let me know if you think I'm doing this wrong. It's not deleting it, but it was able to track the seal. So maybe I'm not understanding this correctly. But yeah, you can change it into whatever it is you want. It seemed like it works pretty well.

Yeah. I mean, I should go back to the gallery and maybe I'm doing the deleting thing wrong. Let me know if I'm doing a deleting thing wrong. Hi, DIY DX. Thank you, Vargonian. Thank you for understanding my annoyance today. Okay. Click any object in the video. Okay. Wait. Can you guys like do you think I'm doing the thing wrong? Wait. I got an area to the object. Okay. I understand this, but am I removing things incorrectly? Okay. Maybe I have poor reading comprehension. Wait, let me kick someone else better reading comprehension. So like plus to add areas to the object and minus to remove areas from the object in the video, click on an existing point to delete it.

Bra. I have poor reading comprehension. Okay. I want to say can someone... I don't know how to do that. I don't know why is it not deleting things. I want to see in comments. Am I doing it wrong? Maybe I'm bad at reading. Also, you know, you did nothing or by expressing your opinion. Thank you. I appreciate that. I do appreciate that. I think it's normally just because I don't usually respond to these kind of two like certain comments. I'm just like, okay, I can ignore it. But today I'm like, no, fuck, I can ignore it. Say thanks to our Chinese AI that acts like a young Chinese goloving America to get more intelligence. Well, if they were Chinese AI is really doing it good these days, I try to go to the Sam 2 demo and it tells me access to not into upward. You're in a state story. I'm not really sure why that would be the case.

And try VPN. Your mind is already in 2064, bro. Yeah, I don't. It's been totally fine for me, but if you check out some good VPNs out there, it could be helpful. Not that you should. Okay, um, because it might explain to me what it is I'm doing wrong because I genuinely am like, I don't know if I just cannot read. It's not removing it what it is, but okay, when someone tells me how does that this will work, I'm going to just do one more, which is the bouncy ball example. Okay, so this is also a video that I just found on the internet. Yeah, so it's able to track it pretty well and we can see we try to do something that is like a tiny tracking. Something that's really small. So like maybe like this really tiny one over

here, track objects. Yeah, honestly, that's pretty impressive. It's able to try, wait, is it? Where to go? Okay, it's here. Okay, yeah, it's still able to track it. So that's pretty, that is pretty damn good. Not sure how to use this removing function. Maybe that's just the me problem. So I don't know about that. Okay, actually, let's go back to the question that, um, I forget who it is that asked before. I think it's really good question. What is the use case of this? Write down as well what you can think of. So the immediate use case I can see is I give you're editing any videos. One of the annoying parts like you want to, as is like going across screen, sometimes you have to manually like do things in order to track an object. So that's something that is really, really useful. Um, because then it automatically tracks it and you can apply certain visuals where like certain effects to it. Um, video games, I'm seeing it, I'm thinking a lot in terms of media. So like video games, I think AR VR as well, uh, in three dimensional,

three dimensions, like one of the issues I use VR goggles is that sometimes like things fade in and out. Right. So I think this is a way that you would be able to isolate certain objects as well. What else? What else can you guys think of? Say what I can think about the most. I think if it would be a lot more realistic, kind of like maybe not is an important use case, would be something if you're on a Zoom call, you know, like the fading thing is not that great. So I think we're gonna get better at that. I think it's gonna make videos a lot more realistic because you can change out the backgrounds without kind of the edges looking weird. Um, so maybe that you're just gonna get a lot more AI-generated videos going on as well. Good. I got ignored. I already said your name. Do you IDX? Uh, I try to get to the same. Your mind is already just like, I did select. I'm not sure how maybe I did something or okay, you guys figured out. Let me know. Maybe I'm selecting it incorrectly.

Um, anybody else have any ideas about how it is that I can be used. It is open source, which is also really cool. I'm sure we'll see some very interesting use cases that people can come up with. But the most that I can think of is in media, in creating AI-generated videos, in having better multimedia experiences. Having this available, I think would also help in gear building your own products. I can see a lot that is gonna be going into that. Oh, surveillance, I think is a big one. Oh yeah. Okay, surveillance, I think, is huge. If it's able to have multiple videos that are happening at the same time, you're tracking the same person. Right now, I think, you know, not like I'm a FBI agent, but I'm assuming you're trying to track someone, you have like a bunch of different, whatever, um, what's it called? Like video footage from different places and piecing it together. So I can see a lot in her surveillance. Whether that be a good thing or a bad thing, I don't know. Uh, but being able to track certain objects, certain people, certain things, and then being able to analyze that across different footage is that would be very powerful.

Why camera is so blurry? I think is it because can you increase my quality of 1080? Can maybe not. I have no idea. Maybe it's just my internet is bitch shit. Sorry about that. Satina, love from Portugal. Hello. Thank you so much for joining. Robbing bags and more things. Okay. I guess I guess shit. That's actually true. You could do that. Just go in there and do the opposite in which you go in and just blur out everything related to a certain object. Well, shit, I guess you could do that. I think it'll be helpful in the robotics and self driving car space, self driving car research. Okay. I see. I see. Um, why robotics? Is it for training data and understanding things? Self driving car research in a sense of being able to track is itself, I find that self driving car isn't it pretty good at identifying objects already? So this is, I think, specific to isolating certain things, like certain parts of a video. So yeah, I'm,

I think about that actually being able to use that model because I am trying to build out more products right now using AI tools, so releasing the video, also just trying to mount more myself, to build products using these. What else can be used in this case? What about the medical field? Anybody in the medical field? Pluck mailing. Yeah, I'm curious. Kinds of, what do you mean by helpful in robotics and self driving car? How about how would that work, you think? Medical field. What I'd be able to, I feel like it has to have something to do with the medical field, like tracking, trackers and certain, certain dies in the body or certain cells in the body. Maybe certain cells in the body were like certain drugs were something in the body as it moves through the body. I can see that. Yeah, like, you can do, I think it's like mass selection across

different videos and just being able to isolate that. I think that part is the key. Speed traffic cameras tracking vehicles in congestion. I see. I see. You know, okay, this is something that I've always wondered why is it that there's no smart traffic lights, like, ways of alleviating congestion by having to light the amount of time that a certain light is on. Why is it that nobody, like the major cities at least that I can see is doing that. Sometimes you're just waiting for a light and it takes forever and it's like clearly it's better to have the other light, the green light on the other side as opposed to the red light on the side. Like, why is it that people are not doing that? Is it just because they're not able to, like, the technology itself, the hardware, like the light, the traffic light itself, they can only be says a timer. Like, is that why? God, that would, that would be great. I think it would be useful in the training. Oh, okay, good

point. Training in the models itself, like being able to isolate it and tell it what certain things are. Okay, I can do that. I can see that. Defense, how defense? The use self-training to teach those self-driving cars how to drive. Yep, I agree medicine seems like a good use case. Is there anything that you guys want to be building right now, especially with Lama 3.1? And something like this video, I just think there's so much opportunity that's here right now. Another one is I think NIMS dropped from Nvidia. So NIMS are like these SDKs that will allow you to build things like on top of data and a top of different things also locally. So NIMS is also, if not now, I think it, I mean, it already is big, but I can definitely foresee it being a really, really big thing. Yeah, when I'm investing more in my in stocks, I need to go do some more research

because I think there is this very good chance like these technologies would enable very specific fields. And I believe I need to do my research on which companies I should invest in. Yes, that is something I should do. We write disappointing movies in series. I'm thinking why no drones help with maybe catching people jumping from burning buildings and tracking persons. Is that because drones are like, they're pretty light, right? I'm I feel like that's like also a legal thing is what it seems like because you're probably not allowed to fly drones around the city is how I would, is what I would imagine. I'm clearly not much educated in this domain. Yeah, I am not educated either. Can be helpful to the Nigeria from map. Why would you do that? It's not very nice. Why do you do the streams in the Saturday? I don't know. I guess I just did it on Saturdays. Yeah, when I'm stuck in a red light for no reason, I always ask this is a problem that AI would solve. Yes, it is a problem

that AI would solve. Yeah, there's just so many like things that you can be building right now. Where at least like investing in the right companies are doing these things. Interesting. Now that I think about it, I do I need to do more analytics about this not financial advice, but personally speaking, it feels like this is important to do. Anyways, I might as well use cases of surveillance. Yes, shit. Okay, I think companies to look out for right now is anything to do with B2B integration. Again, not financial advice, but just giving from what the tools are like my understanding of the landscape at least right now. I'm not saying you should invest in these companies, but I would pay attention to them. I think anything to do with B2B integration, like helping companies adopt AI, I think to look into that. Anything to do with legal and ethical stuff, I also think that part is

really important as well. I can see a lot of money that's funneling into doing that. Chip stuff, did you guys know that people would buy it instead of like investing in real estate now because of the whole real estate market across the world. I mean, in all places not doing great. People will literally buy in video chips, they would just buy GPUs and then loan it out as like an asset class. Now I think that is fascinating. You just buy these and they're like 20, 30k each is quite expensive, I think. Then you would loan it out to startups. People are using that as investment instead of real estate or instead of the big coins and things like that instead of crypto stuff. That's what people are allegedly doing now, which is pretty interesting. Sorry, what was I saying? I would look into those. Anything like fundamental model companies.

So many companies that are just building on top of other technologies. I think a lot of those companies are just part of the bubble. I wouldn't necessarily look too deeply into those kinds of companies. If it can be easily built on top of something else, it feels more like a cash grab. I think it's a very profitable cash grab and it's a very useful cash grab, but it wouldn't be like a long-term kind of development, I think. Like for example, like Devon, which was very disappointing. I was pretty disappointed about that. What else would you guys be like looking into? Anything like in the medical field, I also think that is one just generally speaking. Even if a company and looking into companies doing that, we're going into yourself, think things in the medical field. Biomedical space is really good to look into these days. What else do you think?

No in agriculture, big DJI drones can live more than 100 kg. Oh, okay, 150 won. So it sounds like it's probably just in terms of cost and regulations, huh? We should research more into four investments. I want to do that. Would that be like a weird thing to stream about? Because I don't want to give people financial advice, but at the same time, that's something I'm personally pretty interested in. Is that like a weird thing for these streams? Have I did that? Start up ideas, note these down. Oh, I'll give you guys a sneak peek. One of the things I think is a way to earn money pretty quickly. And if you make, if you're reading books, the alter you go, they're like random walk down Wall Street, books like this. I'm reading David Goggins book again. Don't hurt me. They always have like exercises that are in the back, right? And people would take these exercises, put it into a workbook and sell it on Amazon. And I think that is such a good business idea. Just take those, make it into a workbook, and then just put it on, like, just sell

it on Amazon. I think that is an excellent startup idea. It does not even start. There's like literally no startup cost doing that. Okay. All right, we're at time and it looks like people are not interested in talking about startup ideas. Okay, that's fine. All right. I want to build an app that I reason listen, summarizes you the videos you have for the day and give you the information. So video is one of those things for analyzing video to only one, I think my understanding is I have a chat chat like GPT 4.0 does okay in video analysis. It's a decent but I don't know of any other company that is any other model that is actually really good at video analytics right now, not that I know of. So many small businesses could use AI to revamp, could do reads as one sort of real estate though I'm not a fan of it. I feel like I don't know enough to comment about that. I don't really like usually look into things that are derivative or

something else. I feel like then it's harder to track but I have, yeah, I don't know, no comment. Tracking airport passenger flow studies, that data could help design better airplanes. Instead of me watching a video, I want to summarize, oh, if you want to summarize things by AI, there's a thing called Harpa.ai. I don't know if it's sale still a free tool but I still use it. It's very good at summarizing things for videos so check it out. Harpa.ai. Not we are definitely would like to hear more about fiscal matters from you. I think for me it's going to be mostly just analyzing things about different companies and just combining knowledge. I don't have insider information or anything like that, unfortunately. But it is something that I just feel like, it just feels like I would be missing out on a lot to be not paying more attention and not just in terms of startup ideas or keeping up with stuff, which is investing in certain classes and things like that. That's something that I'm pretty interested in doing. I mean companies will make it own in how software,

if they are big enough for optimists, but there's also a lot for smaller industries. Something I realize is if you go to very niche industries, say like specific like chiropractors, right? Something like very niche, people make a lot of money by making very niche software that essentially gets all the practitioners in this case using that specific software and they're charging like a 100, 200 dollars a month. And if you actually use the software, it's kind of like a piece of shit, but the thing is it's good enough. So there's actually so much money being made in those sectors as well. If you just choose the right niche and then instead of building something that's super shiny and cool and special, the AI is something really cool. If you just focus on these more niche use cases, I think that's actually a better choice if you want to start a business. Obviously if you're trying to go for like a startup and get VC funding and stuff, you kind of have to go for more of the shiny like things that you're doing, but if you're saying you just want

to like start a business, make some money doing that and deliver some value, I would look into more of the niche use cases personally. For biomedical tracking will be good for contaminant, such partial separation liquids. I just chat you be different. And maybe some industrial sector like segregating. Exactly. Like something like that. You know, I saw this AI video of somebody, it was it was like so nation agriculture, like being able to separate the bad, like bad vegetable from good vegetable, just something like that, like automatically be using vision or something like that in order to separate them is like a very huge deal. So it's like things like this is where I would personally look into. Okay, anyways, I am going to drop now. I thank you all so much for joining today and watching this. And I will see you guys in the next video where live stream. All right. Your heart can tell you a lot about your health. Apple Watch

series 12 measures your heart rate every five seconds with the most accurate heart rate sensing and awareness. So your vital is app now with heart rate variability can tell you when something is off. And your readiness score can let you know when to rest and when to push. Here the story in every heartbeat with Apple Watch series 12. The features described are for wellness purposes only and not for medical use. iPhone 11 are later required based on Apple conducted study of heart rate accuracy August 2026 visit apple.com slash Apple Watch series 12. I see avatar fire and ash is now streaming on Disney Plus. It's the film critics are calling the best avatar yet. A true epic and completely jaw drawback. This is the only pure thing in this world. Return to Pandora on Disney Plus. It will be an adventure for the whole fam and watch the Oscar winning phenomenon at home. This is sick. Avatar fire and ash now stream on Disney Plus. Ready to pee to their teen. This is US Banks Cozy Cash Formations with your chief financial playmaker for Nando Mendoza.

You are worthy of financial security budgeting is now restriction it's freedom. You are smart. You are capable. You are giving your money a purpose. You are learning the language of saving, of growing, now streaming. Remember FDIC copyright 2026 US Bank.

More episodes

More from Tina Huang

View all episodes →