NVIDIA Generative AI

Cloudadorn Academy
  • 0.0 (0)
  • Technology
  • Complete

Every vendor wants to sell Ana a model. Ana runs Kestrel, a company that sells shelf-scanning cameras to warehouses, and both she and Kestrel are made up. She isn't technical, so she asks her friend Shelley what to believe, and Shelley brings each brochure, quote and strange email to Rob, who explains the machinery underneath. Over twenty-three episodes they open the nesting dolls of generative AI, follow a sentence through attention, watch a model learn and fail, and climb the ladder of ways to make a model yours, from asking differently to RAG and LoRA. Then the model gets eyes and ears, through CLIP, diffusion, a hospital scan, a car braking in the drizzle and a demo that describes a crushed box. The last stretch walks NVIDIA's stack: the chips and CUDA, TensorRT, serving with Dynamo, NeMo, agents and the price of thinking longer, guardrails, robots trained in warehouses that don't exist, why NVIDIA is hard to leave, and how to check whether a product was really replaced. Where the story lives elsewhere, other invented people step in, like Joan and Denise. No technical background needed. More at cloudadorn.com.

  1. Episode 1

    S1E01. Nesting Dolls: What Generative AI Actually Contains

    Three brochures land on Ana's desk after a trade show. One promises AI powered shelf counting, one says machine learning inside, and one says generative AI, built on a foundation model, in gold letters. Ana and her company Kestrel, which sells cameras that count what's on warehouse shelves, were made up. Rob pictures the words as painted Russian dolls, artificial intelligence outside, then machine learning, deep learning and generative AI, where smaller means more specific, not better. You'll understand why a spam filter sorts while a generative model makes, why a foundation model isn't one of the dolls, using CLIP and NVIDIA's Cosmos, and how self-supervised learning lets text be its own answer key. Along the way come ImageNet in 2012 and the 2017 paper Attention Is All You Need. Next, a forklift manual that leaves a chat assistant thinking for ages.

  2. Episode 2

    S1E02. Attention: How a Model Reads a Sentence

    With one page of a forklift manual pasted in, Ana's brake question came back almost at once. With the whole manual, she had time to make tea. Ana and Kestrel, her warehouse camera company, are made up. Rob follows her question from the moment she presses enter: text cut into pieces smaller than words, each placed on a map of how words get used and stamped with its position, because the machinery sees a heap, not a line. He pictures attention as a group chat where every piece listens to every other, which is why double the length means four times the comparing. You'll also hear how NVIDIA's Nemotron 3 Nano mixes in cheaper blocks, and why one deck saying thirty-one billion and another saying three billion can both be honest. Coming up, an overnight message from a contractor whose training run went wrong.

  3. Episode 3

    S1E03. How a Model Learns: Forward Pass, Loss, and Backpropagation

    Error went up overnight, run is ruined, reads the message Ana's contractor sent at six in the morning. Ana and her camera company Kestrel are made up, and the contractor is teaching the cameras to spot dented boxes. Rob explains what training is: one loop of guessing, boiling the wrongness down to a single number called the loss, and handing blame backwards through the layers, which is backpropagation. Expect to learn why confidently wrong costs more than hesitant, and to hear a whispered message that gets quieter at every handover, plus the fixes that let networks a hundred and fifty two layers deep learn. Then the knobs a person sets before training starts, above all the learning rate and warmup. Next, a dent checker that was right nearly every time on its test photos and still waved through a pallet of crushed boxes.

  4. Episode 4

    S1E04. When Training Goes Wrong: Overfitting, Bad Data, and Metrics That Lie

    On day two at a client's warehouse, a new checker built to spot dented boxes waved through an entire pallet of crushed ones, even though its report said it was right nearly every time. Ana and her shelf-camera company Kestrel were made up. Rob explains why accuracy hides the failure you care about when the thing you're hunting is rare, and how precision and recall point at false alarms and misses. A pair of learner drivers, one who only knows the road to work and one who quit after a lesson, separate overfitting from underfitting, and a locked drawer of photos tells them apart. You'll also hear about early stopping, a yellow damage sticker that may have given the answer away, and a 2021 study led by Curtis Northcutt that found wrong labels in ten standard test collections. Up next, a cat asleep on a warehouse shelf.

  5. Episode 5

    S1E05. The Architecture Zoo: CNN, RNN, Transformer, GAN, VAE, Diffusion

    A cat is asleep on a warehouse shelf, and the shelf camera reported everything fine, because it was only ever taught to look for dented boxes. The camera is one of Kestrel's, the company Ana runs, and both she and Kestrel are imaginary. Rob treats the model names two vendors pitch as shapes of machine, each built for one kind of problem. Hear how a CNN slides a tiny window across a photo, how an autoencoder trained only on normal shelves finds the odd one out by rebuilding it badly, and why latent space is the recipe rather than the cake. For fake dented boxes, he weighs a GAN, including NVIDIA's StyleGAN and the risk of mode collapse, against diffusion, citing the 2021 paper Diffusion Models Beat GANs and a 2020 measurement by Song, Meng and Ermon. Next, a vendor's quote with a lot of zeros.

  6. Episode 6

    S1E06. The Cheapest Thing That Works: Prompting Before Fine-Tuning

    A vendor has quoted Ana months of work and a number with a lot of zeros to fine-tune a model on every document Kestrel has ever written. Ana and Kestrel, her warehouse camera firm, are as imaginary as each other, but her three wishes for a support assistant are ordinary: answer questions about a returns policy that keeps changing, write in Kestrel's friendly, dry voice, and fill in a ticket form exactly. Rob lays out a ladder of ways to make a general model do your job, in price order, and says climb only when the rung below has failed. You'll understand why the form needs a stencil, why RAG suits a policy that moves, where LoRA earns its place for the voice, and what full fine-tuning and pretraining are for. Temperature isn't training at all. Coming up, an expensive retrain that can no longer write a simple list.

  7. Episode 7

    S1E07. Teaching an Old Model New Tricks: Transfer Learning, LoRA, and RLHF

    Please complete section four of the damage report form. That was the whole reply Kestrel's support assistant gave when a warehouse manager asked for a short list for his night shift. Ana and her camera company are made up, and a vendor had just retrained an entire model on everything Kestrel ever wrote. Rob uses hiring for a café to explain why starting from a lab's model makes sense, then decodes the delivery note: fine tuning, RLHF learning from people's clicks, as in InstructGPT, and a simpler method that skips the judge. Learn about catastrophic forgetting, why whole-model retraining needs so much memory, and how LoRA leaves the base model frozen and puts the change in a small file. A study called Fine-Tuning or Retrieval explains why the catalogue still came out wrong. Next, an assistant that looks things up and still gets the freezer aisle wrong.

  8. Episode 8

    S1E08. Open Book: RAG Done Honestly

    Asked whether the bracket warranty covers a camera in the freezer aisle, Kestrel's support assistant said fully covered, quoting the manual. One line further down, the manual says it doesn't apply in freezer aisles. Ana and Kestrel are imaginary, and her team wants a bigger model. Rob says hold the money, because retrieval augmented generation, or RAG, is two machines bolted together, one that fetches pages and one that writes. You'll understand embeddings as addresses for meaning, how chunking can cut an exception off its rule, and the silent trap of mixing embedding models. The papers Fine-Tuning or Retrieval and Lost in the Middle show why looking things up reduces hallucination without ending it, and two checks, faithfulness and whether the right pieces came back near the top, point at which machine failed. Up next, a comparison sheet with a red circle round the wrong vendor.

See All Season 1 (23)

About

Every vendor wants to sell Ana a model. Ana runs Kestrel, a company that sells shelf-scanning cameras to warehouses, and both she and Kestrel are made up. She isn't technical, so she asks her friend Shelley what to believe, and Shelley brings each brochure, quote and strange email to Rob, who explains the machinery underneath. Over twenty-three episodes they open the nesting dolls of generative AI, follow a sentence through attention, watch a model learn and fail, and climb the ladder of ways to make a model yours, from asking differently to RAG and LoRA. Then the model gets eyes and ears, through CLIP, diffusion, a hospital scan, a car braking in the drizzle and a demo that describes a crushed box. The last stretch walks NVIDIA's stack: the chips and CUDA, TensorRT, serving with Dynamo, NeMo, agents and the price of thinking longer, guardrails, robots trained in warehouses that don't exist, why NVIDIA is hard to leave, and how to check whether a product was really replaced. Where the story lives elsewhere, other invented people step in, like Joan and Denise. No technical background needed. More at cloudadorn.com.

Information

  • Creator
    Cloudadorn Academy
  • Episodes
    23
  • Seasons
    1
  • Rating
    Clean
  • Copyright
    © Cloudadorn Academy
  • Show Website
    NVIDIA Generative AI