Épisodes

  • Fine-Tune Llama 3 706B Model Locally

    Fine-Tune Llama 3 706B Model Locally
    Jun 15 2026
    The hum of a server. The click of a terminal. Forget the API; this is about running the colossal Llama 3 model yourself. A 706B parameter AI, in-house, offline. Host Nick Creighton maps the frontier of true data privacy, where your internal documents and codebases never leave your control. This is the blueprint for building a private AI data center, covering the hardware reality check and the parallel power of fine-tuning and RAG. Ready to ship it? Listen here. Read the companion post: [Link to blog post]
    Afficher plus Afficher moins
    17 min
  • Fine Tune Llama 3 For Sql Query Generation Tutorial

    Fine Tune Llama 3 For Sql Query Generation Tutorial
    Jun 15 2026
    You trust the AI to write your SQL. It's going to cost you. This week, we fine-tune Llama 3 in an afternoon for under fifty dollars, turning a brittle liability into a precise, private tool. It’s about the gap between a convincing answer and a correct one. The quiet panic when a hallucinated column sends your dashboard off a cliff. We’re moving past generic chatbots and into something that understands your schema, your logic, your stakes. Build the detail-oriented, credible alternative. Full walkthrough in the companion post. The next query it writes could save your morning. Listen.
    Afficher plus Afficher moins
    15 min
  • Llm Evaluation Metrics Explained 2024

    Llm Evaluation Metrics Explained 2024
    Jun 8 2026
    Build Log, with Nick Creighton. This week, the models went quiet. The outputs, once reliable, turned bland and hollow. When your systems falter and hope is your only strategy, it’s time to move past the demo. Nick recounts the death of the "vibe check"—that quick, gut-feeling review that fails when you’re not looking. He spent the last three months building a real validation pipeline, shifting from fragile prompts to a system that actually earns its keep. This is about fighting the silent decay of AI performance, about replacing theory with a foundation that holds while you sleep. For more detail on the validation build, find the companion post [link]. Listen to the full episode.
    Afficher plus Afficher moins
    9 min
  • Openai Api Vs Local Llama 3 Cost 2024

    Openai Api Vs Local Llama 3 Cost 2024
    Jun 8 2026
    Signal Notes. March 25th, 2024. A cold number on the dashboard at dawn. The hum of a server, the quiet click of a key. The cost of intelligence is plummeting, a 92% freefall in 14 months. The gap between cloud and local inference has narrowed to a sliver, a decimal point on a billing report. The raw data from Nick’s production run: $347.22 for the API, $412.00 for the rented hardware, plus the hidden tax of maintenance scripts and library conflicts. It’s a story told in tokens and receipts, not theory. A vibe of pragmatic calculation. The quiet awe of a shifting landscape. Read the numbers: [companion blog post link] Listen to the math.
    Afficher plus Afficher moins
    25 min
  • Ai Agent Frameworks Vs Traditional Automation 2024

    Ai Agent Frameworks Vs Traditional Automation 2024
    Jun 5 2026
    The old map is obsolete. It's being replaced by a compass. Traditional automation is a brittle, minimum-wage workforce. AI agents are something else entirely—navigators that work off-road. Nick put both systems to the test over three months across thirteen sites. The results weren't close. It’s a fundamental architectural shift happening right now in production, moving from rigid step-by-step processes to adaptive, goal-oriented execution. This isn't a future prediction. It's a present-tense reality measured in hours saved and systems that don't break. Dive into the data and the details in our companion post. Grab your headphones. Let's build.
    Afficher plus Afficher moins
    9 min
  • Rag Vs Fine-Tuning For Document Qa 2024

    Rag Vs Fine-Tuning For Document Qa 2024
    Jun 5 2026
    Build Log. I’m Nick. If you’re using a fine-tuned model for document Q&A, you’re likely burning cash for worse results. This is the critical build-vs-buy decision for your AI’s brain, and it’s a weekly invoice that decides if your project lives or dies. GPU costs are falling, but fine-tuning API prices haven’t. The real killer? Knowledge cutoffs. A perfectly formatted, completely wrong answer from a model trained on last year’s docs. RAG solves this inherently. New docs hit the vector store, and seconds later, your AI knows. No retraining. No extra cost. A three-month production test. The winner wasn’t close. Read the full breakdown on the blog. Listen to the episode.
    Afficher plus Afficher moins
    9 min
  • Fine-Tuning Transformers Vs Lora Vs Qlora 2024

    Fine-Tuning Transformers Vs Lora Vs Qlora 2024
    Jun 5 2026
    The old guard is out. The headlines make it sound like custom AI needs a bank of supercomputers and a team of PhDs. What if it doesn’t? Build Log, with Nick Creighton. A quiet story of shipping. This week, we move past the hype to the real workbench. The goalposts have moved. We’re talking about fine-tuning that’s faster, cheaper, and shockingly accessible—practical for the rest of us, running in the background of everything we build. Full breakdown: [Link to blog post] See how it fits together. Listen to Build Log.
    Afficher plus Afficher moins
    8 min
  • Local Ai Deployment Cost Analysis 2024

    Local Ai Deployment Cost Analysis 2024
    Jun 5 2026
    Build Log. Nick Creighton. A quiet rebellion against the cloud. The real cost of AI isn't in the API docs—it’s in the monthly bill. Nick just pulled his AI workflow in-house, deploying a private agent for his entire content network. The price tag? Under fifty bucks. This is about taking back control. It’s the hum of a local server, not the silent drain of a metered service. A breakdown of the hardware, the models, and the math they don’t want you to see. The vibe is autonomy. For the full cost analysis, see the companion post. Listen to the quiet hum of your own machine.
    Afficher plus Afficher moins
    8 min