|
AI is everywhere, from coding assistants to chatbots, but what's really happening under the hood? It often feels like a "black box," but it doesn't have to be.
In this episode, Allen sits down with Manning author and AI expert Emmanuel Maggiori to demystify the core concepts behind Large Language Models (LLMs). Emmanuel, author of "The AI Pocket Book," breaks down the entire pipeline - from the moment you type a prompt to the second you get a response. He explains complex topics like tokens, embeddings, context windows, and the controversial training methods that make these powerful tools possible.
IN THIS EPISODE
00:00 - Welcome & Why "The AI Pocket Book" is a Must-Read 15:20 - The Basic LLM Pipeline Explained 8:05 - What Are Tokens? 21:30 - Understanding the Context Window 25:50 - How Embeddings Represent Meaning 35:45 - Controlling Creativity with Temperature 39:30 - How LLMs Learn From Internet Data 45:25 - Fine-Tuning with Human Feedback (RLHF) 51:15 - Why AI Hallucinates 56:45 - When Not to Use
|