How do LLMs work and where does the knowledge live?
This presentation explores how large language models (LLMs) store knowledge and what happens during both training and inference (interaction with the model). You'll learn about the structures used to persist knowledge, both in a human-like and a very abstract way. By the end, you'll have a solid understanding of how LLMs work, along with some of the key techniques that make them effective. The goal is to demystify the technology and lift the veil on the 'magic' often associated with LLMs.
