06-02, 10:00–10:50 (Europe/Brussels), Kleine Spoel (0.31)
While most people have played with LLMs, a lot less is known about how they work. This talk takes a look just below the surface: not a full dive into the math, but a look at the typical architecture and the processing steps. I will also demonstrate how you can run a model locally using llama.cpp.
Software engineer interested in Python, Linux, embedded systems and retro computing.