Logotyp KTH Logotyp Karolinska Institutet Logotyp Region Stockholm

Mostly Harmless LLMs: Demystifying ChatGPT

AI generated picture illustrating mystified AI

by Dennis Hein (Department of Physics, KTH)

Tuesday Nov 5, 15.00-16.00

Location: Seminar room FA32, Albanova Main building, 3rd floor, Roslagstullsbacken 21

Abstract:
This talk attempts to demystify large language models (LLMs) like ChatGPT by exploring their fundamental training processes and internal workings. We will cover how these models are trained on vast amounts of text, how they process and generate language, and what gives them their seemingly magical capabilities. The presentation concludes with practical strategies for effectively using these tools in real-world applications.

Videoconference: https://kth-se.zoom.us/j/69955116658