Slide deck: drive.google.com/file/d/1DGXb...
One year after the release of GPT-4, large language models (LLMs) remain the most exciting topic in AI. While much about their qualitative capabilities remain poorly understood, there are some areas where we can quantitatively measure, bound, and forecast their behavior. This tutorial will introduce the topic of LLMs through 5 key formulas in their study. The goal is to provide a survey of the area by focusing on diverse efforts to formalize the remarkable nature of observed phenomena.
Instructor:
Sasha Rush, Associate Professor, Cornell Tech, Cornell University