Language Models and Next‑Word PredictionCornell Course

In this course, you will use Python to quantify the next-word predictions of large language models (LLMs) and understand how these models assign probabilities to text. You'll compare raw scores from LLMs, transform them into probabilities, and explore uncertainty measures like entropy. You'll also build n-gram language models, handle unseen words, and interpret log probabilities to avoid numerical underflow.

By the end of this course, you will be able to evaluate entire sentences for their likelihood, implement your own model confidence checks, and decide when and how to suggest completions for real-world text applications.

You are required to have completed the following course or have equivalent experience before taking this course:

• LLM Tools, Platforms, and Prompts