Description: Lecture outlining some of the capabilities of LLMs when evaluating them on various NLP tasks. LLMs do shockingly well at a wide range of these tasks, including exceeding state-of-the-art performance in translation, language modelling, and question answering. This lecture gives an overview of tasks in NLP (independent of large language models), a sense of how well GPT-3 works, and a taste for the art of prompt engineering.
Authors: Percy Liang, Tatsunori Hashimoto, Christopher Re
Date Created: 1/22/2023
Last Updated: 2/22/2024
# Flashcards: 22
Tags: Machine Learning