(Retrieval & Vision) Augmented Language Models

Large language models, on their own, have important limitations. I will discuss the trend of "augmenting" language models in various ways to overcome those limitations. Specifically, I will focus on two kinds of augmentation: retrieval augmented generation (RAG) and associated approaches; and vision-augmented language models enhanced to see (LENS).

Speaker: Douwe Kiela, Stanford University

See weblink for instructions to gain entry to the building.

Room 126

This talk was originally scheduled for January 08, 2024