From the course: Hands-On AI: Introduction to Retrieval-Augmented Generation (RAG)

Unlock this course with a free trial

Join today to access over 24,700 courses taught by industry experts.

Introduction to vector databases

Introduction to vector databases

- [Instructor] Vector databases are used to store, index, and compare vectors. The main usage for vector databases is to store and retrieve the vector embeddings that we talked about in the last video. One of the things that separates vector databases from a traditional SQL database is the search method. Namely, vectors are indexed in a certain way so that we can retrieve them based on some sort of similarity search method. When we input data into a vector database, all vector inputs are assigned an ID. You can, and in most cases automatically do, store metadata in your vector database. For example, for RAG applications, you have to store the text along with the vector, and most frameworks do this automatically. Metadata refers to any data that is not the vector embedding itself or the ID. The basic process for retrieving data from a vector database is simple. The distance between your input vector and the vector stored in the vector database is computed. This distance can be measured…

Contents