Vector embedding is the process of transforming data or tokens such as texts, images, audio, and videos into numerical vectors by embedding models. Each embedding model is specific to the large language model (LLM), therefore it is critical that the correct embedding be used for the appropriate model.

Why is vector embedding important?

Vector embedding is important because it captures the underlying meaning, structure, or feature of the data in a way that can be understood and quickly compared. It is critical for powerful semantic searches, recommendations, clustering, and reasoning across multiple filetypes.