Embeddings and vector databases
In Chapter 15, we discussed the importance of embeddings and latent spaces, and I explained how they can be created in different ways. One way is when a generative model learns them intrinsically during its training process, and another is when we use specific types of models to create them explicitly.
I also touched on why we would want to explicitly create them since they can be processed more efficiently and are a more suitable format for ML use cases. In this context, when we create an embedding for something, we are simply creating a vector of numeric values to represent it (how we actually do that is a more advanced topic we will cover shortly).
Another concept I briefly touched upon was the importance of relationships between embeddings in the vector space. For example, the proximity of embeddings to each other in the vector space can reflect the similarity between the concepts they represent. Let’s examine this relationship in more...
