Retrieval & RAGintermediate12 min · 9 steps
Embedding Models
How text becomes a vector, why you rarely need all of its dimensions, and how to pick and upgrade the model behind your index.
dims1,536
per vector6,144B
datarequest
metrics are illustrativeKey ideaStep 1 of 9
An embedder turns text into a point in space
An embedding model reads a chunk and outputs a fixed-length vector — often 1,536 numbers. It is trained so that texts with similar meaning land close together.
Standard embedders are bi-encoders: one vector per chunk, computed once at indexing time. At query time you embed the question the same way and look for its neighbours.
Lesson 3 of 37 in Retrieval & RAG