Embedding Model
A specialized type of AI that converts text into numerical vectors for efficient semantic search.
What is an embedding model?
An embedding model is a specialized type of artificial intelligence that serves as a translator between human speech and the world of numbers. Its task is to take a piece of text (a word, sentence, or paragraph) and calculate its vector - a numerical representation of meaning.
How embedding models work
Modern embedding models operate in a space with hundreds to thousands of dimensions. For example, the OpenAI model (text-embedding-3-small) uses 1,536 dimensions. Each section of text is assigned 1,536 coordinates that precisely define its meaning.
Use in RAG
- When storing, documents are converted to vectors in a vector database
- When querying, the user's question is converted to a vector for comparison with stored data
- Enables semantic search - searching by meaning, not just keywords