liveBook · Manning

FAISS

As explained in LLM Customization and Fine-Tuning

A vector search library used here as an in-process index implementation and as a backend for IVF.

FAISS is an open-source index library from Meta that runs inside your process. It is one option between a flat NumPy array and an operational vector database. section 4.5.3

Choose the index layer to fit corpus scale and deployment needs:

  • Use a flat NumPy array for a prototype with tens of thousands of chunks; it searches exhaustively but scales linearly. section 4.5.3
  • Consider an in-process library such as FAISS for hundreds of thousands of chunks. Its IVF index is suited to corpora sharded across many machines. section 4.5.3
  • Use a vector database when multi-tenant production needs millions of chunks and features such as audit logging. section 4.5.3

How LLM Customization and Fine-Tuning covers it

How deep it goes
Identifies FAISS and IVF use cases but does not detail their implementation or configuration.
Examples
Compares NumPy for tens of thousands of chunks, FAISS for hundreds of thousands, and vector databases for millions.
What you can do after
Evaluate index-layer choices by corpus scale, sharding, and operational requirements.
How it connects
Connects FAISS with vector indexes, IVF, approximate nearest-neighbor search, and vector databases.

Also explained in

Compare how 7 books explain FAISS →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of LLM Customization and Fine-Tuning. The section links point to the parts of the book it draws on.