RAG Retriever : Exact vs. Compressed Index?

lhoestq · November 9, 2020, 1:42pm

Hi ! exact vs compressed refers to the quantization used for the FAISS index. The compressed one uses an IVF index with product quantization and requires significantly less RAM than the exact one. To reproduce the RAG papers result you will need the exact one though.
Note that I will update this week the parameters of both index so that the exact one uses the same as RAG’s paper, and also to have an optimized compressed one.

Topic		Replies	Views
Using RAG with local documents Models	3	2655	April 21, 2021
RAG Model performance does not match paper Models	0	284	February 5, 2021
RAG Model for QA Models	1	1067	December 30, 2023
RAG Example and Word-Level contributions Models	4	1585	October 12, 2020
Vector DB - Exhaustive search in RAG Intermediate	0	190	November 14, 2023

RAG Retriever : Exact vs. Compressed Index?

Related Topics