3 ms·
I think RAG is out of favor because models have a much larger context these days, so the loss of information density from vectorization isn't worth it, and does
by prophesi 1y ago
I think RAG is out of favor because models have a much larger context these days, so the loss of information density from vectorization isn't worth it, and doesn't fetch the information surrounding what's retrieved.
- simonw 1y agoThat's true if you use RAG to mean "extra context found via vector search". I think vector search has shown to be a whole lot more expensive than regular FTS or even grep, so these days a search tool for the model which uses FTS or grep/rg or vectors or a combination of those is the way to go.