✳flâneur — a map of the web's best reading

Rerankers and Two-Stage Retrieval | Pinecone

pinecone.io · 2,401 words · saved by 2 readers

Retrieval Augmented Generation (RAG) is an overloaded term. It promises the world, but after developing a RAG pipeline, there are many of us left wondering why it doesn't work as well as we had expected. As with most tools, RAG is easy to use but hard to master. The truth is that there is more to RAG than putting documents into a vector DB and adding an LLM on top. That can work, but it won't always. This ebook aims to tell you what to do when out-of-the-box RAG doesn't work. In this first chapter, we'll look at what is often the easiest and fastest to implement solution for suboptimal RAG pipelines — we'll be learning about rerankers. Before jumping into the solution, let's talk about the problem. With RAG, we are performing a semantic search across many text documents — these could be tens of thousands up to tens of billions of documents. To ensure fast search times at scale, we typically use vector search — that is, we transform our text into vectors, place them all into a vector sp

Rerankers and Two-Stage Retrieval Jump to section Recall vs. Context Windows Power of Rerankers Implementing Two-Stage Retrieval with Reranking References Retrieval Augmented Generation (RAG) is an overloaded term. It promises the world, but after developing a RAG pipeline, there are many of us left wondering why it doesn't work as well as we had expected. As with most tools, RAG is easy to use but hard to master. The truth is that there is more to RAG than putting documents into a vector DB and adding an LLM on top. That can work , but it won't always. This ebook aims to tell you what to do w

Explore this link on the map →

saved by

related reading