How to build a retrieval-augmented generation (RAG) AI system in Java (Spring Boot + LangChain4j)

[@mustiosrs]

Yo I have a question: You've specified some arbitrary embedding model to convert the text chunks into vectors to be stored into the database. Are these vectors also sent to GPT-4 as context? Doesn't GPT-4 only accept their proprietary embedding models such as ada-002 and their new v3 models?

[@vladislavknyshov3482]

Hey, if I got it right, embeddings are only used to find document chunks similar to the question. Then those chunks provided to the model as text context

[@mustiosrs]

@@vladislavknyshov3482 Thanks, I misinterpreted the API of OpenAI, apparentlt the embedding model is completely detached from the LLM AI

[@MiguelAngel-ug3bh]

"and it all of a sudden knows everything about Vaadin, and can put me out of a job. So I don`t know why I created that" 🤣

[@HadraouiZakaria]

Resource code plz