Retrieval Augmented Generation (RAG) using a FAISS vector database, LangChain and OpenAI GPT-4 for inference.
Check the generic instructions for requirements at parent README.md.
It loads data/scalexi.txt information file, split it in chunks, send them to OpenAIEmbeddings to convert them into embeddings, and stores it in FAISS.
Then, calls ConversationalRetrievalChain.from_llm to answer a prompt, giving it access to retrieve information from the vector database.
Run it with:
make