Retrieval-Augmented Generation(RAG) framework to generate answers from local knowledge base using Open source LLM (zephyr-7b-beta/ Mistral-7B-Instruct-v0.2)

Blog Post here.

Naive RAG

Vector Store: FAISS
Local Knowledge Base: Federal Open Market Committee (FOMC) meeting documents for the years 2020-2023
LLM : HuggingFaceH4/zephyr-7b-beta & Mistral-7B-Instruct-v0.2
Local documents in pdf

NoteBooks:
- 1a: Creating a Vector Store database (FAISS) from pdf documents (local knowledge base)
- 1b: RAG implementation using transformers pipeline
- 1c: RAG implementetion using langchain llm chain with ReRanker
- 1d : Conversational RAG implementation

2a : Graph RAG implemention using Neo4j Graph and Open source LLM(Mistral-7B-Instruct-v0.2)

Colab Notebooks.