Skip to content

tamoghna21/RAG_LLM

Repository files navigation

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

Graph RAG

Notebooks

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

Colab Notebooks.

Releases

No releases published

Packages

No packages published