Implementation of Question Answers model through RAG

This project uses OpenAI's embedding vector for storing data through API call. It required OPENAI_API_KEY to be set in environment variable. It uses RAG with the combination of LangChain for retrival of embedding vector from vector database.

Folder structure

notebooks:

• It consists of jupyter notebook
• Different test been tried w.r.t to Generative AI models (like flan, qa, bert etc)

resources:

• It consists of extra resources (like templates; sample questions)
• It also consists of chromaDB on-disk files

configs:

• configuration files

main.py:

• It's a main function call

Technology Stack

• fastAPI been used as a microservice
• Langchain for building pipeline across sections of generative AI model
• ChromaDB for vector database storage
• OpenAI for API call
• RAG for retrieval
• FLAN-T5 for fine-tuning model
• BERT for fine-tuning model (roberta flavours)