The project aims to create an application that assists users with knowledge retrievals. The method to do so is by having the LLM be augmented by a knowledge repository via the RAG (Retrieval-Augmented Generation).
Tools used are OpenAI LLM, ChromaDB, LangChain, and StreamLit (frontend)
The goal of the project is to have a Most Viable Product that shows an LLM being able to assist users with finding key information quickly from a store of over 100 PDFs, each having 30-100 pages. During the testing, publicly available graduate mathematics lecture notes were used so the application is not bounded by any knowledge domain.
Short-term and long-term memories are used to augment user queries in the same manner as above. This allows more "targeted" answers, where context is preserved.
