This repository showcases some of the potential ways adding context by retrieving relevant parts of documents may benefit the applicability of LLMs within the financial sector where recent information is highly relevant.
Folder structure (updated on 2024-06-10)
- data - data used to add recent information; Limited preprocessing has be done to make the information fit for LLMs.
- data_preprocessing - files used for preprocessing the raw data.
- graphrag_test_case - experiment done with graphrag
- vector_stores - multiple vector stores using different embedding techniques
The main code implementation can be found in RAG.ipynb