Building a Financial RAG Engine with LLaMA

This repository implements a Retrieval Augmented Generation (RAG) engine named LlamaRAG, specifically designed for tasks in the financial domain. It leverages the power of the LLaMA 270B model, making it a valuable tool for open-source large language model development.

Here's a breakdown of what you'll find:

• Code demonstrating how LLaMA can be used for financial question answering tasks using RAG techniques.
• Integration with libraries like Sentence Transformer and Streamlit for efficient document embedding and a user-friendly interface.
• A Streamlit web application showcasing functionalities like summarizing financial performance, extracting entities, and analyzing sentiment.

Important Note: Due to the computationally intensive nature of large language models, I strongly recommend running this project in a GPU-accelerated environment.

This project can serve as a valuable resource for developers interested in exploring LLaMA and its applications in financial analysis and other open-source settings.

Startup

1. Clone this repo git clone https://github.com/usamabuttar/LlamaRAG
2. Go into the directory cd LlamaRAG
3. Startup jupyter by running jupyter lab in a terminal or command prompt
4. Update the auth_token variable in the notebook.
5. Hit Ctrl + Enter to run through the notebook!
6. If you want to start up the streamlit app run streamlit run app.py (make sure you update your auth token in there as well!)

Other References

-Llama 2 70b Chat Model Card:hugging face model card on the model used for the video.

-Llama Index Doco:sick library used for RAG.

Who, When, Why?

👨🏾‍💻 Author: Usama Buttar
📅 Version: 1.x
📜 License: This project is licensed under the MIT license. Feel free to use it, just don't do bad things with it.