A Knowledge Processing Framework using Foundation Models based on RAG

[https://arxiv.org/abs/2401.00544]

This is an example code to be included as supplementary material to the following article: "A Reliable Knowledge Processing Framework for Combustion Science using Foundation Models". 2024

Authors: - Vansh Sharma, Venkat Raman

Affiliation: - APCL Group - Department of Aerospace Engineering, University of Michigan, Ann Arbor

Steps:

1. Install the packages using pip - the list is provided. It is advised to create a separate python environment for working with the code.

   Venv steps (in VS Code):

   a. Create environment using: Cmd + Shift + P and then Create Environment (venv)

   b. Activate venv using: source .venv/bin/activate

   c. Install packages using: pip install -r "requirements.txt"

2. Download the LLM model from link provided: https://huggingface.co/TheBloke/Llama-2-7B-Chat-GGML/tree/main For this work you can use - llama-2-7b-chat.ggmlv3.q8_0

3. Use the correct document-data path in create_database.py file and run the code.

4. Once the database is created, update the database path in file - framework_demo.py and run the code for Q&A.

Note - the correct path to LLM model needs to be provided in framework_demo.py

Please cite this work as : @misc{sharma2024reliable, title={A Reliable Knowledge Processing Framework for Combustion Science using Foundation Models}, author={Vansh Sharma and Venkat Raman}, year={2024}, eprint={2401.00544}, archivePrefix={arXiv}, primaryClass={cs.AI} }

Additional notes on Python version and Package List

1. Code runs with Python 3.11.9 with venv
2. Omitted packages for compatibility:
   • chromaviz==0.0.4
   • gmsh==4.12.0.dev1
   • ray==2.4.0
   • urllib==31.26.16
3. Modified packages for compatibility:
   • grpcio==1.49.1 --> grpcio==1.56.0