A large language model, based off Meta's open-source Llama 3.1-8B-Instruct, which uses a corpus of various religious texts to generate parables across different cultural traditions.
A project for Brian Greene's Origins and Meaning course, taken Fall 2025.
Uses RAG (Retrival-augmented generation) to enrich the prompt. We draw from the following four texts, sourced from https://github.com/Traves-Theberge/sacred-scriptures-mcp.
- Bible - Christianity
- Dhammapada - Buddhism
- Qur'an - Islam
- Tao Te Ching - Taoism
A detailed explanation and walkthrough of my data preparation and methods are detailed in usage.ipynb.
To get the program working on your local machine, you will need Ollama installed on your computer alongside an installation of Llama 3.1-8B-Instruct or another Llama of choice.
Then, simply run:
python parableGPT.pyto start the session.
