GG: A Generative Role Playing Game using LRMs and LightRAG (a fork of GraphRAG)
Augmenting Finetuned Chain of Thought capable Large Reasoning Models with Retrieval Augmented Generation and Knowledge Graphs to drive a role-playing game with infinite narrative choices and track game world developments.
- Data Engineering - Luao Zhang
- Synthetic Data - Jiaheng Yang
- Model Training - Xiao Ying
- RAG System - Srihari
- Streamlit Frontend - Junhan Guan
Built as a group project in part for the required project component in CT5121 - Advanced Natural Language Processing module at the University of Galway.
- Streamlit
- Ollama
- LightRAG
We use the JerichoWorld Dataset as a base and generate Synthetic Data out of it to train our Large Reasoning Model.