An LLM-Augmented Dialogue System
Dialogue systems in games have been around since some of the earliest implementations of video games, all the way back to text-based adventures. They have progressively become more intimate and personalized as developers have sought more natural and tailored conversations. However, even the most intricate dialogue systems require large teams of game designers to write millions of lines of dialogue [citation needed] to cover every imaginable situation, which is inherently an impossible task. Inevitably, corners must be cut. With LLMs dominating the most recent advancements in AI, it is natural to consider methods of augmenting dialogue systems by incorporating them somewhere in the pipeline. In our project, we plan to research existing implementations of LLM-augmented dialogue systems and use the knowledge we uncover to implement one of our own.
This project is for CSC-570: Critical Creative Computing at Cal Poly
- Nathan McCutchen (@N8WM)
- Mihika Nigam (@MihikaNigam)
- A running LLM studio with
lmstudio-community/Meta-Llama-3-8B-Instruct-GGUFdeployed. - Python 3.x installed.
-
Clone the repository:
git clone https://github.com/YourUsername/OpenNPC.git cd OpenNPC/src -
Install the required packages:
pip install -r requirements.txt
-
Run the main application:
python3 main.py