PokémonLLM

Final project for my COSC490: Introduction to Large Language Models course. This project is a Pokémon battle assistant chatbot built using Meta's LLaMA-3.1-8B-Instruct model as its foundation and ran locally with Ollama. To enhance the accuracy and legality of the generated responses, I implemented techniques such as post-generation filtering, Retrieval-Augmented Generation (RAG), and fine-tuning. The assistant includes a user-friendly web interface and can be accessed in the browser via a Flask server.

⚠️ Note: You must have Ollama installed to run the model locally. Ollama is required to load the GGUF quantized model and serve it for the chatbot.

⚠️ Note: You must download the llama-3.1-8b-instruct-q4_k_m.gguf from HuggingFace file and place it in /llama3-gguf. Your file structure should look like

llama3-gguf/
├── llama-3.1-8b-instruct-q4_k_m.gguf
└── Modelfile

UI

Structure

.
├── chatbot_core.py                         # PokemonLLM logic: RAG, post-generation filtering
├── flask_app.py                            # Flask server entry point 
├── pokemon_gen12_moveset.json              # legal moveset data for gen 1 & 2 used in RAG 
├── README.md                                     
├── llama3-gguf                             # llama3 model and Modelfile used to create assistant via Ollama  
│   ├── llama-3.1-8b-instruct-q4_k_m.gguf
│   └── Modelfile
├── static                                  # CSS, images, fonts
│   ├── global.css
│   ├── PixelifySans-VariableFont_wght.ttf
│   ├── pokeball-removebg-preview.png
│   └── pokewalk.gif
└── templates                               # HTML and JS
    └── index.html

Tech Stack

Frontend:

• HTML5
• CSS3

Backend:

• Python 3.10+
• Flask

Model Serving:

• Ollama (LLaMA 3.1 8B Instruct via GGUF)
• llama.cpp / GGUF model integration

LLM Enhancements:

• Post-generation filtering
• Retrieval-Augmented Generation (RAG)
• Fine-tuning (limited scale)

How to Run

1. Create and Activate environment

- python3 -m venv venv
- source venv/bin/activate

2. Install dependencies

- pip install -r requirements.txt

3. Create Assistant Model with modelfile

- cd to /llama3-gguf
- run ollama create pokemon-assistant-v3 -f Modelfile

4. Run Flask server

- Ensure Ollama server is running - Ollama serve &
- python flask_app.py

Authors

• William Lamuth