A chatbot that answers questions about the game Monopoly using open source LLM from Ollama. The models are downloaded to your desktop and the chatbot can be run without an internet connection. The data is stored in a Vector Data Store using ChromaDB. https://docs.trychroma.com/docs/overview/introduction
This is built off of my Monopoly Chatot: https://github.com/acflippo/Monopoly_Chatbot
The special domain knowledge I'm using is a downloaded play instructions (as this is a RAG implementation for Monopoly) that I've saved as annie_monopoly_vault.txt which I've removed special characters that would make the embedded Markdown text look strange. That's a bit of prepocessing going on here outside of this code base.
Below is a screenshot of the bot input and reply. I'll create a video on the interactions perhaps later. I've tried to trick the bot into discussing other topics but so far so good (see below 👇).
For the latest instructions, go https://docs.streamlit.io/get-started/installation
- In your virtual python environment, run this
pip install streamlit
-
Install Ollama to your machine by following instructions here: https://ollama.com/
-
Download models you'll need
ollama pull mxbai-embed-large
ollama pull dolphin-llama3
- Create the ChromaDB to store your embeddings into a named folder
python create_db.py
- You can test if your ChromaDB is working
python load_db_test.py
Note: You can alternatively use larger models such as Ollama3.3 or DeepSeek-r1 but I've found that those models require a large amount of memory so you might want to move to a cloud server instead of your own personal desktop or laptop.
- Run the app
streamlit run monopoly_rag_chromadb.py
- Have Fun! Try your own special domain knowledge and let me know how it goes. You can contact me at www.datashep.io