DnD_Chatbot

A simple chatbot that uses retrieval augmented generation to give me personalized spellcasting advice for my Dungeons and Dragons character.

Step 1: Set up environment

Create a conda enviroment from the requirements.txt file with the following command:

conda install --file requirements.txt

Step 2: Download LLaMA model locally

By setting up the LLM locally, we do not have to use an API key to generate completions. Use the bash script to get a small model and launch the server locally. The server will be accessible on your computer at http://127.0.0.1:8080/.

bash SetupLLM_locally.sh

Step 3: Launch UI and interact with SpellBot

Launch the model and UI with the following code:

python RunSpellBot.py

The UI will be designed according to index.html and will be accessible locally at http://127.0.0.1:5000! The code will use the SpellsTable.csv to supplement it's responses (using Retrieval Augmented Generation).

Typically, you will get a response in 10 seconds.

See example conversation below!