README.md

AlchemyBot: GPT-4 + Stable Diffusion powered Homestuck Alchemy Discord chatbot

What is this?

A Discord bot that allows users to generate text + image of combinations of two items like in the webcomic Homestuck, using LLM and diffusion AI technologies.

How to use

1. Create a virtual environment and install the requirements:

python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt

2. Create a Discord bot and get its token. See this guide for more information.
3. Create a .env file from .env.example and fill in the Discord and OpenAI tokens.
4. If you want image generation, download AUTOMATIC1111's Stable Diffusion UI and keep it up at the default port. You might need to install the RemBG extension, and the models linked in the blog post. If you don't want image generation, you might need to lightly edit the code in alchemy_bot.py.
5. Run the bot:

python3 bot.py

6. Invite the bot to your server and launch it with D--> alchemize ITEM1 && ITEM2. alchemize3 uses GPT-3.5 and alchemize4 uses GPT-4.

Learn more in the blog post.

Credits

• Clue, Tensei and the Homestuck Discord for varied help.
• The good parts of Homestuck for the inspiration.
• pyrrhicPachiderm for his Alchemy JS library.
• AUTOMATIC1111 for his amazing Stable Diffusion UI.
• OpenAI and their LLM team.
• The creators of Fantassified Icons for their surprisingly good icon model.