A quick demo of a RAG pipeline that answers questions about a book, using excerpts from the book as it's only source of knowledge. Uses Cohere as the underlying LLM and Embedding Vector provider.

I used Romeo and Juliet from Project Gutenberg as an example epub, you can find and download that here: https://www.gutenberg.org/ebooks/1513.epub.noimages

1. Install pyenv (brew install pyenv on OSX)
2. Run ./deps.sh
3. Create a new file named .env and add a variable with your Cohere API key COHERE_API_KEY
4. Run python main.py and follow the prompts!

This was a pretty quick and rudimentary take at building this pipeline, I would like to add better metdata and chapter name parsing as it's really dependent on the name of the internal HTML files right now