README.md

metadata

title: Swords & Wizardry RAG over Rulebook
emoji: 📚
colorFrom: yellow
colorTo: purple
sdk: gradio
sdk_version: 4.26.0
app_file: app.py
pinned: false

Retrieval Augmented Generation for Wizards & Wizardy Rule Sets

Project Overview

Welcome to the Retrieval Augmented Generation (RAG) project for Tabletop Role-Playing Game (TTRPG) rule sets. This project, by Alan Meigs, aims to explore the potential of RAG techniques in parsing and interacting with complex PDF documents, specifically focusing on TTRPG rule sets.

Objectives

• Experiment with RAG: Utilize Retrieval Augmented Generation to enhance the understanding and interaction with TTRPG rule sets.
• PDF Parsing: Develop methods to effectively parse and extract meaningful information from PDF documents.
• Contextual Understanding: Provide comprehensive context to language models to improve the quality of responses related to TTRPG rules.

Key Features - Soon Avaiable on DungeonMind.net

• Page-Aware Chunking: Implement a custom text splitter that retains page information, allowing for precise referencing and context building.
• Enhanced JSON Summaries: Load and utilize document and page-level summaries to provide enriched context for language models.
• Embedding and Retrieval: Use advanced embedding models to convert text chunks into embeddings, facilitating efficient retrieval of relevant information.
• Interactive Chatbot: Deploy a Gradio-based chatbot interface to interact with the parsed rule sets, providing users with accurate and contextually relevant answers.

How to use

• Chat with the Rulebook: Use the chatbot to ask questions about the rulebook.

How to Find Me

• LinkedIn
• GitHub
• DungeonMind.net