README.md

Multi-Document Agent System (MDAS)

Overview

MDAS leverages a distributed agent-based architecture to enhance document processing through a smart partitioning of textual documents. By encapsulating the semantic meaning of document partitions in semi-autonomous agents, it offers parallel querying and reasoning across a comprehensive knowledge space.

System Architecture

Detailed documentation for the system architecture can be found in the following sections:

• Document Agents
• Qdrant Vector Database
• Vector Embeddings

Query Processing

Explore how MDAS processes queries through these detailed documents:

• Master Agent
• Query Planner
• Reranking Module

Optimization Techniques

Understand the optimization techniques used in MDAS:

• Bootstrapped Few-Shot Learning

Getting Started

To set up the MDAS:

git clone https://github.com/jmanhype/DSPy-Multi-Document-Agents.git
cd DSPy-Multi-Document-Agents
pip install -r requirements.txt

Usage

Run the system with:

python main.py

Contributing

Contributions are welcome! Please fork the project, make your changes, and submit a pull request.

License

This project is licensed under the MIT License - see the LICENSE file for details.