martin-fabbri/graph-llm-agents

A hub for exploring the integration of graph-based orchestration with LLMs, focusing on the development and experimentation of multi-actor LLM applications.

Graph LLM Agents

Overview

This is a hub for exploring the integration of graph-based orchestration with LLMs. This repository is a collection of experiments, notebooks, and projects aimed at innovating and understanding the capabilities and applications of multi-actor LLM systems. It serves as a space for documenting a journey of discovery and innovation within the AI Agents domain.

Objective

The primary goal of this playground is to explore the dynamics of coordinating LLMs using graph-based structures, enabling sophisticated workflows and interactions among multiple computational actors.

Features

• Experiment Notebooks
• Project Implementations: Complete project directories showcasing applications of graph-LLM integration, from concept to execution.
• Documentation & Guides: In-depth explanations of the experiments and projects, including setup instructions, methodologies, and findings.
• Resources & References: Curated list of resources, articles, and papers that provide additional context and support for the experiments conducted.

WIP

• LangGraph Exploration
• LLM Powered Autonomous Agents

Backlog

• Adversarial Attacks on LLMs
• Efficient Tool Use with Chain-of-Abstraction Reasoning

Setup Virtual Environment with Poetry

Step 1: Installing Poetry

pip install poetry

Step 2: Configuring Poetry

Set Poetry to create virtual environments within your project directory:

poetry config virtualenvs.in-project true

Step 3: Initialize the Project

poetry init

Step 4: Install Project

poetry install