Awesome AI Agents › Generative Agents Simulation
This project is a working, locally runnable, and low-cost implementation of Generative Agents, which are interactive simulacra of human behavior based on generative large language models. It is inspired by the research paper "Generative Agents: Interactive Simulacra of Human Behavior" and aims to simulate human-like behavior in a virtual environment. The repository provides models distributed as Jupyter notebooks that can be run locally or on Google Colab, with a recommendation to use Jupyter Lab for local execution. The models are designed to work best with 16GB VRAM but also support environments with less than 8GB VRAM, making them accessible for personal computers and edge devices. The current implementation simulates the town of Phandalin from an introductory Dungeons & Dragons 5th Edition adventure, offering a more free-form and complex setting than the original paper's simple scenario. The project includes stable models available in the Release folder and work-in-progress models with the latest features in the WIP folder. Additionally, there is a WIP library for game simulation purposes. One of the main challenges addressed by this project is the high cost and resource demand of running high-quality instruction models like GPT-3, which are typically required for such generative agents. Instead, this implementation uses low-parameter, locally runnable models to reduce costs and increase accessibility. The project anticipates improvements with the next generation of instruction-tuned models. Future development plans include summarizing agent decisions using emojis, creating a family of questions to better compress agent contexts, and exploring additional prompt layers to enhance context compression. Overall, this project provides a practical and accessible approach to simulating human-like behavior using generative language models in interactive environments.
https://github.com/mkturkcan/generative-agents
This project provides a simulation framework for generative agents that mimic believable human behaviors within an interactive game environment, enabling running, saving, replaying, and demonstrating agent-based simulations.
Voyager is an open-ended embodied agent powered by large language models that autonomously explores and learns diverse skills in Minecraft through lifelong learning and an evolving skill library.
SFighterAI is a deep reinforcement learning-based AI agent designed to master and beat the final boss in Street Fighter II: Special Champion Edition using only game screen pixel data.
A comprehensive Godot framework for creating smooth and complex AI motion using steering behaviors in both 2D and 3D games.
WorldX turns a single sentence into a running simulated world, generating maps and characters with LLMs and letting autonomous agents act, remember and form stories inside it.
Control room for LLM agent societies: pause and scrub a running simulation, question individual residents, inject instructions into the next step, and export replayable experiment packs.
PyGame Learning Environment (PLE) is a Python-based reinforcement learning platform that provides a variety of games and tools to facilitate the development and testing of reinforcement learning algorithms.
GPTRPG is a proof-of-concept project featuring a GPT-based AI agent that autonomously interacts within a simple RPG-like environment using the OpenAI API.