Awesome AI Agents › Generative Agents Simulation
JARVIS-1 is an advanced open-world agent designed to operate within the Minecraft universe, capable of perceiving multimodal inputs such as visual observations and human instructions. It leverages pre-trained multimodal language models to translate these inputs into sophisticated plans, which are then executed through goal-conditioned controllers. A key innovation of JARVIS-1 is its multimodal memory system, which integrates both pre-trained knowledge and experiential data from actual gameplay, enabling more effective planning and decision-making. This agent is highly versatile, able to complete over 200 distinct tasks ranging from simple, short-term objectives like chopping trees to complex, long-term goals such as obtaining a diamond pickaxe. JARVIS-1 demonstrates near-perfect performance on short-horizon tasks and significantly outperforms current state-of-the-art agents in long-horizon tasks, achieving up to five times greater reliability in challenging scenarios. The project is Linux-based and recommends using Anaconda for environment management, with dependencies including JDK 8 and specific Python packages. Users can run the agent in both graphical and headless modes, with plans to release additional features such as dynamic memory learning and multimodal descriptors to enhance the agent's understanding of the visual world. JARVIS-1 builds upon related projects like STEVE-1, Minedojo, and MC-TextWorld, integrating their capabilities to create a comprehensive and generalist Minecraft agent. The project is supported by a detailed research paper available on arXiv, which provides in-depth insights into the methodologies and performance of the agent. Overall, JARVIS-1 represents a significant step forward in the development of generalist AI agents capable of complex planning and control in open-world environments using multimodal inputs and memory-augmented language models.
https://github.com/CraftJarvis/JARVIS-1
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.
A locally runnable, low-cost implementation of generative agents simulating human-like behavior in an interactive Dungeons & Dragons setting using large language models.