Awesome AI Agents › Generative Agents Simulation
MineStudio is a comprehensive and streamlined package designed for the development of AI agents within the Minecraft environment. It offers a suite of tools and APIs that facilitate rapid and efficient creation, training, and deployment of Minecraft AI agents. The package includes a customizable Minecraft simulator built on top of MineRL, enabling users to simulate Minecraft gameplay scenarios for AI training and testing. It provides a specialized trajectory data structure for efficient storage and retrieval of arbitrary trajectory segments, which is crucial for handling large datasets of gameplay demonstrations and agent interactions. MineStudio also features a template for Minecraft policy models along with a collection of baseline models, allowing developers to build and experiment with different AI policies easily. The package supports both offline and online training pipelines: offline training enables pre-training of agents using existing datasets, while online training offers reinforcement learning implementations that support memory-based policies and include mechanisms for simulator crash recovery, ensuring robustness during training. For inference, MineStudio provides a parallelized and distributed framework based on Ray, facilitating scalable and efficient deployment of trained AI agents. Additionally, it includes benchmarking tools to automate and batch-test a variety of Minecraft tasks, helping users evaluate agent performance across different challenges. The project is actively maintained and encourages contributions. It supports installation via PyPI and offers a Docker image for containerized use. The package requires Python 3.10+, JDK 8, and rendering tools like VirtualGL or Xvfb for the simulator. MineStudio also integrates with Hugging Face, providing access to datasets converted from OpenAI's Video Pre-Training project and pretrained model checkpoints, making it easier for users to start training or inference with high-quality resources. Overall, MineStudio is a powerful and user-friendly platform for researchers and developers interested in Minecraft AI agent development, combining simulation, data management, model training, inference, and benchmarking in one cohesive package.
https://github.com/CraftJarvis/MineStudio
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.