The mario-ai project offers a comprehensive pipeline for developing AI agents to master Super Mario Bros. using neuroevolution. By integrating a Python-based NEAT implementation with the OpenAI Gym SuperMario environment, it allows users to define custom fitness criteria, mutation rates, and network topologies. During training, the framework evaluates generations of neural networks, selects high-performing genomes, and provides real-time visualization of both gameplay and network evolution. Additionally, it supports saving and loading trained models, exporting champion genomes, and generating detailed performance logs. Researchers, educators, and hobbyists can extend the codebase to other game environments, experiment with evolutionary strategies, and benchmark AI learning progress across different levels.

multiagent_envs delivers a modular set of Python-based environments tailored for multi-agent reinforcement learning research and development. It includes scenarios like cooperative navigation, predator-prey, social dilemmas, and competitive arenas. Each environment lets you define the number of agents, observation features, reward functions, and collision dynamics. The framework integrates seamlessly with popular RL libraries such as Stable Baselines and RLlib, allowing vectorized training loops, parallel execution, and easy logging. Users can extend existing scenarios or create new ones by following a simple API, accelerating experimentation with algorithms like MADDPG, QMIX, and PPO in a consistent, reproducible setup.

MultiAgentModel provides a unified API to define custom environments and agent classes for multi-agent scenarios. Developers can specify observation and action spaces, reward structures, and communication channels. Built-in support for popular RL algorithms like PPO, DQN, and A2C allows training with minimal configuration. Real-time visualization tools help monitor agent interactions and performance metrics. The modular architecture ensures easy integration of new algorithms and custom modules. It also includes a flexible configuration system for hyperparameter tuning, logging utilities for experiment tracking, and compatibility with OpenAI Gym environments for seamless portability. Users can collaborate on shared environments and replay logged sessions for analysis.

Vanilla Agents is a lightweight PyTorch-based framework that delivers modular and extensible implementations of core reinforcement learning agents. It supports algorithms like DQN, Double DQN, PPO, and A2C, with pluggable environment wrappers compatible with OpenAI Gym. Users can configure hyperparameters, log training metrics, save checkpoints, and visualize learning curves. The codebase is organized for clarity, making it ideal for research prototyping, educational use, and benchmarking new ideas in RL.

The YGO-Agent framework allows researchers and enthusiasts to develop AI bots that play the Yu-Gi-Oh card game using reinforcement learning. It wraps the YGOPRO game simulator into an OpenAI Gym-compatible environment, defining state representations such as hand, field, and life points, and action representations including summoning, spell/trap activation, and attacking. Rewards are based on win/loss outcomes, damage dealt, and game progress. The agent architecture uses PyTorch to implement DQN, with options for custom network architectures, experience replay, and epsilon-greedy exploration. Logging modules record training curves, win rates, and detailed move logs for analysis. The framework is modular, enabling users to replace or extend components such as the reward function or action space.