MARL-DPP is an open-source framework enabling multi-agent reinforcement learning (MARL) with enforced diversity through Determinantal Point Processes (DPP). Traditional MARL approaches often suffer from policy convergence to similar behaviors; MARL-DPP addresses this by incorporating DPP-based measures to encourage agents to maintain diverse action distributions. The toolkit provides modular code for embedding DPP in training objectives, sampling policies, and managing exploration. It includes ready-to-use integration with standard OpenAI Gym environments and the Multi-Agent Particle Environment (MPE), along with utilities for hyperparameter management, logging, and visualization of diversity metrics. Researchers can evaluate the impact of diversity constraints on cooperative tasks, resource allocation, and competitive games. The extensible design supports custom environments and advanced algorithms, facilitating exploration of novel MARL-DPP variants.

Multi-Agent Reinforcement Learning by alaamoheb is a comprehensive open-source library designed to facilitate the development, training, and evaluation of multiple agents acting in shared environments. It includes modular implementations of value-based and policy-based algorithms such as DQN, PPO, MADDPG, and more. The repository supports integration with OpenAI Gym, Unity ML-Agents, and the StarCraft Multi-Agent Challenge, allowing users to experiment in both research and real-world inspired scenarios. With configurable YAML-based experiment setups, logging utilities, and visualization tools, practitioners can monitor learning curves, tune hyperparameters, and compare different algorithms. This framework accelerates experimentation in cooperative, competitive, and mixed multi-agent tasks, streamlining reproducible research and benchmarking.

This repository provides a complete suite of multi-agent reinforcement learning algorithms—including MADDPG, DDPG, PPO, and more—integrated with standard benchmarks like the Multi-Agent Particle Environment and OpenAI Gym. It features customizable environment wrappers, configurable training scripts, real-time logging, and performance evaluation metrics. Users can easily extend algorithms, adapt to custom tasks, and compare policies across cooperative and adversarial settings with minimal setup.

RL Shooter is a Python-based framework that integrates ViZDoom with OpenAI Gym APIs to create a flexible reinforcement learning environment for FPS games. Users can define custom scenarios, maps, and reward structures to train agents on navigation, target detection, and shooting tasks. With configurable observation frames, action spaces, and logging facilities, it supports popular deep RL libraries such as Stable Baselines and RLlib, enabling clear performance tracking and reproducibility across experiments.

CommNet is a research-oriented library that implements the CommNet architecture, allowing multiple agents to share hidden states at each timestep and learn to coordinate actions in cooperative environments. It includes PyTorch model definitions, training and evaluation scripts, environment wrappers for OpenAI Gym, and utilities for customizing communication channels, agent counts, and network depths. Researchers and developers can use CommNet to prototype and benchmark inter-agent communication strategies on navigation, pursuit–evasion, and resource-collection tasks.

RL-Agents is a research-grade reinforcement learning framework built on PyTorch that bundles popular RL algorithms across value-based, policy-based, and actor-critic methods. The library features a modular agent API, GPU acceleration, seamless integration with OpenAI Gym, and built-in logging and visualization tools. Users can configure hyperparameters, customize training loops, and benchmark performance with a few lines of code, making RL-Agents ideal for academic research, prototyping, and industrial experimentation.

Text-to-Reward provides a pipeline to train reward models that map text-based task descriptions or feedback into scalar reward values for RL agents. Leveraging transformer-based architectures and fine-tuning on collected human preference data, the framework automatically learns to interpret natural language instructions as reward signals. Users can define arbitrary tasks via text prompts, train the model, and then incorporate the learned reward function into any RL algorithm. This approach eliminates manual reward shaping, boosts sample efficiency, and enables agents to follow complex multi-step instructions in simulated or real-world environments.