Open-source reinforcement-learning software
278 open-source libraries, tools and applications tagged reinforcement-learning, ranked by how closely reinforcement-learning is their primary purpose, then by adoption and maintenance health. Each entry links to a page with sources, use cases and when to choose or avoid it.
278 products, primary matches first, then adoption-weighted; health v2 shown.
| Product | Health v2 | Stars | Maturity |
|---|---|---|---|
| tinkoff-ai/CORL CORL is a research-oriented Python library providing thoroughly benchmarked single-file implementations of state-of-the-art offline and off… | 10 | 1374 | maintenance |
| Replicable-MARL/MARLlib MARLlib is a unified multi-agent reinforcement learning library built on Ray and RLlib with PyTorch, providing implementations of many MARL… | 23 | 1346 | maintenance |
| ikostrikov/pytorch-a3c A PyTorch implementation of the Asynchronous Advantage Actor Critic (A3C) reinforcement learning algorithm from the DeepMind paper 'Asynchr… | 32 | 1338 | maintenance |
| vietnh1009/Super-mario-bros-PPO-pytorch A PyTorch implementation of the Proximal Policy Optimization (PPO) algorithm that trains an agent to play Super Mario Bros via OpenAI Gym. … | 32 | 1302 | maintenance |
| Khrylx/PyTorch-RL A PyTorch library implementing deep reinforcement learning policy gradient algorithms (TRPO, PPO, A2C) and Generative Adversarial Imitation… | 32 | 1286 | maintenance |
| pfnet/pfrl PFRL is a PyTorch-based deep reinforcement learning library implementing state-of-the-art algorithms such as DQN, Rainbow, PPO, SAC, and A3… | 51 | 1277 | maintenance |
| google-research/planet Open-source implementation of the PlaNet agent, a purely model-based reinforcement learning algorithm that solves control tasks from pixels… | 10 | 1263 | maintenance |
| NeymarL/ChineseChess-AlphaZero An open-source implementation of AlphaZero/AlphaGo Zero reinforcement learning methods applied to Chinese chess (Xiangqi). It includes self… | 23 | 1249 | maintenance |
| chainer/chainerrl ChainerRL is a deep reinforcement learning library implementing state-of-the-art algorithms like DQN, Rainbow, PPO, SAC, and A3C in Python … | 23 | 1203 | maintenance |
| facebookresearch/House3D House3D is a virtual 3D environment of over 45k fully annotated indoor scenes from the SUNCG dataset, built for training embodied AI agents… | 10 | 1199 | maintenance |
| aitorzip/DeepGTAV DeepGTAV is a C++ plugin for Grand Theft Auto V that converts the game into a vision-based self-driving car research environment. It expose… | 32 | 1192 | maintenance |
| flow-project/flow Flow is a computational framework for deep reinforcement learning and control experiments in traffic microsimulation, built on top of the S… | 23 | 1192 | maintenance |
| openai/procgen A benchmark of 16 procedurally-generated, game-like Gym environments for measuring how quickly reinforcement learning agents learn generali… | 52 | 1191 | maintenance |
| huawei-noah/SMARTS SMARTS is a simulation platform for multi-agent reinforcement learning research focused on autonomous driving, emphasizing realistic and di… | 23 | 1135 | maintenance |
| andyzeng/visual-pushing-grasping PyTorch reference implementation of Visual Pushing and Grasping (VPG), which trains robotic agents via self-supervised deep reinforcement l… | 32 | 1113 | maintenance |
| nicrusso7/rex-gym Rex-gym provides OpenAI Gym environments for the open-source SpotMicro quadruped robot, built on PyBullet simulation, along with a PPO lear… | 32 | 1103 | maintenance |
| MatthewJA/Inverse-Reinforcement-Learning A Python library implementing selected inverse reinforcement learning (IRL) algorithms, including linear programming IRL, maximum entropy I… | 23 | 1092 | maintenance |
| ntasfi/PyGame-Learning-Environment PyGame Learning Environment (PLE) is a Python library providing a reinforcement learning environment with a suite of PyGame-based games, mi… | 32 | 1067 | maintenance |
| facebookresearch/mbrl-lib MBRL-Lib is a PyTorch-based toolbox from Facebook Research for developing model-based reinforcement learning algorithms. It provides interc… | 10 | 1063 | maintenance |
| isaac-sim/OmniIsaacGymEnvs A collection of GPU-accelerated reinforcement learning environments built on NVIDIA Isaac Sim and the omni.isaac.gym framework, with PPO tr… | 10 | 1049 | maintenance |
| Shunichi09/PythonLinearNonlinearControl A Python library implementing linear and nonlinear control theory algorithms such as MPC, iLQR, DDP, MPPI, and CEM using only NumPy and Sci… | 23 | 1043 | maintenance |
| Kismuz/btgym BTGym is a Python library that wraps the Backtrader algorithmic trading engine in an OpenAI Gym environment API, enabling event-driven back… | 32 | 1037 | maintenance |
| google-deepmind/scalable_agent A TensorFlow implementation of IMPALA (Importance Weighted Actor-Learner Architectures) for scalable distributed deep reinforcement learnin… | 32 | 1034 | maintenance |
| facebookresearch/CompilerGym CompilerGym is a Python library of OpenAI Gym-compatible reinforcement learning environments for compiler and program optimization tasks, c… | 10 | 1021 | maintenance |
| Continual-Intelligence/SEAL SEAL (Self-Adapting LLMs) is a research framework from MIT CSAIL that trains language models via reinforcement learning to generate their o… | 31 | 1870 | experimental |
| KhoomeiK/LlamaGym LlamaGym is a Python library that simplifies fine-tuning LLM-based agents with online reinforcement learning in Gym-style environments. It … | 26 | 1253 | experimental |
| SakanaAI/self-adaptive-llms Transformer² is a research framework from SakanaAI that adapts large language models to unseen tasks in real-time by selectively adjusting … | 22 | 1225 | experimental |
| Jiayi-Pan/TinyZero TinyZero is a minimal reproduction of DeepSeek R1-Zero, showing that a 3B base language model can develop self-verification and search abil… | 49 | 13248 | abandoned |
| karpathy/convnetjs ConvNetJS is a JavaScript library for defining and training neural networks, including convolutional networks and an experimental Deep Q Le… | 23 | 11208 | abandoned |
| openai/universe Universe is an OpenAI library that wraps games, websites, and other applications into OpenAI Gym environments so AI agents can be trained a… | 10 | 7491 | abandoned |
| janhuenermann/neurojs neurojs is a JavaScript deep learning framework for the browser with a focus on reinforcement learning, supporting deep Q-networks and acto… | 32 | 4367 | abandoned |
| facebookresearch/ReAgent ReAgent is an open-source end-to-end platform for applied reinforcement learning developed at Facebook, built in Python with PyTorch for tr… | 74 | 3722 | abandoned |
| tensorforce/tensorforce Tensorforce is an open-source deep reinforcement learning framework built on TensorFlow, emphasizing modular, configurable library design f… | 64 | 3305 | abandoned |
| xbpeng/DeepMimic DeepMimic is a C++ simulation framework with a Python (SWIG/TensorFlow) wrapper that trains simulated humanoid characters to imitate motion… | 53 | 3110 | abandoned |
| rll/rllab rllab is a Python framework for developing and evaluating reinforcement learning algorithms, with implementations of algorithms like TRPO, … | 32 | 3079 | abandoned |
| openai/multiagent-particle-envs A simple multi-agent particle world environment with continuous observations and discrete actions, used in the paper 'Multi-Agent Actor-Cri… | 10 | 2775 | abandoned |
| devsisters/DQN-tensorflow A TensorFlow implementation of the Deep Q-Network (DQN) algorithm from the DeepMind paper 'Human-Level Control through Deep Reinforcement L… | 32 | 2586 | abandoned |
| openai/roboschool Roboschool is an open-source robot simulation library providing physics-based Gym environments for reinforcement learning research, includi… | 10 | 2170 | abandoned |
| openai/maddpg Reference implementation of the MADDPG (Multi-Agent Deep Deterministic Policy Gradient) algorithm from OpenAI's paper on mixed cooperative-… | 10 | 1981 | abandoned |
| notadamking/RLTrader RLTrader is a Python library providing a cryptocurrency trading environment built on OpenAI's gym, using deep reinforcement learning to tra… | 23 | 1867 | abandoned |
| kuz/DeepMind-Atari-Deep-Q-Learner The original DQN 3.0 source code published by DeepMind alongside the 2015 Nature paper 'Human-level control through deep reinforcement lear… | 32 | 1830 | abandoned |
| openai/multi-agent-emergence-environments OpenAI's environment generation code for multi-agent reinforcement learning research, implementing environments like Hide and Seek from the… | 10 | 1815 | abandoned |
| geek-ai/MAgent MAgent is a research platform for many-agent reinforcement learning, supporting environments that scale from hundreds to millions of agents… | 32 | 1762 | abandoned |
| openai/neural-mmo Neural MMO is a massively multiagent game environment (MMORPG-style) for training and evaluating intelligent agents, released alongside an … | 10 | 1652 | abandoned |
| openai/evolution-strategies-starter A distributed Python implementation of Evolution Strategies (ES) from OpenAI's paper 'Evolution Strategies as a Scalable Alternative to Rei… | 10 | 1632 | abandoned |
| carpedm20/deep-rl-tensorflow A TensorFlow implementation of several Deep Reinforcement Learning papers, including DQN, Double DQN, and Dueling networks, for training ag… | 32 | 1580 | abandoned |
| deependersingla/deep_trader A Python research project applying deep reinforcement learning (DQN and policy gradient methods) to stock market trading, exploring whether… | 32 | 1506 | abandoned |
| openai/lm-human-preferences OpenAI's research code for the paper 'Fine-Tuning Language Models from Human Preferences', implementing reward model training from human la… | 10 | 1388 | abandoned |
| Atropos Atropos is a Python framework from Nous Research for building reinforcement learning environments that collect and evaluate LLM trajectorie… | 10 | 1351 | abandoned |
| haarnoja/sac The original reference implementation of Soft Actor-Critic (SAC), a deep reinforcement learning algorithm for training maximum entropy poli… | 32 | 1308 | abandoned |
| nv-tlabs/ASE ASE (Adversarial Skill Embeddings) is a research codebase from NVIDIA for training reusable low-level skill controllers for physically simu… | 54 | 1124 | abandoned |
| spragunr/deep_q_rl A Theano/Lasagne-based implementation of the Deep Q-Network (DQN) algorithm from DeepMind's Atari reinforcement learning papers. It trains … | 32 | 1094 | abandoned |
| coreylynch/async-rl A TensorFlow + Keras implementation of asynchronous 1-step Q-learning from the DeepMind 'Asynchronous Methods for Deep Reinforcement Learni… | 32 | 1003 | abandoned |
| ray-project/ray Ray is a unified open-source framework for scaling AI and Python applications, consisting of a core distributed runtime (tasks, actors, obj… | 99 | 44057 | stable |
| pytorch/examples A curated collection of short, high‑quality PyTorch example projects covering vision, text, reinforcement learning, and more. The repositor… | 46 | 24052 | active |
| CARLA Simulator CARLA is an open-source simulator for autonomous driving research built on Unreal Engine, providing open digital assets like urban layouts,… | 67 | 14465 | active |
| AidLux (AidLearning) AidLux (originally AidLearning) is an AIoT development platform that runs a native Ubuntu Linux environment with GUI, deep learning tooling… | 63 | 5812 | active |
| ddbourgin/numpy-ml numpy-ml is a collection of machine learning models and algorithms implemented exclusively in NumPy and the Python standard library, coveri… | 32 | 16323 | maintenance |
| OpenArm OpenArm is a fully open-source 7DOF humanoid robot arm (hardware, firmware, and software) designed for physical AI research and deployment … | 67 | 3605 | active |
| learnhouse/learnhouse LearnHouse is a next-generation open-source learning management system (LMS) for creating, sharing, and selling educational content. It com… | 100 | 2404 | active |
| levy-street/world-of-claudecraft World of ClaudeCraft is a free, open-source classic-style browser MMO built in TypeScript with Three.js, playable online with persistent ch… | 77 | 2302 | active |
| google/trax Trax is an end-to-end deep learning library built on JAX and TensorFlow that focuses on clear code and speed, developed and maintained by t… | 10 | 8306 | maintenance |
| lupantech/AgentFlow AgentFlow is a trainable, tool-integrated agentic framework that coordinates planner, executor, verifier, and generator modules through an … | 44 | 2064 | active |
| SerpentAI Framework Serpent.AI is a Python framework for building game agents—AIs and bots that learn to play any video game you own—turning games into machine… | 10 | 6989 | maintenance |
| allenai/ai2thor AI2-THOR is an open-source platform from the Allen Institute for AI providing near photo-realistic, interactable 3D environments (iTHOR, Ma… | 42 | 1809 | active |
| BindsNET/bindsnet BindsNET is a Python package for simulating spiking neural networks (SNNs) built on PyTorch tensor functionality, running on CPUs or GPUs. … | 81 | 1704 | active |
| fan-ziqi/rl_sar A C++ framework for simulation verification and physical deployment of reinforcement learning policies for robots, supporting quadruped, wh… | 74 | 1481 | active |
| AgentCPM AgentCPM-GUI is an open-source 8B-parameter on-device GUI agent built on MiniCPM-V that takes Android screenshots as input and autonomously… | 44 | 1420 | active |
| IrisRainbowNeko/genshin_auto_fish A Genshin Impact auto-fishing AI built from a YOLOX object detection model (fish and rod landing point localization) and a DQN reinforcemen… | 23 | 4748 | maintenance |
| ElectricAlexis/NotaGen NotaGen is a symbolic music generation model that produces high-quality classical sheet music using LLM-style training paradigms: pre-train… | 29 | 1236 | active |
| Fastbot Fastbot is a model-based GUI testing tool for Android apps that models GUI transitions using machine learning and reinforcement learning to… | 41 | 1210 | active |
| inclusionAI/UI-Venus UI-Venus is a family of open-source multimodal GUI agent models (9B/27B) that perform UI element grounding and task navigation from screens… | 61 | 1055 | active |
| metauto-ai/GPTSwarm GPTSwarm is a Python library for building LLM-based agents as computational graphs, with modules for agent graphs, memory, LLM backends, an… | 52 | 1045 | active |
| MoonshotAI/checkpoint-engine Checkpoint-engine is a lightweight Python middleware for updating model weights in-place across LLM inference engines, a critical step in r… | 78 | 1005 | active |
| facebookresearch/end-to-end-negotiator A PyTorch research codebase from Facebook AI Research implementing end-to-end learning for negotiation dialogues, including models for hier… | 10 | 1400 | maintenance |
| Chatbot_CN Chatbot_CN is a Chinese multi-turn, task-oriented chatbot platform targeting the finance and legal domains (with chitchat support), combini… | 32 | 1291 | maintenance |
| shepherd-agents/shepherd Shepherd is a Python runtime substrate that records LLM agent executions as reversible, Git-like traces so meta-agents can inspect, fork, r… | 77 | 2464 | experimental |
| Genesis-Embodied-AI/RoboGen RoboGen is a self-guided generative robotic agent that autonomously proposes new tasks, generates simulation environments, and learns robot… | 28 | 1228 | experimental |
← prev page 3 / 3