Denys88/rl_games

Health v2 · maintenance only

Adoption not part of the score

1383 stars · 220 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

RL Games is a high-performance reinforcement learning library built on PyTorch, focused on training agents in massively parallel GPU-based simulation environments like Isaac Gym and MuJoCo. It provides implementations of popular RL algorithms (e.g., PPO, A2C) tuned for games and robotics workloads.

Use cases

• train RL agents on GPU-parallel physics simulations
• learn locomotion policies for quadrupeds and humanoids
• train dexterous in-hand manipulation policies
• run self-play training at large scale
• transfer simulated robot policies to real hardware
• benchmark RL algorithms on EnvPool environments

When to choose

• you need high-throughput RL training with thousands of parallel environments
• you work with NVIDIA Isaac Gym, Isaac Lab, or MuJoCo/MJLab robotics tasks
• you want proven PPO implementations used in published robotics research
• you train policies for games or robotics on GPU hardware

When to avoid

• you need a beginner-friendly RL library with extensive tutorials
• you require a broad algorithm zoo beyond PPO-style on-policy methods
• you work outside Python/PyTorch ecosystems
• you need stable, well-documented APIs for production services rather than research training

Facets

library · maturity active

machine-learning reinforcement-learning deep-learning reinforcement-learning machine-learning robotics python cross-platform pytorch ppo a2c isaac-gym mujoco envpool robotics gpu-simulation self-play game-development gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="Denys88/rl_games")