Quick Start#
This section provides a hands-on introduction to reinforcement learning (RL) and OpenEnv through a series of interactive tutorials. Whether you’re new to RL or looking to learn how OpenEnv simplifies building and deploying environments, these tutorials will guide you through the fundamentals.
What is OpenEnv?
OpenEnv is a collaborative effort between Meta, Hugging Face, Unsloth, GPU Mode, Reflection, and other industry leaders to standardize reinforcement learning environments. Our goal is to make environment creation as easy and standardized as model sharing on Hugging Face.
Learning Path#
The tutorials are designed to be followed in sequence, building upon concepts from previous lessons:
Introduction & Quick Start - Understand what OpenEnv is, why it exists, and run your first environment. Includes a comparison with traditional solutions like OpenAI Gym.
Using Environments - Learn how to connect to environments (Hub, Docker, URL), create AI policies, and run evaluations. Work with different games and multi-player scenarios.
Building & Sharing Environments - Create your own custom environment from scratch, package it with Docker, and share it on Hugging Face Hub.
Packaging & Deploying - The complete reference guide for creating, packaging, and deploying custom environments with the
openenvCLI.Contributing to Hugging Face - Publish, fork, and contribute to environments hosted as Hugging Face Spaces.
No GPU Required! All five tutorials run without a GPU.
For GPU-intensive training workflows, see the RL Training Tutorial in the Tutorials section.
Prerequisites#
Before starting, ensure you have:
Basic Python programming knowledge
Python 3.11+ installed
Docker (optional, for container-based deployment)
Running the Tutorials#
You can run these tutorials locally:
# Install OpenEnv
pip install openenv-core
# Run the Python scripts
python plot_01_introduction_quickstart.py
Or view them directly in the documentation with full code output below.
Quick Start
