LeRobot EnvHub

LeRobot's mechanism for sharing simulation environments, publishing environment code to the Hugging Face Hub for one-line loading.

EnvHub is a simulation-environment-sharing feature Hugging Face built into its LeRobot framework. Previously, using someone else's simulated task meant cloning their repository, installing its dependencies, and rewriting code to match their interface; EnvHub lets an author publish environment code as a Hub repository, and a user in LeRobot then loads it by repository name and gets back an environment conforming to the Gymnasium interface (reset/step), ready to use for training or evaluating a policy. Because loading it executes code from that repository, it requires explicitly allowing remote code execution, so it should only be loaded from a trusted source. It lives on the same Hub as LeRobot's datasets and models, making it convenient to publish data, a policy, and an evaluation environment together in one place.

Related

LeRobot (Hugging Face) · Hugging Face · Gymnasium (formerly OpenAI Gym) · Simulation-based Evaluation · LeRobotDataset (LeRobot dataset format)

Sources

LeRobot 文档 EnvHub (Chinese)
huggingface/lerobot (GitHub)

As of

2026-09