google-deepmind/mujoco_playground v0.2.0

Release v0.2.0

Repository: google-deepmind/mujoco_playground

Tag: v0.2.0

Published: 2026-03-16T18:52:37Z

Prerelease: no

Release notes:

[0.2.0] - 2026-03-16

• Added vision-based PPO training configs for CartpoleBalance and

PandaPickCubeCartesian with tuned hyperparameters.

• Update vision notebooks to use MuJoCo Warp.
• Default MuJoCo implementation for all envs is now MuJoCo Warp.
• Renamed deprecated nconmax / nccdmax config fields to naconmax /

naccdmax across all environments, matching the updated MJX API.