Minari
Farama-Foundation
Tools for storing, sharing and using offline reinforcement-learning datasets.
What you can explore
Find maintained dataset workflows for benchmarks migrated from D4RL and inspect episode metadata and environment compatibility.
Before you use it
Minari is a dataset tool, not a hand-control policy. Each dataset has its own version, action space, provenance and license.
Original contributors
Farama Foundation / Minari contributors
License / access: MIT for Farama-owned code · MIT/Apache 2.0 components; dataset terms vary
Keep exploring
D4RL / Adroit legacy benchmarks
Historical offline reinforcement-learning benchmarks, including Adroit dexterous-hand tasks.
Gymnasium Robotics
MuJoCo robotics environments with Gymnasium interfaces, including dexterous hand tasks.
LeRobot
Tools for recording datasets, training policies and exploring robot-learning workflows.
Editorial reference · checked 2026-09-22