Minari
- Software
- Humanoids and robot learning
- Tool
Minari is a Python library for offline reinforcement learning research, offering a store of datasets to read from and tools to write new ones. It works like an offline counterpart of Gymnasium.
Made by Farama Foundation.
Source on GitHub: github.com/Farama-Foundation/Minari (opens in a new tab). Facts read at commit d88d29d (opens in a new tab).
Licenses
- Code
- Apache-2.0 (opens in a new tab)
Read by Robotheca from LICENSE (opens in a new tab) (Apache-2.0) at commit d88d29d (opens in a new tab), on 16 Oct 2026.
- Commercial use
- Yes
What it does
- Installs from PyPI with pip, with an optional extra for all dependencies.
- Provides a command line interface to list, download, and show remote and local datasets.
- Loads datasets such as D4RL/door/human-v2 and iterates over their episodes with observations, actions, rewards, and terminations.
- Collects data from a Gymnasium environment with a DataCollector wrapper and saves it as a new dataset.
- Integrates with the training libraries TorchRL, d3rlpy, and AgileRL.
- Includes a tutorial that recreates the Maze2D datasets from D4RL.
Written by Robotheca from the project's README (opens in a new tab) at commit d88d29d.
Getting started
- Difficulty
- Intermediate
- How to start
- The project's website (opens in a new tab)
Maintenance
- Status
- Active: a commit in the last 90 days
- Newest commit
- 14 Oct 2026 on main
- Latest release
- v0.5.4 (14 Sept 2026)
Checked on GitHub on 16 Oct 2026.
Links
Robots that can run it
No robot on Robotheca is linked to it yet. Links come from a robot's robot.yaml, from its dependency files read as text, or from Robotheca's editors.
Robotheca lists software as a reference: nothing is hosted or run here, and every fact links to where it was read.