# Farama-Foundation/Minari

A standard format for offline reinforcement learning datasets, with popular reference datasets and related utilities

Repository: https://github.com/Farama-Foundation/Minari
Canonical: https://ross.abutalabs.com/products/minari
Homepage: https://minari.farama.org
Language: Python
License: NOASSERTION
License Family: other
Topics: datasets, gymnasium, offline-rl, reinforcement-learning
Last push: 2026-08-04T23:48:43+00:00

## Health v2 (maintenance only)
Score: 66/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 96, release rhythm 8, longevity 100
- inputs: {"age_days": 1427, "days_push": 29, "days_rel": 693, "gap_med": null, "n_releases_24m": 1}
- flags: no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1290, forks 73 (observed 2026-08-28T04:04:15.470148+00:00)

## What it is
Minari is a Python library providing a standard format for offline reinforcement learning datasets, with popular reference datasets and utilities for creating, storing, and loading them. It integrates with Gymnasium and offers a CLI for browsing and downloading datasets like D4RL.

## Use cases
- download offline RL datasets like D4RL
- create offline RL datasets by collecting trajectories from Gymnasium environments
- load and iterate over offline RL episodes for training
- run behavioral cloning or implicit Q-learning experiments
- share offline RL datasets in a standard format

## When to choose
- you need offline RL datasets in a standard format compatible with Gymnasium
- you want to collect your own trajectory datasets from Gymnasium environments
- you want access to reference datasets like D4RL, Minigrid, or PointMaze

## When to avoid
- you need online reinforcement learning environments rather than offline datasets
- you work outside Python or need non-Gymnasium data formats

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, reinforcement-learning, cli, serialization
- domain: reinforcement-learning, machine-learning, artificial-intelligence
- platform: python, cli, cross-platform
- tags: offline-rl, gymnasium, d4rl, datasets, farama, dataset

## Member repositories
- Farama-Foundation/Minari (main) score 66

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:15.470148+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T04:54:50.533455+00:00, confidence not recorded.
  - readme: https://github.com/Farama-Foundation/Minari (fetched 2026-08-28T04:04:15.470148+00:00, sha 84fdbc4c4db1)
  - homepage: https://minari.farama.org (fetched 2026-08-29T12:11:26.993113+00:00, sha 26bb03659b5c)
  - registry_pypi: https://pypi.org/pypi/minari/json (fetched 2026-08-29T12:11:27.002236+00:00, sha 8077884e45c6)
- Data as of 2026-08-30T08:39:29.467469+00:00.