# openai/baselines

OpenAI Baselines: high-quality implementations of reinforcement learning algorithms

Repository: https://github.com/openai/baselines
Canonical: https://ross.abutalabs.com/products/baselines
Language: Python
License: MIT
License Family: permissive
Last push: 2024-08-01T21:31:33+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 3389, "days_push": 762, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 16764, forks 4932 (observed 2026-08-28T04:11:17.180007+00:00)

## What it is
OpenAI Baselines is a collection of high-quality reference implementations of reinforcement learning algorithms such as DQN, PPO, and ACKTR, built on TensorFlow. It is intended for the research community to replicate published results, compare new methods against standard baselines, and build new ideas on top of well-tested algorithm implementations.

## Use cases
- implement reinforcement learning algorithms in python
- replicate published RL paper results like DQN scores
- compare a new RL algorithm against standard baselines
- train agents on Atari and MuJoCo environments
- get reference implementations of PPO and DQN
- build RL research on top of tested algorithm code

## When to choose
- you need canonical, paper-accurate RL algorithm implementations for research
- you want to benchmark a new RL method against established baselines
- you are working with TensorFlow 1.x and classic RL environments

## When to avoid
- you need actively maintained code or modern TensorFlow/PyTorch support
- you want production-ready RL training infrastructure rather than research reference code
- you need the latest RL algorithms or ecosystem integrations like Gymnasium

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, reinforcement-learning, benchmarking
- domain: reinforcement-learning, machine-learning, artificial-intelligence
- platform: python
- tags: reinforcement-learning, dqn, ppo, openai, tensorflow, research-baselines, mujoco, research, linux, macos, gpu

## Member repositories
- openai/baselines (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:11:17.180007+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 2026-08-29T17:04:16.395377+00:00, confidence not recorded.
  - readme: https://github.com/openai/baselines (fetched 2026-08-28T04:11:17.180007+00:00, sha 0f2067ab3812)
  - registry_pypi: https://pypi.org/pypi/baselines/json (fetched 2026-08-29T08:01:51.800136+00:00, sha cdbbb2b2c0a8)
- Data as of 2026-08-30T08:39:29.467469+00:00.