# openai/evolution-strategies-starter

Code for the paper "Evolution Strategies as a Scalable Alternative to Reinforcement Learning"

Repository: https://github.com/openai/evolution-strategies-starter
Canonical: https://ross.abutalabs.com/products/evolution-strategies-starter
Homepage: https://arxiv.org/abs/1703.03864
Language: Python
License: MIT
License Family: permissive
Topics: paper
Archived: true
Last push: 2019-10-31T09:20:13+00:00

## Health v2 (maintenance only)
Score: 10/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 3449, "days_push": 2498, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, archived
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1635, forks 279 (observed 2026-08-28T04:05:14.584129+00:00)

## What it is
A distributed Python implementation of Evolution Strategies (ES) from OpenAI's paper 'Evolution Strategies as a Scalable Alternative to Reinforcement Learning'. It uses a master-worker architecture designed to scale to over a thousand parallel workers on AWS EC2, including spot instances.

## Use cases
- run distributed evolution strategies for reinforcement learning
- train RL policies on MuJoCo humanoid tasks
- scale black-box optimization across many CPUs
- reproduce results from the ES paper
- run RL experiments on AWS spot instances

## When to choose
- you want to experiment with evolution strategies as an RL alternative
- you need a reference implementation of the scalable ES algorithm
- you have AWS infrastructure and want massively parallel black-box optimization

## When to avoid
- you need maintained or updated code (the repo is archived)
- you want a modern RL framework with GPU support
- you lack an AWS account or Mujoco license

## Facets
- artifact type: library
- maturity: abandoned
- function: machine-learning, reinforcement-learning, concurrency
- domain: machine-learning, artificial-intelligence, cloud-computing
- platform: python, cloud
- tags: evolution-strategies, distributed-training, research-code, aws, mujoco, black-box-optimization, linux

## Member repositories
- openai/evolution-strategies-starter (main) score 10

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:14.584129+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 2026-08-30T03:47:01.991161+00:00, confidence not recorded.
  - readme: https://github.com/openai/evolution-strategies-starter (fetched 2026-08-28T04:05:14.584129+00:00, sha a1b192a2ff2d)
  - homepage: https://arxiv.org/abs/1703.03864 (fetched 2026-08-29T11:20:13.698454+00:00, sha e193ed980e93)
  - site_page: https://info.arxiv.org/about/donate.html (fetched 2026-08-29T11:20:13.707602+00:00, sha cca9c3a11c56)
  - site_page: https://info.arxiv.org/about/ourmembers.html (fetched 2026-08-29T11:20:13.711113+00:00, sha 47cbc55ff1de)
  - site_page: https://info.arxiv.org/about (fetched 2026-08-29T11:20:13.713104+00:00, sha a1f16f915a9a)
  - site_page: https://info.arxiv.org/labs/index.html (fetched 2026-08-29T11:20:13.709409+00:00, sha b14a8d05a0ec)
- Data as of 2026-08-30T08:39:29.467469+00:00.