# gensyn-ai/rl-swarm

A fully open source framework for creating RL training swarms over the internet.

Repository: https://github.com/gensyn-ai/rl-swarm
Canonical: https://ross.abutalabs.com/products/rl-swarm
Homepage: https://www.gensyn.ai
Language: Python
License: MIT
License Family: permissive
Last push: 2026-01-05T21:15:30+00:00

## Health v2 (maintenance only)
Score: 54/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 60, release rhythm 56, longevity 39
- inputs: {"age_days": 554, "days_push": 240, "days_rel": 294, "gap_med": 4.0, "n_releases_24m": 25}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1679, forks 615 (observed 2026-08-28T04:05:21.062472+00:00)

## What it is
RL Swarm is an open-source, permissionless peer-to-peer framework for running reinforcement learning training swarms over the internet, built on the GenRL library. Participants train models collaboratively on shared tasks (currently code generation via the CodeZero environment) and can earn on-chain identity and progress tracking on the Gensyn Testnet.

## Use cases
- join a decentralized RL training swarm from a consumer laptop
- train LLMs collaboratively with peers over the internet
- run a swarm node on a GPU in the cloud
- participate in code-generation RL tasks with on-chain identity
- create custom multi-agent reinforcement learning swarms
- contribute compute to crowd-sourced model training runs

## When to choose
- you want to participate in or build decentralized, peer-to-peer RL training
- you want to run collaborative LLM fine-tuning on consumer hardware or cloud GPUs
- you need a composable multi-agent multi-stage RL environment framework
- you want on-chain identity and reputation for training participation

## When to avoid
- you need a single-node, fully local training pipeline without networking
- you require production-grade supervised fine-tuning with guaranteed convergence
- you cannot run Docker or manage Python environments
- you need an officially running swarm right now, since none are currently active

## Facets
- artifact type: framework
- maturity: active
- function: llm-training, reinforcement-learning, agent-framework, p2p
- domain: machine-learning, large-language-models, artificial-intelligence, microservices, blockchain
- platform: python, cross-platform
- tags: reinforcement-learning-swarm, decentralized-training, peer-to-peer-ml, gensyn-testnet, collaborative-training, multi-agent-rl, docker, linux, macos, gpu

## Member repositories
- gensyn-ai/rl-swarm (main) score 54

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:21.062472+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 2026-08-30T03:41:31.163136+00:00, confidence not recorded.
  - readme: https://github.com/gensyn-ai/rl-swarm (fetched 2026-08-28T04:05:21.062472+00:00, sha d3b64c899b6e)
  - homepage: https://www.gensyn.ai (fetched 2026-08-29T11:14:45.744813+00:00, sha 0ac898f1cacd)
- Data as of 2026-08-30T08:39:29.467469+00:00.