# NousResearch/atropos

Atropos is a Language Model Reinforcement Learning Environments framework for collecting and evaluating LLM trajectories through diverse environments

Repository: https://github.com/NousResearch/atropos
Canonical: https://ross.abutalabs.com/products/nousresearch-atropos
Language: Python
License: MIT
License Family: permissive
Archived: true
Last push: 2026-07-04T17:39:01+00:00

## Health v2 (maintenance only)
Score: 10/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 90, release rhythm 62, longevity 35
- inputs: {"age_days": 491, "days_push": 60, "days_rel": 176, "gap_med": 59, "n_releases_24m": 4}
- flags: archived
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1350, forks 397 (observed 2026-08-28T04:04:28.034048+00:00)

## What it is
Atropos is a Python framework from Nous Research for building reinforcement learning environments that collect and evaluate LLM trajectories. It is now archived and no longer maintained.

## Use cases
- train LLMs with reinforcement learning in custom environments
- collect model trajectories for RLHF-style training
- evaluate language models across diverse task environments
- build an RL gym for language models
- generate rollouts from LLMs for fine-tuning

## When to choose
- you want to study or fork an LLM RL environments framework
- you need a reference design for trajectory collection environments
- you are doing research compatible with its existing environment implementations

## When to avoid
- you need maintained software with bug fixes and security patches
- you want a production RL training pipeline with active support
- you need compatibility with the latest LLM training stacks

## Facets
- artifact type: framework
- maturity: abandoned
- function: reinforcement-learning, llm-training, machine-learning, benchmarking
- domain: reinforcement-learning, large-language-models, machine-learning, artificial-intelligence
- platform: python
- tags: llm-rl-gym, trajectory-collection, rl-environments, archived, nous-research, linux, gpu

## Member repositories
- NousResearch/atropos (main) score 10

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:28.034048+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 2026-08-30T04:42:15.680463+00:00, confidence not recorded.
  - readme: https://github.com/NousResearch/atropos (fetched 2026-08-28T04:04:28.034048+00:00, sha 917434807163)
- Data as of 2026-08-30T08:39:29.467469+00:00.