What is disco-rl-pytorch?

lucidrains/disco-rl-pytorch — explained in plain English

Analysis updated 2026-06-24

★ 16PythonAudience · researcherComplexity · 4/5Setup · moderate

Mindmap

mindmap
  root((repo))
    What It Does
      DiscoRL algorithm
      Auto RL discovery
      Reference implementation
    Research Basis
      Nature 2025 paper
      David Silver DeepMind
      Test-time training link
    Tech Stack
      Python
      PyTorch
    Status
      Work in progress
      Minimal documentation
    Audience
      RL researchers
      AI paper readers

mindmap root((repo)) What It Does DiscoRL algorithm Auto RL discovery Reference implementation Research Basis Nature 2025 paper David Silver DeepMind Test-time training link Tech Stack Python PyTorch Status Work in progress Minimal documentation Audience RL researchers AI paper readers

Click or tap to explore — scroll the page freely

What do people build with it?

USE CASE 1

Study the DiscoRL algorithm from the 2025 Nature paper by reading a clean PyTorch reference implementation.

USE CASE 2

Experiment with automated reinforcement learning algorithm discovery by running DiscoRL on your own environments.

What is it built with?

PythonPyTorch

How does it compare?

	lucidrains/disco-rl-pytorch	adya84/ha-world-cup-2026	afk-surf/safeclipper
Stars	16	16	16
Language	Python	Python	Python
Setup difficulty	moderate	easy	moderate
Complexity	4/5	2/5	3/5
Audience	researcher	general	developer

Figures from each repo's GitHub metadata at analysis time.

How do you get it running?

Difficulty · moderate Time to first run · 1h+

Work in progress with minimal documentation, requires reading the DiscoRL paper to understand the algorithm before using the code.

So what is it?

This repository is a PyTorch implementation of DiscoRL, short for Discovering state-of-the-art reinforcement learning algorithms. The research it is based on was published in Nature in 2025 and represents the last work David Silver completed at DeepMind. Reinforcement learning is a field of AI where a system learns by trial and error, receiving rewards for good actions and penalties for bad ones, DiscoRL is a method for automatically discovering which learning algorithms perform best rather than relying on human-designed ones. The repository is marked as a work in progress, and the README is minimal: it contains a diagram, a brief description, and citation references for the underlying research paper and a related paper on test-time training. There is no setup guide, usage documentation, or code walkthrough provided at this stage. The project comes from lucidrains, a prolific open-source contributor known for implementing recent AI research papers in PyTorch as learning and reference resources.

Copy-paste prompts

Prompt 1

I am reading the DiscoRL Nature 2025 paper by David Silver. Walk me through how the disco-rl-pytorch implementation maps to the core algorithm described in the paper.

Prompt 2

I want to run disco-rl-pytorch on a custom Gym environment. What interface does my environment need to implement to plug in?

Prompt 3

The repository is marked as a work in progress. Which parts of the DiscoRL algorithm are already implemented and which are still missing or stubbed out?

Frequently asked questions