Eclectic-Sheep/sheeprl

Distributed Reinforcement Learning accelerated by Lightning Fabric

What it solves

SheepRL is a PyTorch-based reinforcement learning (RL) framework designed to be simple and scalable. It addresses the common problem of tight coupling between RL algorithms and their environments, allowing users to easily decouple them so that algorithms can be applied to a wider variety of environments beyond the standard gym interface.

How it works

The framework leverages Lightning Fabric to accelerate training and scale across multiple devices. It provides a set of pre-implemented RL algorithms (such as PPO, SAC, and Dreamer) that can be configured via a CLI. Users can run experiments by specifying the algorithm and environment through command-line arguments, and the framework handles the interaction loop between the agent and the environment.

Who it’s for

It is intended for researchers and developers working with reinforcement learning who need a scalable, easy-to-use library that supports a wide range of observation types (vector, pixel) and action spaces (continuous, discrete, multi-discrete).

Highlights

  • Broad Algorithm Support: Includes implementations of A2C, PPO, SAC, DroQ, and the Dreamer series (V1, V2, V3) along with Plan2Explore.
  • Extensive Environment Integration: Supports a variety of environments including Atari, Mujoco, MineRL, MineDojo, DIAMBRA, and Crafter.
  • Scalable Infrastructure: Built with Lightning Fabric for easy acceleration and multi-device distribution.
  • Detailed Benchmarking: Provides performance comparisons against Stable Baselines3 across various algorithms and hardware configurations.

Related

  • Project
    Lightning-AI/pytorch-lightning

    A deep learning framework that abstracts away the engineering boilerplate of PyTorch, enabling researchers to pretrain and finetune AI models at scale across various hardware.

  • Project
    meta-pytorch/torchforge

    A PyTorch-native agentic RL library that separates infrastructure concerns from model concerns to simplify scalable reinforcement learning experimentation.

  • Project
    kengz/SLM-Lab

    A modular PyTorch framework for Deep Reinforcement Learning that allows users to configure and run RL experiments via JSON specs across various Gymnasium environments.

  • Project
    yihaosun1124/OfflineRL-Kit

    A PyTorch library for offline reinforcement learning that provides a standardized framework and implementations of state-of-the-art model-free and model-based algorithms.

  • Project
    Denys88/rl_games

    A high-performance reinforcement learning library implemented in PyTorch that optimizes training throughput for robotics, multi-agent systems, and complex simulations.