abcdrl 0.2.0a4

Modular Single-file Reinfocement Learning Algorithms Library

abcdRL (Implement a RL algorithm in four simple steps)

English | 简体中文

abcdRL is a Modular Single-file Reinforcement Learning Algorithms Library that provides modular design without strict and clean single-file implementation.

When reading the code, understand the full implementation details of the algorithm in the single file quickly; When modifying the algorithm, benefiting from a lightweight modular design, only need to focus on a small number of modules.

abcdRL mainly references the single-file design philosophy of vwxyzjn/cleanrl and the module design of PaddlePaddle/PARL.

Documentation ➡️ docs.abcdrl.xyz

Roadmap🗺️ #57

🚀 Quickstart

Open the project in Gitpod🌐 and start coding immediately.

Using Docker📦:

# 0. Prerequisites: Docker & Nvidia Drive & NVIDIA Container Toolkit
# 1. Run DQN algorithm
docker run --rm --gpus all sdpkjc/abcdrl python abcdrl/dqn.py

For detailed installation instructions 👀

🐼 Features

• 👨‍👩‍👧‍👦 Unified code structure
• 📄 Single-file implementation
• 🐷 Low code reuse
• 📐 Minimizing code differences
• 📈 Tensorboard & Wandb support
• 🛤 PEP8(code style) & PEP526(type hint) compliant

🗽 Design Philosophy

• "Copy📋", not "Inheritance🧬"
• "Single-file📜", not "Multi-file📚"
• "Features reuse🛠", not "Algorithms reuse🖨"
• "Unified logic🤖", not "Unified interface🔌"

✅ Implemented Algorithms

Weights & Biases Benchmark Report ➡️ report.abcdrl.xyz

• Deep Q Network (DQN)
• Deep Deterministic Policy Gradient (DDPG)
• Twin Delayed Deep Deterministic Policy Gradient (TD3)
• Soft Actor-Critic (SAC)
• Proximal Policy Optimization (PPO)

---

• Double Deep Q Network (DDQN)
• Prioritized Deep Q Network (PDQN)

Citing abcdRL

@misc{zhao_abcdrl_2022,
    author = {Yanxiao, Zhao},
    month = {12},
    title = {{abcdRL: Modular Single-file Reinforcement Learning Algorithms Library}},
    url = {https://github.com/sdpkjc/abcdrl},
    year = {2022}
}