Stable Baselines3 RL Colab Notebooks

강화학습을 수행하는데 있어 보다 효율적으로 진행할 수 있도록 도와주는 라이브러리에는 여러가지가 있지만 Stable Baselines를 무시할 수 없습니다. 본래 OpenAI사에서 개발한 Baselines를 보다 개선하여 탄생하게 된 라이브러리입니다.
현재는 Baselines3까지 나와있는 상태이고요. 기본적인 사용법과 예제는 Colab버전으로 친절하게 제공되고 있습니다.

 

Colab notebooks part of the documentation of Stable Baselines3 reinforcement learning library

Those notebooks are independent examples.

Documentation: https://stable-baselines3.readthedocs.io/

Content

• Getting Started Colab Notebook
• Saving, loading Colab Notebook
• Multiprocessing Colab Notebook
• Monitor Training Colab Notebook
• Atari games Colab Notebook
• PyBullet: Normalizing input features Colab Notebook
• Pre-training using Behavior Cloning Colab Notebook
• RL Baselines3 Zoo Colab Notebook
• Hindsight Experience Replay Colab Notebook
• Advanced Saving and Loading Colab Notebook
• Track Experiments with Weights and Biases Colab Notebook

RL Tutorial JNRR19

We recommend looking at rl-tutorial-jnrr19 for a more complete tutorial.

1. Getting Started Colab Notebook
2. Gym Wrappers, saving and loading models Colab Notebook
3. Multiprocessing Colab Notebook
4. Callbacks and hyperparameter tuning Colab Notebook
5. Creating a custom gym environment Colab Notebook