#4949th of 56 in ML frameworks and libraries
stable-baselines3
PyTorch implementations of reinforcement learning algorithms with a scikit-learn-like API
- Stars
- 13.9k
- License
- MIT
- Last commit
- Sep 2026
- Last release
- Jun 2026
- Language
- Python
Missed the latest GitHub check; these facts may be older.
Overview
Stable Baselines3 provides tested PyTorch implementations of RL algorithms including PPO, A2C, DQN, DDPG, SAC, TD3 and HER, driven through a sklearn-style `learn()` and `predict()` interface. It works with Gymnasium environments, supports Dict observation spaces, custom policies and callbacks, and logs to TensorBoard. Development is limited to maintenance; newer algorithms live in SB3 Contrib and a Jax port, SBX.
Who it is for: RL researchers and engineers needing reliable baseline algorithm implementations
Strengths
- Consistent sklearn-like API across algorithms; a PPO run is a one-line call
- Supports Gymnasium envs, Dict observations, custom policies, callbacks and TensorBoard
- Type hints, high test coverage, and benchmark results documented per algorithm
- Companion RL Zoo provides training scripts and tuned hyperparameters
Weaknesses
- Assumes prior RL knowledge; README says not to use it without practice
- Maintenance mode only; no major new features planned for the core
- TRPO, QR-DQN, TQC, RecurrentPPO live in the separate SB3 Contrib repo
- No technical support or consulting offered by maintainers
What it needs
- GPU optional
- Docker
- Needs PyTorch >= 2.8, Gymnasium, Python 3.10+
- Models: PPO, A2C, DQN, DDPG, SAC
Also in ML frameworks and libraries
See all 56| Rank | Project | Score |
|---|---|---|
| 1 | transformersPython library defining open models for text, vision, audio and multimodal tasks | 95 out of 100 |
| 2 | pytorchPython tensor library with GPU support and tape-based autograd for deep learning | 94 out of 100 |
| 3 | ultralyticsPython library and CLI for training and running YOLO vision models | 92 out of 100 |
| #4 | supervisionPython toolkit for loading, annotating and counting computer vision detections | 89 out of 100 |
| #5 | pytorch-image-modelsPyTorch library of image models, pretrained weights and reference training scripts | 89 out of 100 |