ChainerRL is a deep reinforcement learning library that implements various state-of-the-art deep reinforcement algorithms in Python using Chainer, a flexible deep learning framework.
ChainerRL is tested with Python 2.7+ and 3.5.1+. For other requirements, see requirements.txt.
ChainerRL can be installed via PyPI:
pip install chainerrl
It can also be installed from the source code:
python setup.py install
Refer to Installation for more information on installation.
You can try ChainerRL Quickstart Guide first, or check the examples ready for Atari 2600 and Open AI Gym.
For more information, you can refer to ChainerRL's documentation.
| Algorithm | Discrete Action | Continous Action | Recurrent Model | CPU Async Training |
|---|---|---|---|---|
| DQN (including DoubleDQN etc.) | ✓ | ✓ (NAF) | ✓ | x |
| Categorical DQN | ✓ | x | ✓ | x |
| Rainbow | ✓ | x | ✓ | x |
| IQN | ✓ | x | x | x |
| DDPG | x | ✓ | ✓ | x |
| A3C | ✓ | ✓ | ✓ | ✓ |
| ACER | ✓ | ✓ | ✓ | ✓ |
| NSQ (N-step Q-learning) | ✓ | ✓ (NAF) | ✓ | ✓ |
| PCL (Path Consistency Learning) | ✓ | ✓ | ✓ | ✓ |
| PPO | ✓ | ✓ | x | x |
| TRPO | ✓ | ✓ | x | x |
Following algorithms have been implemented in ChainerRL:
- A3C (Asynchronous Advantage Actor-Critic)
- ACER (Actor-Critic with Experience Replay)
- Asynchronous N-step Q-learning
- Rainbow
- Categorical DQN
- IQN
- DQN (including Double DQN, Persistent Advantage Learning (PAL), Double PAL, Dynamic Policy Programming (DPP))
- DDPG (Deep Deterministic Policy Gradients) (including SVG(0))
- PGT (Policy Gradient Theorem)
- PCL (Path Consistency Learning)
- PPO (Proximal Policy Optimization)
- TRPO (Trust Region Policy Optimization)
Q-function based algorithms such as DQN can utilize a Normalized Advantage Function (NAF) to tackle continuous-action problems as well as DQN-like discrete output networks.
The following papers have been implemented in ChainerRL:
- Playing Atari with Deep Reinforcement Learning
- Human-level control through Deep Reinforcement Learning
- Deep Reinforcement Learning with Double Q-learning
- Prioritized Experience Replay
- Dueling Network Architectures for Deep Reinforcement Learning
- Asynchronous Methods for Deep Reinforcement Learning
- A Distributional Perspective on Reinforcement Learning
- Implicit Quantile Networks for Distributional Reinforcement Learning
- Rainbow: Combining Improvements in Deep Reinforcement Learning
- Increasing the Action Gap: New Operators for Reinforcement Learning
- Noisy Networks for Exploration
- Continuous control with deep reinforcement learning
- Proximal Policy Optimization Algorithms
- Trust Region Policy Optimization
- Sample Efficient Actor-Critic with Experience Replay
- Bridging the Gap Between Value and Policy Based Reinforcement Learning
ChainerRL has a set of accompanying visualization tools in order to aid developers' ability to understand and debug their RL agents. With this visualization tool, the behavior of ChainerRL agents can be easily inspected from a browser UI.
Environments that support the subset of OpenAI Gym's interface (reset and step methods) can be used.
Any kind of contribution to ChainerRL would be highly appreciated! If you are interested in contributing to ChainerRL, please read CONTRIBUTING.md.