Tianshou


<h1 align="center">Tianshou</h1>

![PyPI](https://img.shields.io/pypi/v/tianshou)
![Unittest](https://github.com/thu-ml/tianshou/workflows/Unittest/badge.svg?branch=master)
[![Documentation Status](https://readthedocs.org/projects/tianshou/badge/?version=latest)](https://tianshou.readthedocs.io/en/latest/?badge=latest)
[![GitHub stars](https://img.shields.io/github/stars/thu-ml/tianshou)](https://github.com/thu-ml/tianshou/stargazers)
[![GitHub forks](https://img.shields.io/github/forks/thu-ml/tianshou)](https://github.com/thu-ml/tianshou/network)
[![GitHub issues](https://img.shields.io/github/issues/thu-ml/tianshou)](https://github.com/thu-ml/tianshou/issues)
[![GitHub license](https://img.shields.io/github/license/thu-ml/tianshou)](https://github.com/thu-ml/tianshou/blob/master/LICENSE)

**Tianshou**(天授) is a reinforcement learning platform based on pure PyTorch. Unlike existing reinforcement learning libraries, which are mainly based on TensorFlow, have many nested classes, unfriendly API, or slow-speed, Tianshou provides a fast-speed framework and pythonic API for building the deep reinforcement learning agent. The supported interface algorithms include:


- [Policy Gradient (PG)](https://papers.nips.cc/paper/1713-policy-gradient-methods-for-reinforcement-learning-with-function-approximation.pdf)
- [Deep Q-Network (DQN)](https://storage.googleapis.com/deepmind-media/dqn/DQNNaturePaper.pdf)
- [Double DQN (DDQN)](https://arxiv.org/pdf/1509.06461.pdf) + n-step returns
- [Advantage Actor-Critic (A2C)](http://incompleteideas.net/book/RLbook2018.pdf)
- [Deep Deterministic Policy Gradient (DDPG)](https://arxiv.org/pdf/1509.02971.pdf)
- [Proximal Policy Optimization (PPO)](https://arxiv.org/pdf/1707.06347.pdf)
- [Twin Delayed DDPG (TD3)](https://arxiv.org/pdf/1802.09477.pdf)
- [Soft Actor-Critic (SAC)](https://arxiv.org/pdf/1812.05905.pdf)

Tianshou supports parallel workers for all algorithms as well. All of these algorithms are reformatted as replay-buffer based algorithms.

Tianshou is still under development. More algorithms are going to be added and we always welcome contributions to help make Tianshou better. If you would like to contribute, please check out [CONTRIBUTING.md](/CONTRIBUTING.md).

## Installation

Tianshou is currently hosted on [PyPI](https://pypi.org/project/tianshou/). You can simply install Tianshou with the following command:

```bash
pip3 install tianshou
```

## Documentation

The tutorials and API documentation are hosted on [https://tianshou.readthedocs.io](https://tianshou.readthedocs.io). It is under construction currently.

The example scripts are under [test/discrete](/test/discrete) (CartPole) and [test/continuous](/test/continuous) (Pendulum).

## Why Tianshou?

### Fast-speed

Tianshou is a lightweight but high-speed reinforcement learning platform. For example, here is a test on a laptop (i7-8750H + GTX1060). It only uses 3 seconds for training a agent based on vanilla policy gradient on the CartPole-v0 task.

![testpg](docs/_static/images/testpg.gif)

We select some of famous (>1k stars) reinforcement learning platforms. Here is the benchmark result for other algorithms and platforms on toy scenarios:

| RL Platform      | [Tianshou](https://github.com/thu-ml/tianshou)               | [Baselines](https://github.com/openai/baselines)             | [Ray/RLlib](https://github.com/ray-project/ray/tree/master/rllib/) | [PyTorch DRL](https://github.com/p-christ/Deep-Reinforcement-Learning-Algorithms-with-PyTorch) | [rlpyt](https://github.com/astooke/rlpyt)                    |
| ---------------- | ------------------------------------------------------------ | ------------------------------------------------------------ | ------------------------------------------------------------ | ------------------------------------------------------------ | ------------------------------------------------------------ |
| GitHub Stars     | [![GitHub stars](https://img.shields.io/github/stars/thu-ml/tianshou)](https://github.com/thu-ml/tianshou/stargazers) | [![GitHub stars](https://img.shields.io/github/stars/openai/baselines)](https://github.com/openai/baselines/stargazers) | [![GitHub stars](https://img.shields.io/github/stars/ray-project/ray)](https://github.com/ray-project/ray/stargazers) | [![GitHub stars](https://img.shields.io/github/stars/p-christ/Deep-Reinforcement-Learning-Algorithms-with-PyTorch)](https://github.com/p-christ/Deep-Reinforcement-Learning-Algorithms-with-PyTorch/stargazers) | [![GitHub stars](https://img.shields.io/github/stars/astooke/rlpyt)](https://github.com/astooke/rlpyt/stargazers) |
| Algo \ Task | PyTorch                                                      | TensorFlow                                                   | TF/PyTorch                                                   | PyTorch                                                      | PyTorch                                                      |
| PG - CartPole    | 9.03±4.18s                                                   | None                                                         |                                                              | None                                                         |                                                              |
| DQN - CartPole   | 20.94±11.38s                                                 | 1046.34±291.27s                                              |                                                              | 175.55±53.81s                                                |                                                              |
| A2C - CartPole   | 11.72±3.85s                                                  | *(~1612s)                                                    |                                                              | Runtime Error                                                |                                                              |
| PPO - CartPole   | 35.25±16.47s                                                 | *(~1179s)                                                    |                                                              | 29.16±15.46s                                                 |                                                              |
| DDPG - Pendulum  | 46.95±24.31s                                                 | *(>1h)                                                       |                                                              | 652.83±471.28s                                               | 172.18±62.48s                                                |
| TD3 - Pendulum   | 48.39±7.22s                                                  | None                                                         |                                                              | 619.33±324.97s                                               | 210.31±76.30s                                                |
| SAC - Pendulum   | 38.92±2.09s                                                  | None                                                         |                                                              | 808.21±405.70s                                               | 295.92±140.85s                                               |

*: Could not reach the target reward threshold in 1e6 steps in any of 10 runs. The total runtime is in the brackets.

All of the platforms use 10 different seeds for testing. We erase those trials which failed for training. The reward threshold is 195.0 in CartPole and -250.0 in Pendulum over consecutive 100 episodes' mean returns. 

### Reproducible

Tianshou has unit tests. Different from other platforms, **the unit tests include the full agent training procedure for all of the implemented algorithms**. It will be failed once it cannot train an agent to perform well enough on limited epochs on toy scenarios. The unit tests secure the reproducibility of our platform. 

Check out the [GitHub Actions](https://github.com/thu-ml/tianshou/actions) page for more detail.

### Elegant and Flexible

Currently, the overall code of Tianshou platform is less than 1500 lines. Most of the implemented algorithms are less than 100 lines of python code. It is quite easy to go through the framework and understand how it works. We provide many flexible API as you wish, for instance, if you want to use your policy to interact with environment with `n` steps:

```python
result = collector.collect(n_step=n)
```

If you have 3 environment in total and want to collect 1 episode in the first environment, 3 for third environment:

```python
result = collector.collect(n_episode=[1, 0, 3])
```

If you want to train the given policy with a sampled batch:

```python
result = policy.learn(collector.sample(batch_size))
```

You can check out the [documentation](https://tianshou.readthedocs.io) for further usage.

## Quick Start

This is an example of Policy Gradient. You can also run the full script under [test/discrete/test_pg.py](/test/discrete/test_pg.py).

First, import the relevant packages:

```python
import gym, torch, numpy as np, torch.nn as nn
from torch.utils.tensorboard import SummaryWriter

from tianshou.policy import PGPolicy
from tianshou.env import SubprocVectorEnv
from tianshou.trainer import onpolicy_trainer
from tianshou.data import Collector, ReplayBuffer
```

Define some hyper-parameters:

```python
task = 'CartPole-v0' 
seed = 1626 
lr = 3e-4 
gamma = 0.9 
epoch = 10 
step_per_epoch = 1000 
collect_per_step = 10 
repeat_per_collect = 2 
batch_size = 64 
train_num = 8 
test_num = 100 
device = 'cuda' if torch.cuda.is_available() else 'cpu'
writer = SummaryWriter('log/pg')  # tensorboard is also supported!
```

Define the network:

```python
class Net(nn.Module):
    def __init__(self, layer_num, state_shape, action_shape=0, device='cpu'):
        super().__init__()
        self.device = device
        self.model = [
            nn.Linear(np.prod(state_shape), 128),
            nn.ReLU(inplace=True)]
        for i in range(layer_num):
            self.model += [nn.Linear(128, 128), nn.ReLU(inplace=True)]
        if action_shape:
            self.model += [nn.Linear(128, np.prod(action_shape))]
        self.model = nn.Sequential(*self.model)

    def forward(self, s, state=None, info={}):
        if not isinstance(s, torch.Tensor):
            s = torch.tensor(s, device=self.device, dtype=torch.float)
        batch = s.shape[0]
        s = s.view(batch, -1)
        logits = self.model(s)
        return logits, state
```

Make envs and fix seed:

```python
env = gym.make(task)
state_shape = env.observation_space.shape or env.observation_space.n
action_shape = env.action_space.shape or env.action_space.n
train_envs = SubprocVectorEnv([lambda: gym.make(task) for _ in range(train_num)])
test_envs = SubprocVectorEnv([lambda: gym.make(task) for _ in range(test_num)])
np.random.seed(seed)
torch.manual_seed(seed)
train_envs.seed(seed)
test_envs.seed(seed)
```

Setup policy and collector:

```python
net = Net(3, state_shape, action_shape, device).to(device)
optim = torch.optim.Adam(net.parameters(), lr=lr)
policy = PGPolicy(net, optim, torch.distributions.Categorical, gamma)
train_collector = Collector(policy, train_envs, ReplayBuffer(20000))
test_collector = Collector(policy, test_envs)
```

Let's train it:

```python
result = onpolicy_trainer(policy, train_collector, test_collector, epoch, step_per_epoch, collect_per_step, repeat_per_collect, test_num, batch_size, stop_fn=lambda x: x >= env.spec.reward_threshold, writer=writer)
```

Saving / loading trained policy (it's exactly the same as PyTorch nn.module):

```python
torch.save(policy.state_dict(), 'pg.pth')
policy.load_state_dict(torch.load('pg.pth', map_location=device))
```

Watch the performance with 35 FPS:

```python3
collecter = Collector(policy, env)
collecter.collect(n_episode=1, render=1/35)
```

Looking at the result saved in tensorboard: (on bash script)

```bash
tensorboard --logdir log/pg
```

## Citing Tianshou

If you find Tianshou useful, please cite it in your publications.

```latex
@misc{tianshou,
  author = {Jiayi Weng},
  title = {Tianshou},
  year = {2020},
  publisher = {GitHub},
  journal = {GitHub repository},
  howpublished = {\url{https://github.com/thu-ml/tianshou}},
}
```

## TODO

- [ ] More examples on [mujoco, atari] benchmark
- [ ] Prioritized replay buffer
- [ ] RNN support
- [ ] Imitation Learning
- [ ] Multi-agent
- [ ] Distributed training

## Miscellaneous

Tianshou was previously a reinforcement learning platform based on TensorFlow. You can checkout the branch [`priv`](https://github.com/thu-ml/tianshou/tree/priv) for more detail.