cherry-rl 0.0.9

PyTorch Reinforcement Learning Framework for Researchers

---

Cherry is a reinforcement learning framework for researchers built on top of PyTorch.

Unlike other reinforcement learning implementations, cherry doesn't implement a single monolithic interface to existing algorithms. Instead, it provides you with low-level, common tools to write your own algorithms. Drawing from the UNIX philosophy, each tool strives to be as independent from the rest of the framework as possible. So if you don't like a specific tool, you don’t need to use it.

Features

• Pythonic and modular interface à la Pytorch.
• Support for tabular (!) and function approximation algorithms.
• Various OpenAI Gym environment wrappers.
• Helper functions for popular algorithms. (e.g. A2C, DDPG, TRPO, PPO, SAC)
• Logging, visualization, and debugging tools.
• Painless and efficient distributed training on CPUs and GPUs.
• Unit, integration, and regression tested, continuously integrated.

Installation

Note Cherry is considered in alpha release. Some stuff might break in the future.

For the latest release of cherry, run the following command in your favorite shell.

pip install cherry-rl

For the cutting edge version, you can run the following commands.

1. Clone the repo: git clone https://github.com/seba-1511/cherry
2. cd cherry
3. pip install -e .

Requirements

• torch >= 1.0.0
• gym >= 0.15

Development Guidelines

• The master branch is always working, considered stable.
• The dev branch should always work and is ahead of master, considered cutting edge.
• To implement a new functionality: branch dev into your_name/functionality_name, implement your functionality, then pull request to dev. It will be periodically merged into master.

Usage

The following snippet demonstrates some of the tools offered by cherry.

import cherry as ch

# Wrapping environments
env = ch.envs.Logger(env, interval=1000)  # Prints rollouts statistics
env = ch.envs.Normalized(env, normalize_state=True, normalize_reward=False)  
env = ch.envs.Torch(env)  # Converts actions/states to tensors

# Storing and retrieving experience
replay = ch.ExperienceReplay()
replay.append(old_state, action, reward, state, done, info = {
    'log_prob': mass.log_prob(action),  # Can add any variable/tensor to the transitions
    'value': value
})
replay.actions  # Tensor of all stored actions
replay.states  # Tensor of all stored states
replay.empty()  # Removes all stored experience

# Discounting and normalizing rewards
rewards = ch.rewards.discount(GAMMA, replay.rewards, replay.dones)
rewards = ch.utils.normalize(rewards)

# Sampling rollouts per episode or samples
env = envs.Runner(env)
replay = env.run(get_action, steps=100)  # alternatively: episodes=10

Concrete examples are available in the examples/ folder.

Documentation

Documentation and tutorials are available on cherry’s website: http://seba-1511.github.io/cherry.

TODO

Some functionalities that we might want to implement.

• parallelize environments and a way to handle it with ExperienceReplay,
• VisdomLogger as a dashboard to debug an implementation,
• example with reccurent net,
• minimal but complete documentation,
• GPU implementations.

Acknowledgements

Cherry draws inspiration from many reinforcement learning implementations, including

• OpenAI Baselines,
• John Schulman's implementations
• Ilya Kostrikov's implementations,
• Shangtong Zhang's implementations,
• Dave Abel's implementations,
• Vitchyr Pong's implementations,
• Kai Arulkumaran's implementations,
• RLLab.

Why 'cherry' ?

Because it's the sweetest part of the cake.