cpprb is a python (CPython) module providing replay buffer classes for reinforcement learning.
Major target users are researchers and library developers.
You can build your own reinforcement learning algorithms together with your favorite deep learning library (e.g. TensorFlow, PyTorch).
cpprb forcuses speed, flexibility, and memory efficiency.
By utilizing Cython, complicated calculations (e.g. segment tree for prioritized experience replay) are offloaded onto C++. (The name cpprb comes from “C++ Replay Buffer”.)
In terms of API, initially cpprb referred to OpenAI Baselines’ implementation. The current version of cpprb has much more flexibility. Any NumPy compatible types of any numbers of values can be stored (as long as memory capacity is sufficient). For example, you can store the next action and the next next observation, too.
cpprb requires following softwares before installation.
- C++17 compiler (for installation from source)
- GCC (maybe 7.2 and newer)
- Visual Studio (2017 Enterprise is fine)
- Python 3
- pip
Additionally, here are user’s good feedbacks for installation at Ubuntu. (Thanks!)
2.1 Install from PyPI (Recommended)
The following command installs cpprb together with other dependencies.
pip install cpprbDepending on your environment, you might need sudo or --user flag
for installation.
On supported platflorms (Linux x86-64, Windows amd64, and macOS x86_64), binary packages hosted on PyPI can be used, so that you don’t need C++ compiler. On the other platforms, such as 32bit or arm-architectured Linux and Windows, you cannot install from binary, and you need to compile by yourself. Please be patient, we plan to support wider platforms in future.
If you have any troubles to install from binary, you can fall back to
source installation by passing --no-binary option to the above pip
command. (In order to avoid NumPy source installation, it is better to
install NumPy beforehand.)
pip install numpy
pip install --no-binary cpprbFirst, download source code manually or clone the repository;
git clone https://gitlab.com/ymd_h/cpprb.gitThen you can install in the same way;
cd cpprb
pip install .For this installation, you need to convert extended Python (.pyx) to C++ (.cpp) during installation, it takes longer time than installation from PyPI.
Basic usage is following step;
- Create replay buffer (
ReplayBuffer.__init__) - Add transitions (
ReplayBuffer.add)- Reset at episode end (
ReplayBuffer.on_episode_end)
- Reset at episode end (
- Sample transitions (
ReplayBuffer.sample)
Here is a simple example for storing standard environment (aka. obs,
act, rew, next_obs, and done).
from cpprb import ReplayBuffer
buffer_size = 256
obs_shape = 3
act_dim = 1
rb = ReplayBuffer(buffer_size,
env_dict ={"obs": {"shape": obs_shape},
"act": {"shape": act_dim},
"rew": {},
"next_obs": {"shape": obs_shape},
"done": {}})
obs = np.ones(shape=(obs_shape))
act = np.ones(shape=(act_dim))
rew = 0
next_obs = np.ones(shape=(obs_shape))
done = 0
for i in range(500):
rb.add(obs=obs,act=act,rew=rew,next_obs=next_obs,done=done)
if done:
# Together with resetting environment, call ReplayBuffer.on_episode_end()
rb.on_episode_end()
batch_size = 32
sample = rb.sample(batch_size)
# sample is a dictionary whose keys are 'obs', 'act', 'rew', 'next_obs', and 'done'(See also API reference)
| Name | Type | Optional | Discription |
|---|---|---|---|
size | int | No | Buffer size |
env_dict | dict | Yes (but unusable) | Environment definition (See here) |
next_of | str or array-like of str | Yes | Memory compression (See here) |
stack_compress | str or array-like of str | Yes | Memory compression (See here) |
default_dtype | numpy.dtype | Yes | Fall back data type |
Nstep | dict | Yes | Nstep configuration (See here) |
mmap_prefix | str | Yes | mmap file prefix (See here) |
Flexible environment values are defined by env_dict when buffer
creation. The detail is described at document.
Since stored values have flexible name, you have to pass to
ReplayBuffer.add member by keyword.
cpprb provides buffer classes for building following algorithms.
| Algorithms | cpprb class | Paper |
|---|---|---|
| Experience Replay | ReplayBuffer | L. J. Lin |
| Prioritized Experience Replay | PrioritizedReplayBuffer | T. Schaul et. al. |
| Multi-step (Nstep) Learning | ReplayBuffer, PrioritizedReplayBuffer | |
| Multiprocess Learning (Ape-X) | MPReplayBuffer MPPrioritizedReplayBuffer | D. Horgan et. al. |
| Large Batch Experience Replay (LaBER) | LaBERmean, LaBERlazy, LaBERmax | T. Lahire et al. |
| Reverse Experience Replay (RER) | ReverseReplayBuffer | E. Rotinov |
| Hindsight Experience Replay (HER) | HindsightReplayBuffer | M. Andrychowicz et al. |
cpprb features and its usage are described at following pages:
- Flexible Environment
- Multi-step add
- Prioritized Experience Replay
- Nstep Experience Replay
- Memory Compression
- Map Large Data on File
- Multiprocess Learning (Ape-X)
- Save/Load Transitions
One of the most distinctive design of cpprb is column-oriented flexibly defined transitions. As far as we know, other replay buffer implementations adopt row-oriented flexible transitions (aka. array of transition class) or column-oriented non-flexible transitions.
In deep reinforcement learning, sampled batch is divided into
variables (i.e. obs, act, etc.). If the sampled batch is
row-oriented, users (or library) need to convert it into
column-oriented one. (See doc, too)
cpprb can accept addition of multiple transitions simultaneously. This
design is convenient when batch transitions are moved from local
buffers to a global buffer. Moreover it is more efficient because of
not only removing pure-Python for loop but also suppressing
unnecessary priority updates for PER. (See doc, too)
We try to minimize dependency. Only NumPy is required during its execution. Small dependency is always preferable to avoid dependency hell.
Any contribution are very welcome!
Bigger commumity makes development more active and improve cpprb.
- Star GitLab repository (and/or GitHub Mirror)
- Publish your code using cpprb
- Share this repository to your friend and/or followers.
When you have any problems or requests, you can check Discussions on github.com. If you still cannot find any information, you can post your own.
We keep issues on GitLab.com and users are still allowed to open issues, however, we mainly use the place as development issue tracker.
cpprb follows local rules:
- Branch Name
- “HotFix_***” for bug fix
- “Feature_***” for new feature implementation
- docstring
- Must for external API
- Numpy Style
- Unit Test
- Put test code under “test/” directory
- Can test by
python -m unittest <Your Test Code>command - Continuous Integration on GitLab CI configured by
.gitlab-ci.yaml
- Open an issue and associate it to Merge Request
Step by step instruction for beginners is described at here.
- keiohta/TF2RL
- TensorFlow2.x Reinforcement Learning
- 【強化学習】cpprb で Experience Replay を簡単に！| Qiita
- 【強化学習】Ape-X の高速な実装を簡単に！| Qiita
- 【強化学習】自作ライブラリでDQN | Qiita
- 【強化学習】Ape-Xの高速化を実現 | Zenn
- 【強化学習】cpprb に遷移のファイル保存機能を追加 | Zenn
We would be very happy if you cite cpprb in your papers.
@misc{Yamada_cpprb_2019,
author = {Yamada, Hiroyuki},
month = {1},
title = {{cpprb}},
url = {https://gitlab.com/ymd_h/cpprb},
year = {2019}
}- 3rd Party Papers citing cpprb
- E. Aitygulov and A. I. Panov, “Transfer Learning with Demonstration Forgetting for Robotic Manipulator”, Proc. Comp. Sci. 186 (2021), 374-380, https://doi.org/10.1016/j.procs.2021.04.159
- T. Kitamura and R. Yonetani, “ShinRL: A Library for Evaluating RL Algorithms from Theoretical and Practical Perspectives”, NeurIPS Deep RL Workshop (2021) (arXiv, code)