d3rlpy: An offline deep reinforcement learning library

d3rlpy is an offline deep reinforcement learning library for practitioners and researchers.

```py import d3rlpy

dataset, env = d3rlpy.datasets.get_dataset("hopper-medium-v0")

prepare algorithm

sac = d3rlpy.algos.SACConfig(compile_graph=True).create(device="cuda:0")

train offline

sac.fit(dataset, n_steps=1000000)

train online

sac.fitonline(env, nsteps=1000000)

ready to control

actions = sac.predict(x) ```

• Documentation: https://d3rlpy.readthedocs.io
• Paper: https://arxiv.org/abs/2111.03788

[!IMPORTANT] v2.x.x introduces breaking changes. If you still stick to v1.x.x, please explicitly install previous versions (e.g. pip install d3rlpy==1.1.1).

Key features

:zap: Most Practical RL Library Ever

• offline RL: d3rlpy supports state-of-the-art offline RL algorithms. Offline RL is extremely powerful when the online interaction is not feasible during training (e.g. robotics, medical).
• online RL: d3rlpy also supports conventional state-of-the-art online training algorithms without any compromising, which means that you can solve any kinds of RL problems only with d3rlpy.

:beginner: User-friendly API

• zero-knowledge of DL library: d3rlpy provides many state-of-the-art algorithms through intuitive APIs. You can become a RL engineer even without knowing how to use deep learning libraries.
• extensive documentation: d3rlpy is fully documented and accompanied with tutorials and reproduction scripts of the original papers.

:rocket: Beyond State-of-the-art

• distributional Q function: d3rlpy is the first library that supports distributional Q functions in the all algorithms. The distributional Q function is known as the very powerful method to achieve the state-of-the-performance.
• data-prallel distributed training: d3rlpy is the first library that supports data-parallel distributed offline RL training, which allows you to scale up offline RL with multiple GPUs or nodes. See example.

Installation

d3rlpy supports Linux, macOS and Windows.

Dependencies

Installing d3rlpy package will install or upgrade the following packages to satisfy requirements: - torch>=2.5.0 - tqdm>=4.66.3 - gym>=0.26.0 - gymnasium==1.0.0 - click - colorama - dataclasses-json - h5py - structlog - typing-extensions - scikit-learn

PyPI (recommended)

$ pip install d3rlpy

Anaconda

$ conda install conda-forge/noarch::d3rlpy

Docker

$ docker run -it --gpus all --name d3rlpy takuseno/d3rlpy:latest bash

Supported algorithms

| algorithm | discrete control | continuous control | |:-|:-:|:-:| | Behavior Cloning (supervised learning) | :whitecheckmark: | :whitecheckmark: | | Neural Fitted Q Iteration (NFQ) | :whitecheckmark: | :noentry: | | Deep Q-Network (DQN) | :whitecheckmark: | :noentry: | | Double DQN | :whitecheckmark: | :noentry: | | Deep Deterministic Policy Gradients (DDPG) | :noentry: | :whitecheckmark: | | Twin Delayed Deep Deterministic Policy Gradients (TD3) | :noentry: | :whitecheckmark: | | Soft Actor-Critic (SAC) | :whitecheckmark: | :whitecheckmark: | | Batch Constrained Q-learning (BCQ) | :whitecheckmark: | :whitecheckmark: | | Bootstrapping Error Accumulation Reduction (BEAR) | :noentry: | :whitecheckmark: | | Conservative Q-Learning (CQL) | :whitecheckmark: | :whitecheckmark: | | Advantage Weighted Actor-Critic (AWAC) | :noentry: | :whitecheckmark: | | Critic Reguralized Regression (CRR) | :noentry: | :whitecheckmark: | | Policy in Latent Action Space (PLAS) | :noentry: | :whitecheckmark: | | TD3+BC | :noentry: | :whitecheckmark: | | Policy Regularization with Dataset Constraint (PRDC) | :noentry: | :whitecheckmark: | | Implicit Q-Learning (IQL) | :noentry: | :whitecheckmark: | | Calibrated Q-Learning (Cal-QL) | :noentry: | :whitecheckmark: | | ReBRAC | :noentry: | :whitecheckmark: | | Decision Transformer | :whitecheckmark: | :whitecheckmark: | | Q-learning Decision Transformer (QDT) | :noentry: | :whitecheckmark: | | Transformer Actor-Critic with Regularization (TACR) | :noentry: | :whitecheckmark: | | Gato | :construction: | :construction: |

Supported Q functions

• [x] standard Q function
• [x] Quantile Regression
• [x] Implicit Quantile Network

Benchmark results

d3rlpy is benchmarked to ensure the implementation quality. The benchmark scripts are available reproductions directory. The benchmark results are available d3rlpy-benchmarks repository.

Examples

MuJoCo

```py import d3rlpy

prepare dataset

dataset, env = d3rlpy.datasets.get_d4rl('hopper-medium-v0')

prepare algorithm

cql = d3rlpy.algos.CQLConfig(compile_graph=True).create(device='cuda:0')

train

cql.fit( dataset, n_steps=100000, evaluators={"environment": d3rlpy.metrics.EnvironmentEvaluator(env)}, ) ```

See more datasets at d4rl.

Atari 2600

```py import d3rlpy

prepare dataset (1% dataset)

dataset, env = d3rlpy.datasets.getataritransitions( 'breakout', fraction=0.01, num_stack=4, )

prepare algorithm

cql = d3rlpy.algos.DiscreteCQLConfig( observationscaler=d3rlpy.preprocessing.PixelObservationScaler(), rewardscaler=d3rlpy.preprocessing.ClipRewardScaler(-1.0, 1.0), compile_graph=True, ).create(device='cuda:0')

start training

cql.fit( dataset, n_steps=1000000, evaluators={"environment": d3rlpy.metrics.EnvironmentEvaluator(env, epsilon=0.001)}, ) ```

See more Atari datasets at d4rl-atari.

Online Training

```py import d3rlpy import gym

prepare environment

env = gym.make('Hopper-v3') eval_env = gym.make('Hopper-v3')

prepare algorithm

sac = d3rlpy.algos.SACConfig(compile_graph=True).create(device='cuda:0')

prepare replay buffer

buffer = d3rlpy.dataset.createfiforeplay_buffer(limit=1000000, env=env)

start training

sac.fitonline(env, buffer, nsteps=1000000, evalenv=evalenv) ```

Tutorials

Try cartpole examples on Google Colaboratory!

• offline RL tutorial:
• online RL tutorial:

More tutorial documentations are available here.

Contributions

Any kind of contribution to d3rlpy would be highly appreciated! Please check the contribution guide.

Community

| Channel | Link | |:-|:-| | Issues | GitHub Issues |

[!IMPORTANT] Please do NOT email to any contributors including the owner of this project to ask for technical support. Such emails will be ignored without replying to your message. Use GitHub Issues to report your problems.

Projects using d3rlpy

| Project | Description | |:-:|:-| | MINERVA | An out-of-the-box GUI tool for offline RL | | SCOPE-RL | An off-policy evaluation and selection library |

Roadmap

The roadmap to the future release is available in ROADMAP.md.

Citation

The paper is available here. @article{d3rlpy, author = {Takuma Seno and Michita Imai}, title = {d3rlpy: An Offline Deep Reinforcement Learning Library}, journal = {Journal of Machine Learning Research}, year = {2022}, volume = {23}, number = {315}, pages = {1--20}, url = {http://jmlr.org/papers/v23/22-0017.html} }

Acknowledgement

This work started as a part of Takuma Seno's Ph.D project at Keio University in 2020.

This work is supported by Information-technology Promotion Agency, Japan (IPA), Exploratory IT Human Resources Project (MITOU Program) in the fiscal year 2020.