Skip to content

CleanRL v1 Release

🎉 We are thrilled to announce the v1.0.0 CleanRL Release. Along with our CleanRL paper's recent publication in Journal of Machine Learning Research, our v1.0.0 release includes reworked documentation, new algorithm variants, support for google's new ML framework JAX, hyperparameter tuning utilities, and more. CleanRL has come a long way making high-quality deep reinforcement learning implementations easy to understand and reproducible. This release is a major milestone for the project and we are excited to share it with you. Over 90 PRs were merged to make this release possible. We would like to thank all the contributors who made this release possible.

More detailed release notes are available at v1.0.0b1, v1.0.0b2, and v1.0.0.

Reworked documentation

One of the biggest change of the v1 release is the added documentation at Having great documentation is important for building a reliable and reproducible project. We have reworked the documentation to make it easier to understand and use. For each implemented algorithm, we have documented as much as we can to promote transparency:

Here is a list of the algorithm variants and their documentation:

Algorithm Variants Implemented
Proximal Policy Gradient (PPO), docs, docs, docs, docs, docs, docs, docs, docs, docs, docs
Deep Q-Learning (DQN), docs, docs, docs, docs
Categorical DQN (C51), docs, docs
Soft Actor-Critic (SAC), docs
Deep Deterministic Policy Gradient (DDPG), docs, docs
Twin Delayed Deep Deterministic Policy Gradient (TD3), docs, docs
Phasic Policy Gradient (PPG), docs
Random Network Distillation (RND), docs

We also improved the contribution guide to make it easier for new contributors to get started. We are still working on improving the documentation. If you have any suggestions, please let us know in the GitHub Issues.

New algorithm variants, support for JAX

We now support JAX-based learning algorithm variants, which are usually faster than the torch equivalent! Here are the docs of the new JAX-based DQN, TD3, and DDPG implementations:

For example, below are the benchmark of DDPG + JAX (see docs here for further detail):

Other new algorithm variants include multi-GPU PPO, PPO prototype that works with Isaac Gym, multi-agent Atari PPO, and refactored PPG and PPO-RND implementations:

Tooling improvements

We love tools! The v1.0.0 release comes with a series of DevOps improvements, including pre-commit utilities, CI integration with GitHub to run end-to-end test cases. We also make available a new hyperparameter tuning tool and a new tool for running benchmark experiments.


We added a pre-commit utility to help contributors to format their code, check for spelling, and removing unused variables and imports before submitting a pull request (see Contribution guide for more detail).

To ensure our single-file implementations can run without error, we also added CI/CD pipeline which now runs end-to-end test cases for all the algorithm variants. The pipeline also tests builds across different operating systems, such as Linux, macOS, and Windows (see here as an example). GitHub actions are free for open source projects, and we are very happy to have this tool to help us maintain the project.

Hyperparameter tuning utilities

We now have preliminary support for hyperparameter tuning via optuna (see docs), which is designed to help researchers to find a single set of hyperparameters that work well with a kind of games. The current API looks like below:

import optuna
from cleanrl_utils.tuner import Tuner
tuner = Tuner(
        "CartPole-v1": [0, 500],
        "Acrobot-v1": [-500, 0],
    params_fn=lambda trial: {
        "learning-rate": trial.suggest_loguniform("learning-rate", 0.0003, 0.003),
        "num-minibatches": trial.suggest_categorical("num-minibatches", [1, 2, 4]),
        "update-epochs": trial.suggest_categorical("update-epochs", [1, 2, 4, 8]),
        "num-steps": trial.suggest_categorical("num-steps", [5, 16, 32, 64, 128]),
        "vf-coef": trial.suggest_uniform("vf-coef", 0, 5),
        "max-grad-norm": trial.suggest_uniform("max-grad-norm", 0, 5),
        "total-timesteps": 100000,
        "num-envs": 16,

Benchmarking utilities

We also added a new tool for running benchmark experiments. The tool is designed to help researchers to quickly run benchmark experiments across different algorithms environments with some random seeds. The tool lives in the cleanrl_utils.benchmark module, and the users can run commands such as:

OMP_NUM_THREADS=1 xvfb-run -a python -m cleanrl_utils.benchmark \
    --env-ids CartPole-v1 Acrobot-v1 MountainCar-v0 \
    --command "poetry run python cleanrl/ --cuda False --track --capture-video" \
    --num-seeds 3 \
    --workers 5

which will run the script with --cuda False --track --capture-video arguments across 3 random seeds for 3 environments. It uses multiprocessing to create a pool of 5 workers run the experiments in parallel.

What’s next?

It is an exciting time and new improvements are coming to CleanRL. We plan to add more JAX-based implementations, huggingface integration, some RLops prototypes, and support Gymnasium. CleanRL is a community-based project and we always welcome new contributors. If there is an algorithm or new feature you would like to contribute, feel free to chat with us on our discord channel or raise a GitHub issue.

More JAX implementations

More JAX-based implementation are coming. Antonin Raffin, the core maintainer of Stable-baselines3, SBX, and rl-baselines3-zoo, is contributing an optimized Soft Actor Critic implementation in JAX ( vwxyzjn/cleanrl#300) and TD3+TQC, and DroQ ( vwxyzjn/cleanrl#272. These are incredibly exciting new algorithms. For example, DroQ is extremely sample effcient and can obtain ~5000 return in HalfCheetah-v3 in just 100k steps (tracked sbx experiment).

Huggingface integration

Huggingface Hub 🤗 is a great platform for sharing and collaborating models. We are working on a new integration with Huggingface Hub to make it easier for researchers to share their RL models and benchmark them against other models ( vwxyzjn/cleanrl#292). Stay tuned! In the future, we will have a simple snippet for loading models like below:

import random
from typing import Callable

import gym
import numpy as np
import torch

def evaluate(
    model_path: str,
    make_env: Callable,
    env_id: str,
    eval_episodes: int,
    run_name: str,
    Model: torch.nn.Module,
    device: torch.device,
    epsilon: float = 0.05,
    capture_video: bool = True,
    envs = gym.vector.SyncVectorEnv([make_env(env_id, 0, 0, capture_video, run_name)])
    model = Model(envs).to(device)

    obs = envs.reset()
    episodic_returns = []
    while len(episodic_returns) < eval_episodes:
        if random.random() < epsilon:
            actions = np.array([envs.single_action_space.sample() for _ in range(envs.num_envs)])
            q_values = model(torch.Tensor(obs).to(device))
            actions = torch.argmax(q_values, dim=1).cpu().numpy()
        next_obs, _, _, infos = envs.step(actions)
        for info in infos:
            if "episode" in info.keys():
                print(f"eval_episode={len(episodic_returns)}, episodic_return={info['episode']['r']}")
                episodic_returns += [info["episode"]["r"]]
        obs = next_obs

    return episodic_returns

if __name__ == "__main__":
    from huggingface_hub import hf_hub_download

    from cleanrl.dqn import QNetwork, make_env

    model_path = hf_hub_download(repo_id="cleanrl/CartPole-v1-dqn-seed1", filename="q_network.pth")


How do we know the effect of a new feature / bug fix? DRL is brittle and has a series of reproducibility issues — even bug fixes sometimes could introduce performance regression (e.g., see how a bug fix of contact force in MuJoCo results in worse performance for PPO). Therefore, it is essential to understand how the proposed changes impact the performance of the algorithms.

We are working a prototype tool that allows us to compare the performance of the library at different versions of the tracked experiment ( vwxyzjn/cleanrl#307). With this tool, we can confidently merge new features / bug fixes without worrying about introducing catastrophic regression. The users can run commands such as:

python -m cleanrl_utils.rlops --exp-name ddpg_continuous_action \
    --wandb-project-name cleanrl \
    --wandb-entity openrlbenchmark \
    --tags 'pr-299' 'rlops-pilot' \
    --env-ids HalfCheetah-v2 Walker2d-v2 Hopper-v2 InvertedPendulum-v2 Humanoid-v2 Pusher-v2 \
    --output-filename compare.png \
    --scan-history \
    --metric-last-n-average-window 100 \

which generates the following image

Support for Gymnasium

Farama-Foundation/Gymnasium is the next generation of openai/gym that will continue to be maintained and introduce new features. Please see their announcement for further detail. We are migrating to gymnasium and the progress can be tracked in vwxyzjn/cleanrl#277.

Also, the Farama foundation is working a project called Shimmy which offers conversion wrapper for deepmind/dm_env environments, such as dm_control and deepmind/lab. This is an exciting project that will allow us to support deepmind/dm_env in the future.


CleanRL has benefited from the contributions of many awesome folks. I would like to cordially thank the core dev members @dosssman @yooceii @Dipamc @kinalmehta @bragajj for their efforts in helping maintain the CleanRL repository. I would also like to give a shout-out to our new contributors @cool-RR, @Howuhh, @jseppanen, @joaogui1, @ALPH2H, @ElliotMunro200, @WillDudley, and @sdpkjc.

We always welcome new contributors to the project. If you are interested in contributing to CleanRL (e.g., new features, bug fixes, new algorithms), please check out our reworked contributing guide.

New CleanRL Supported Publications

  • Md Masudur Rahman and Yexiang Xue. "Bootstrap Advantage Estimation for Policy Optimization in Reinforcement Learning." In Proceedings of the IEEE International Conference on Machine Learning and Applications (ICMLA), 2022.
  • Weng, Jiayi, Min Lin, Shengyi Huang, Bo Liu, Denys Makoviichuk, Viktor Makoviychuk, Zichen Liu et al. "Envpool: A highly parallel reinforcement learning environment execution engine." In Thirty-sixth Conference on Neural Information Processing Systems Datasets and Benchmarks Track.
  • Huang, Shengyi, Rousslan Fernand Julien Dossa, Antonin Raffin, Anssi Kanervisto, and Weixun Wang. "The 37 Implementation Details of Proximal Policy Optimization." International Conference on Learning Representations 2022 Blog Post Track,
  • Huang, Shengyi, and Santiago Ontañón. "A closer look at invalid action masking in policy gradient algorithms." The International FLAIRS Conference Proceedings, 35.
  • Schmidt, Dominik, and Thomas Schmied. "Fast and Data-Efficient Training of Rainbow: an Experimental Study on Atari." Deep Reinforcement Learning Workshop at the 35th Conference on Neural Information Processing Systems,