July 24, 2019

1023 words 5 mins read

araffin/rl-baselines-zoo

araffin/rl-baselines-zoo

A collection of 100+ pre-trained RL agents using Stable Baselines, training and hyperparameter optimization included.

repo name araffin/rl-baselines-zoo
repo link https://github.com/araffin/rl-baselines-zoo
homepage https://stable-baselines.readthedocs.io/
language Python
size (curr.) 384661 kB
stars (curr.) 521
created 2018-10-31
license MIT License

Build Status

RL Baselines Zoo: a Collection of Pre-Trained Reinforcement Learning Agents

A collection of trained Reinforcement Learning (RL) agents, with tuned hyperparameters, using Stable Baselines.

We are looking for contributors to complete the collection!

Goals of this repository:

  1. Provide a simple interface to train and enjoy RL agents
  2. Benchmark the different Reinforcement Learning algorithms
  3. Provide tuned hyperparameters for each environment and RL algorithm
  4. Have fun with the trained agents!

Enjoy a Trained Agent

If the trained agent exists, then you can see it in action using:

python enjoy.py --algo algo_name --env env_id

For example, enjoy A2C on Breakout during 5000 timesteps:

python enjoy.py --algo a2c --env BreakoutNoFrameskip-v4 --folder trained_agents/ -n 5000

If you have trained an agent yourself, you need to do:

# exp-id 0 corresponds to the last experiment, otherwise, you can specify another ID
python enjoy.py --algo algo_name --env env_id -f logs/ --exp-id 0

Train an Agent

The hyperparameters for each environment are defined in hyperparameters/algo_name.yml.

If the environment exists in this file, then you can train an agent using:

python train.py --algo algo_name --env env_id

For example (with tensorboard support):

python train.py --algo ppo2 --env CartPole-v1 --tensorboard-log /tmp/stable-baselines/

Continue training (here, load pretrained agent for Breakout and continue training for 5000 steps):

python train.py --algo a2c --env BreakoutNoFrameskip-v4 -i trained_agents/a2c/BreakoutNoFrameskip-v4.pkl -n 5000

Note: when training TRPO, you have to use mpirun to enable multiprocessing:

mpirun -n 16 python train.py --algo trpo --env BreakoutNoFrameskip-v4

Hyperparameter Tuning

We use Optuna for optimizing the hyperparameters.

Note: hyperparameters search is not implemented for ACER and DQN for now. when using SuccessiveHalvingPruner (“halving”), you must specify --n-jobs > 1

Budget of 1000 trials with a maximum of 50000 steps:

python train.py --algo ppo2 --env MountainCar-v0 -n 50000 -optimize --n-trials 1000 --n-jobs 2 \
  --sampler tpe --pruner median

Record a Video of a Trained Agent

Record 1000 steps:

python -m utils.record_video --algo ppo2 --env BipedalWalkerHardcore-v2 -n 1000

Current Collection: 120+ Trained Agents!

Scores can be found in benchmark.md. To compute them, simply run python -m utils.benchmark.

Atari Games

7 atari games from OpenAI benchmark (NoFrameskip-v4 versions).

RL Algo BeamRider Breakout Enduro Pong Qbert Seaquest SpaceInvaders
A2C :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark:
ACER :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark:
ACKTR :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark:
PPO2 :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark:
DQN :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark:
TRPO

Additional Atari Games (to be completed):

RL Algo MsPacman
A2C :heavy_check_mark:
ACER :heavy_check_mark:
ACKTR :heavy_check_mark:
PPO2 :heavy_check_mark:
DQN :heavy_check_mark:

Classic Control Environments

RL Algo CartPole-v1 MountainCar-v0 Acrobot-v1 Pendulum-v0 MountainCarContinuous-v0
A2C :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark:
ACER :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: N/A N/A
ACKTR :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark:
PPO2 :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark:
DQN :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: N/A N/A
DDPG N/A N/A N/A :heavy_check_mark: :heavy_check_mark:
SAC N/A N/A N/A :heavy_check_mark: :heavy_check_mark:
TD3 N/A N/A N/A :heavy_check_mark: :heavy_check_mark:
TRPO :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark:

Box2D Environments

RL Algo BipedalWalker-v2 LunarLander-v2 LunarLanderContinuous-v2 BipedalWalkerHardcore-v2 CarRacing-v0
A2C :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark:
ACER N/A :heavy_check_mark: N/A N/A N/A
ACKTR :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark:
PPO2 :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark:
DQN N/A :heavy_check_mark: N/A N/A N/A
DDPG :heavy_check_mark: N/A :heavy_check_mark:
SAC :heavy_check_mark: N/A :heavy_check_mark: :heavy_check_mark:
TD3 :heavy_check_mark: N/A :heavy_check_mark:
TRPO :heavy_check_mark: :heavy_check_mark: :heavy_check_mark:

PyBullet Environments

See https://github.com/bulletphysics/bullet3/tree/master/examples/pybullet/gym/pybullet_envs. Similar to MuJoCo Envs but with a free simulator: pybullet. We are using BulletEnv-v0 version.

Note: those environments are derived from Roboschool and are much harder than the Mujoco version (see Pybullet issue)

RL Algo Walker2D HalfCheetah Ant Reacher Hopper Humanoid
A2C :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark:
ACKTR :heavy_check_mark:
PPO2 :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark:
DDPG :heavy_check_mark: :heavy_check_mark: :heavy_check_mark:
SAC :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark:
TD3 :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark:
TRPO :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark:

PyBullet Envs (Continued)

RL Algo Minitaur MinitaurDuck InvertedDoublePendulum InvertedPendulumSwingup
A2C
ACKTR
PPO2 :heavy_check_mark: :heavy_check_mark: :heavy_check_mark: :heavy_check_mark:
DDPG
SAC :heavy_check_mark: :heavy_check_mark:
TD3 :heavy_check_mark: :heavy_check_mark:
TRPO

MiniGrid Envs

See https://github.com/maximecb/gym-minigrid A simple, lightweight and fast Gym environments implementation of the famous gridworld.

RL Algo Empty FourRooms DoorKey MultiRoom Fetch
A2C
PPO2 :heavy_check_mark: :heavy_check_mark:
DDPG
SAC
TRPO

There are 19 environment groups (variations for each) in total.

Note that you need to specify –gym-packages gym_minigrid with enjoy.py and train.py as it is not a standard Gym environment, as well as installing the custom Gym package module or putting it in python path.

pip install gym-minigrid
python train.py --algo ppo2 --env MiniGrid-DoorKey-5x5-v0 --gym-packages gym_minigrid

This does the same thing as:

import gym_minigrid

Also, you may need to specify a Gym environment wrapper in hyperparameters, as MiniGrid environments have Dict observation space, which is not supported by StableBaselines for now.

MiniGrid-DoorKey-5x5-v0:
  env_wrapper: gym_minigrid.wrappers.FlatObsWrapper

Env Wrappers

You can specify in the hyperparameter config one or more wrapper to use around the environment:

for one wrapper:

env_wrapper: gym_minigrid.wrappers.FlatObsWrapper

for multiple, specify a list:

env_wrapper:
    - utils.wrappers.DoneOnSuccessWrapper:
        reward_offset: 1.0
    - utils.wrappers.TimeFeatureWrapper

Note that you can easily specify parameters too.

Overwrite hyperparameters

You can easily overwrite hyperparameters in the command line, using --hyperparams:

python train.py --algo a2c --env MountainCarContinuous-v0 --hyperparams learning_rate:0.001 policy_kwargs:"dict(net_arch=[64, 64])"

Colab Notebook: Try it Online!

You can train agents online using colab notebook.

Installation

Stable-Baselines PyPi Package

Min version: stable-baselines[mpi] >= 2.10.0

apt-get install swig cmake libopenmpi-dev zlib1g-dev ffmpeg
pip install stable-baselines[mpi] box2d box2d-kengz pyyaml pybullet optuna pytablewriter scikit-optimize

Please see Stable Baselines README for alternatives.

Docker Images

Build docker image (CPU):

./scripts/build_docker.sh

GPU:

USE_GPU=True ./scripts/build_docker.sh

Pull built docker image (CPU):

docker pull stablebaselines/rl-baselines-zoo-cpu

GPU image:

docker pull stablebaselines/rl-baselines-zoo

Run script in the docker image:

./scripts/run_docker_cpu.sh python train.py --algo ppo2 --env CartPole-v1

Tests

To run tests, first install pytest, then:

python -m pytest -v tests/

Citing the Project

To cite this repository in publications:

@misc{rl-zoo,
  author = {Raffin, Antonin},
  title = {RL Baselines Zoo},
  year = {2018},
  publisher = {GitHub},
  journal = {GitHub repository},
  howpublished = {\url{https://github.com/araffin/rl-baselines-zoo}},
}

Contributing

If you trained an agent that is not present in the rl zoo, please submit a Pull Request (containing the hyperparameters and the score too).

Contributors

We would like to thanks our contributors: @iandanforth, @tatsubori @Shade5

comments powered by Disqus