Welcome to Stable Baselines docs! - RL Baselines Made Easy¶
Stable Baselines is a set of improved implementations of Reinforcement Learning (RL) algorithms based on OpenAI Baselines.
Github repository: https://github.com/hill-a/stable-baselines
RL Baselines Zoo (collection of pre-trained agents): https://github.com/araffin/rl-baselines-zoo
RL Baselines zoo also offers a simple interface to train, evaluate agents and do hyperparameter tuning.
You can read a detailed presentation of Stable Baselines in the Medium article: link
Main differences with OpenAI Baselines¶
This toolset is a fork of OpenAI Baselines, with a major structural refactoring, and code cleanups:
- Unified structure for all algorithms
- PEP8 compliant (unified code style)
- Documented functions and classes
- More tests & more code coverage
- Additional algorithms: SAC and TD3 (+ HER support for DQN, DDPG, SAC and TD3)
Installation¶
Prerequisites¶
Baselines requires python3 (>=3.5) with the development headers. You’ll also need system packages CMake, OpenMPI and zlib. Those can be installed as follows
Note
Stable-Baselines supports Tensorflow versions from 1.8.0 to 1.14.0, and does not work on Tensorflow versions 2.0.0 and above. Support for Tensorflow 2 API is planned.
Ubuntu¶
sudo apt-get update && sudo apt-get install cmake libopenmpi-dev python3-dev zlib1g-dev
Mac OS X¶
Installation of system packages on Mac requires Homebrew. With Homebrew installed, run the following:
brew install cmake openmpi
Windows 10¶
We recommend using Anaconda for Windows users for easier installation of Python packages and required libraries. You need an environment with Python version 3.5 or above.
For a quick start you can move straight to installing Stable-Baselines in the next step (without MPI). This supports most but not all algorithms.
To support all algorithms, Install MPI for Windows (you need to download and install msmpisetup.exe) and follow the instructions on how to install Stable-Baselines with MPI support in following section.
Note
Trying to create Atari environments may result to vague errors related to missing DLL files and modules. This is an issue with atari-py package. See this discussion for more information.
Stable Release¶
To install with support for all algorithms, including those depending on OpenMPI, execute:
pip install stable-baselines[mpi]
GAIL, DDPG, TRPO, and PPO1 parallelize training using OpenMPI. OpenMPI has had weird interactions with Tensorflow in the past (see Issue #430) and so if you do not intend to use these algorithms we recommend installing without OpenMPI. To do this, execute:
pip install stable-baselines
If you have already installed with MPI support, you can disable MPI by uninstalling mpi4py
with pip uninstall mpi4py.
Bleeding-edge version¶
With support for running tests and building the documentation.
git clone https://github.com/hill-a/stable-baselines && cd stable-baselines
pip install -e .[docs,tests]
Using Docker Images¶
If you are looking for docker images with stable-baselines already installed in it, we recommend using images from RL Baselines Zoo.
Otherwise, the following images contained all the dependencies for stable-baselines but not the stable-baselines package itself. They are made for development.
Use Built Images¶
GPU image (requires nvidia-docker):
docker pull stablebaselines/stable-baselines
CPU only:
docker pull stablebaselines/stable-baselines-cpu
Build the Docker Images¶
Build GPU image (with nvidia-docker):
make docker-gpu
Build CPU image:
make docker-cpu
Note: if you are using a proxy, you need to pass extra params during build and do some tweaks:
--network=host --build-arg HTTP_PROXY=http://your.proxy.fr:8080/ --build-arg http_proxy=http://your.proxy.fr:8080/ --build-arg HTTPS_PROXY=https://your.proxy.fr:8080/ --build-arg https_proxy=https://your.proxy.fr:8080/
Run the images (CPU/GPU)¶
Run the nvidia-docker GPU image
docker run -it --runtime=nvidia --rm --network host --ipc=host --name test --mount src="$(pwd)",target=/root/code/stable-baselines,type=bind stablebaselines/stable-baselines bash -c 'cd /root/code/stable-baselines/ && pytest tests/'
Or, with the shell file:
./scripts/run_docker_gpu.sh pytest tests/
Run the docker CPU image
docker run -it --rm --network host --ipc=host --name test --mount src="$(pwd)",target=/root/code/stable-baselines,type=bind stablebaselines/stable-baselines-cpu bash -c 'cd /root/code/stable-baselines/ && pytest tests/'
Or, with the shell file:
./scripts/run_docker_cpu.sh pytest tests/
Explanation of the docker command:
docker run -itcreate an instance of an image (=container), and run it interactively (so ctrl+c will work)--rmoption means to remove the container once it exits/stops (otherwise, you will have to usedocker rm)--network hostdon’t use network isolation, this allow to use tensorboard/visdom on host machine--ipc=hostUse the host system’s IPC namespace. IPC (POSIX/SysV IPC) namespace provides separation of named shared memory segments, semaphores and message queues.--name testgive explicitly the nametestto the container, otherwise it will be assigned a random name--mount src=...give access of the local directory (pwdcommand) to the container (it will be map to/root/code/stable-baselines), so all the logs created in the container in this folder will be keptbash -c '...'Run command inside the docker image, here run the tests (pytest tests/)
Getting Started¶
Most of the library tries to follow a sklearn-like syntax for the Reinforcement Learning algorithms.
Here is a quick example of how to train and run PPO2 on a cartpole environment:
import gym
from stable_baselines.common.policies import MlpPolicy
from stable_baselines.common.vec_env import DummyVecEnv
from stable_baselines import PPO2
env = gym.make('CartPole-v1')
# Optional: PPO2 requires a vectorized environment to run
# the env is now wrapped automatically when passing it to the constructor
# env = DummyVecEnv([lambda: env])
model = PPO2(MlpPolicy, env, verbose=1)
model.learn(total_timesteps=10000)
obs = env.reset()
for i in range(1000):
action, _states = model.predict(obs)
obs, rewards, dones, info = env.step(action)
env.render()
Or just train a model with a one liner if the environment is registered in Gym and if the policy is registered:
from stable_baselines import PPO2
model = PPO2('MlpPolicy', 'CartPole-v1').learn(10000)
Define and train a RL agent in one line of code!
Reinforcement Learning Tips and Tricks¶
The aim of this section is to help you doing reinforcement learning experiments. It covers general advice about RL (where to start, which algorithm to choose, how to evaluate an algorithm, …), as well as tips and tricks when using a custom environment or implementing an RL algorithm.
General advice when using Reinforcement Learning¶
TL;DR¶
- Read about RL and Stable Baselines
- Do quantitative experiments and hyperparameter tuning if needed
- Evaluate the performance using a separate test environment
- For better performance, increase the training budget
Like any other subject, if you want to work with RL, you should first read about it (we have a dedicated ressource page to get you started) to understand what you are using. We also recommend you read Stable Baselines (SB) documentation and do the tutorial. It covers basic usage and guide you towards more advanced concepts of the library (e.g. callbacks and wrappers).
Reinforcement Learning differs from other machine learning methods in several ways. The data used to train the agent is collected through interactions with the environment by the agent itself (compared to supervised learning where you have a fixed dataset for instance). This dependence can lead to vicious circle: if the agent collects poor quality data (e.g., trajectories with no rewards), then it will not improve and continue to amass bad trajectories.
This factor, among others, explains that results in RL may vary from one run to another (i.e., when only the seed of the pseudo-random generator changes). For this reason, you should always do several runs to have quantitative results.
Good results in RL are generally dependent on finding appropriate hyperparameters. Recent algorithms (PPO, SAC, TD3) normally require little hyperparameter tuning, however, don’t expect the default ones to work on any environment.
Therefore, we highly recommend you to take a look at the RL zoo (or the original papers) for tuned hyperparameters. A best practice when you apply RL to a new problem is to do automatic hyperparameter optimization. Again, this is included in the RL zoo.
When applying RL to a custom problem, you should always normalize the input to the agent (e.g. using VecNormalize for PPO2/A2C) and look at common preprocessing done on other environments (e.g. for Atari, frame-stack, …). Please refer to Tips and Tricks when creating a custom environment paragraph below for more advice related to custom environments.
Current Limitations of RL¶
You have to be aware of the current limitations of reinforcement learning.
Model-free RL algorithms (i.e. all the algorithms implemented in SB) are usually sample inefficient. They require a lot of samples (sometimes millions of interactions) to learn something useful. That’s why most of the successes in RL were achieved on games or in simulation only. For instance, in this work by ETH Zurich, the ANYmal robot was trained in simulation only, and then tested in the real world.
As a general advice, to obtain better performances, you should augment the budget of the agent (number of training timesteps).
In order to achieve the desired behavior, expert knowledge is often required to design an adequate reward function. This reward engineering (or RewArt as coined by Freek Stulp), necessitates several iterations. As a good example of reward shaping, you can take a look at Deep Mimic paper which combines imitation learning and reinforcement learning to do acrobatic moves.
One last limitation of RL is the instability of training. That is to say, you can observe during training a huge drop in performance.
This behavior is particularly present in DDPG, that’s why its extension TD3 tries to tackle that issue.
Other method, like TRPO or PPO make use of a trust region to minimize that problem by avoiding too large update.
How to evaluate an RL algorithm?¶
Because most algorithms use exploration noise during training, you need a separate test environment to evaluate the performance
of your agent at a given time. It is recommended to periodically evaluate your agent for n test episodes (n is usually between 5 and 20)
and average the reward per episode to have a good estimate.
As some policy are stochastic by default (e.g. A2C or PPO), you should also try to set deterministic=True when calling the .predict() method, this frequently leads to better performance. Looking at the training curve (episode reward function of the timesteps) is a good proxy but underestimates the agent true performance.
Note
We provide an EvalCallback for doing such evaluation. You can read more about it in the Callbacks section.
We suggest you reading Deep Reinforcement Learning that Matters for a good discussion about RL evaluation.
You can also take a look at this blog post and this issue by Cédric Colas.
Which algorithm should I use?¶
There is no silver bullet in RL, depending on your needs and problem, you may choose one or the other. The first distinction comes from your action space, i.e., do you have discrete (e.g. LEFT, RIGHT, …) or continuous actions (ex: go to a certain speed)?
Some algorithms are only tailored for one or the other domain: DQN only supports discrete actions, where SAC is restricted to continuous actions.
The second difference that will help you choose is whether you can parallelize your training or not, and how you can do it (with or without MPI?).
If what matters is the wall clock training time, then you should lean towards A2C and its derivatives (PPO, ACER, ACKTR, …).
Take a look at the Vectorized Environments to learn more about training with multiple workers.
To sum it up:
Discrete Actions¶
Note
This covers Discrete, MultiDiscrete, Binary and MultiBinary spaces
Discrete Actions - Single Process¶
DQN with extensions (double DQN, prioritized replay, …) and ACER are the recommended algorithms. DQN is usually slower to train (regarding wall clock time) but is the most sample efficient (because of its replay buffer).
Discrete Actions - Multiprocessed¶
You should give a try to PPO2, A2C and its successors (ACKTR, ACER).
If you can multiprocess the training using MPI, then you should checkout PPO1 and TRPO.
Continuous Actions¶
Continuous Actions - Single Process¶
Current State Of The Art (SOTA) algorithms are SAC and TD3.
Please use the hyperparameters in the RL zoo for best results.
Continuous Actions - Multiprocessed¶
Take a look at PPO2, TRPO or A2C. Again, don’t forget to take the hyperparameters from the RL zoo for continuous actions problems (cf Bullet envs).
Note
Normalization is critical for those algorithms
If you can use MPI, then you can choose between PPO1, TRPO and DDPG.
Tips and Tricks when creating a custom environment¶
If you want to learn about how to create a custom environment, we recommend you read this page. We also provide a colab notebook for a concrete example of creating a custom gym environment.
Some basic advice:
- always normalize your observation space when you can, i.e., when you know the boundaries
- normalize your action space and make it symmetric when continuous (cf potential issue below) A good practice is to rescale your actions to lie in [-1, 1]. This does not limit you as you can easily rescale the action inside the environment
- start with shaped reward (i.e. informative reward) and simplified version of your problem
- debug with random actions to check that your environment works and follows the gym interface:
We provide a helper to check that your environment runs without error:
from stable_baselines.common.env_checker import check_env
env = CustomEnv(arg1, ...)
# It will check your custom environment and output additional warnings if needed
check_env(env)
If you want to quickly try a random agent on your environment, you can also do:
env = YourEnv()
obs = env.reset()
n_steps = 10
for _ in range(n_steps):
# Random action
action = env.action_space.sample()
obs, reward, done, info = env.step(action)
Why should I normalize the action space?
Most reinforcement learning algorithms rely on a Gaussian distribution (initially centered at 0 with std 1) for continuous actions. So, if you forget to normalize the action space when using a custom environment, this can harm learning and be difficult to debug (cf attached image and issue #473).
Another consequence of using a Gaussian is that the action range is not bounded.
That’s why clipping is usually used as a bandage to stay in a valid interval.
A better solution would be to use a squashing function (cf SAC) or a Beta distribution (cf issue #112).
Note
This statement is not true for DDPG or TD3 because they don’t rely on any probability distribution.
Tips and Tricks when implementing an RL algorithm¶
When you try to reproduce a RL paper by implementing the algorithm, the nuts and bolts of RL research by John Schulman are quite useful (video).
We recommend following those steps to have a working RL algorithm:
- Read the original paper several times
- Read existing implementations (if available)
- Try to have some “sign of life” on toy problems
- Validate the implementation by making it run on harder and harder envs (you can compare results against the RL zoo)
- You usually need to run hyperparameter optimization for that step.
You need to be particularly careful on the shape of the different objects you are manipulating (a broadcast mistake will fail silently cf issue #75) and when to stop the gradient propagation.
A personal pick (by @araffin) for environments with gradual difficulty in RL with continuous actions:
- Pendulum (easy to solve)
- HalfCheetahBullet (medium difficulty with local minima and shaped reward)
- BipedalWalkerHardcore (if it works on that one, then you can have a cookie)
in RL with discrete actions:
- CartPole-v1 (easy to be better than random agent, harder to achieve maximal performance)
- LunarLander
- Pong (one of the easiest Atari game)
- other Atari games (e.g. Breakout)
Reinforcement Learning Resources¶
Stable-Baselines assumes that you already understand the basic concepts of Reinforcement Learning (RL).
However, if you want to learn about RL, there are several good resources to get started:
RL Algorithms¶
This table displays the rl algorithms that are implemented in the stable baselines project, along with some useful characteristics: support for recurrent policies, discrete/continuous actions, multiprocessing.
| Name | Refactored [1] | Recurrent | Box |
Discrete |
Multi Processing |
|---|---|---|---|---|---|
| A2C | ✔️ | ✔️ | ✔️ | ✔️ | ✔️ |
| ACER | ✔️ | ✔️ | ❌ [4] | ✔️ | ✔️ |
| ACKTR | ✔️ | ✔️ | ✔️ | ✔️ | ✔️ |
| DDPG | ✔️ | ❌ | ✔️ | ❌ | ✔️ [3] |
| DQN | ✔️ | ❌ | ❌ | ✔️ | ❌ |
| HER | ✔️ | ❌ | ✔️ | ✔️ | ❌ |
| GAIL [2] | ✔️ | ✔️ | ✔️ | ✔️ | ✔️ [3] |
| PPO1 | ✔️ | ❌ | ✔️ | ✔️ | ✔️ [3] |
| PPO2 | ✔️ | ✔️ | ✔️ | ✔️ | ✔️ |
| SAC | ✔️ | ❌ | ✔️ | ❌ | ❌ |
| TD3 | ✔️ | ❌ | ✔️ | ❌ | ❌ |
| TRPO | ✔️ | ❌ | ✔️ | ✔ | ✔️ [3] |
| [1] | Whether or not the algorithm has be refactored to fit the BaseRLModel class. |
| [2] | Only implemented for TRPO. |
| [3] | (1, 2, 3, 4) Multi Processing with MPI. |
| [4] | TODO, in project scope. |
Note
Non-array spaces such as Dict or Tuple are not currently supported by any algorithm,
except HER for dict when working with gym.GoalEnv
Actions gym.spaces:
Box: A N-dimensional box that contains every point in the action space.Discrete: A list of possible actions, where each timestep only one of the actions can be used.MultiDiscrete: A list of possible actions, where each timestep only one action of each discrete set can be used.MultiBinary: A list of possible actions, where each timestep any of the actions can be used in any combination.
Note
Some logging values (like ep_rewmean, eplenmean) are only available when using a Monitor wrapper
See Issue #339 for more info.
Reproducibility¶
Completely reproducible results are not guaranteed across Tensorflow releases or different platforms. Furthermore, results need not be reproducible between CPU and GPU executions, even when using identical seeds.
In order to make make computations deterministic on CPU, on your specific problem on one specific platform,
you need to pass a seed argument at the creation of a model and set n_cpu_tf_sess=1 (number of cpu for Tensorflow session).
If you pass an environment to the model using set_env(), then you also need to seed the environment first.
Note
Because of the current limits of Tensorflow 1.x, we cannot ensure reproducible results on the GPU yet. We hope to solve that issue with Tensorflow 2.x support (cf Issue #366).
Note
TD3 sometimes fail to have reproducible results for obscure reasons, even when following the previous steps (cf PR #492). If you find the reason then please open an issue ;)
Credit: part of the Reproducibility section comes from PyTorch Documentation
Examples¶
Try it online with Colab Notebooks!¶
All the following examples can be executed online using Google colab 
notebooks:
Basic Usage: Training, Saving, Loading¶
In the following example, we will train, save and load a DQN model on the Lunar Lander environment.
Lunar Lander Environment
Note
LunarLander requires the python package box2d.
You can install it using apt install swig and then pip install box2d box2d-kengz
Note
load function re-creates model from scratch on each call, which can be slow.
If you need to e.g. evaluate same model with multiple different sets of parameters, consider
using load_parameters instead.
import gym
from stable_baselines import DQN
from stable_baselines.common.evaluation import evaluate_policy
# Create environment
env = gym.make('LunarLander-v2')
# Instantiate the agent
model = DQN('MlpPolicy', env, learning_rate=1e-3, prioritized_replay=True, verbose=1)
# Train the agent
model.learn(total_timesteps=int(2e5))
# Save the agent
model.save("dqn_lunar")
del model # delete trained model to demonstrate loading
# Load the trained agent
model = DQN.load("dqn_lunar")
# Evaluate the agent
mean_reward, std_reward = evaluate_policy(model, model.get_env(), n_eval_episodes=10)
# Enjoy trained agent
obs = env.reset()
for i in range(1000):
action, _states = model.predict(obs)
obs, rewards, dones, info = env.step(action)
env.render()
Multiprocessing: Unleashing the Power of Vectorized Environments¶
CartPole Environment
import gym
import numpy as np
from stable_baselines.common.policies import MlpPolicy
from stable_baselines.common.vec_env import SubprocVecEnv
from stable_baselines.common import set_global_seeds, make_vec_env
from stable_baselines import ACKTR
def make_env(env_id, rank, seed=0):
"""
Utility function for multiprocessed env.
:param env_id: (str) the environment ID
:param num_env: (int) the number of environments you wish to have in subprocesses
:param seed: (int) the inital seed for RNG
:param rank: (int) index of the subprocess
"""
def _init():
env = gym.make(env_id)
env.seed(seed + rank)
return env
set_global_seeds(seed)
return _init
if __name__ == '__main__':
env_id = "CartPole-v1"
num_cpu = 4 # Number of processes to use
# Create the vectorized environment
env = SubprocVecEnv([make_env(env_id, i) for i in range(num_cpu)])
# Stable Baselines provides you with make_vec_env() helper
# which does exactly the previous steps for you:
# env = make_vec_env(env_id, n_envs=num_cpu, seed=0)
model = ACKTR(MlpPolicy, env, verbose=1)
model.learn(total_timesteps=25000)
obs = env.reset()
for _ in range(1000):
action, _states = model.predict(obs)
obs, rewards, dones, info = env.step(action)
env.render()
Using Callback: Monitoring Training¶
Note
We recommend reading the Callback section
You can define a custom callback function that will be called inside the agent. This could be useful when you want to monitor training, for instance display live learning curves in Tensorboard (or in Visdom) or save the best agent. If your callback returns False, training is aborted early.
Learning curve of DDPG on LunarLanderContinuous environment
import os
import gym
import numpy as np
import matplotlib.pyplot as plt
from stable_baselines import DDPG
from stable_baselines.ddpg.policies import LnMlpPolicy
from stable_baselines import results_plotter
from stable_baselines.bench import Monitor
from stable_baselines.results_plotter import load_results, ts2xy
from stable_baselines.common.noise import AdaptiveParamNoiseSpec
from stable_baselines.common.callbacks import BaseCallback
class SaveOnBestTrainingRewardCallback(BaseCallback):
"""
Callback for saving a model (the check is done every ``check_freq`` steps)
based on the training reward (in practice, we recommend using ``EvalCallback``).
:param check_freq: (int)
:param log_dir: (str) Path to the folder where the model will be saved.
It must contains the file created by the ``Monitor`` wrapper.
:param verbose: (int)
"""
def __init__(self, check_freq: int, log_dir: str, verbose=1):
super(SaveOnBestTrainingRewardCallback, self).__init__(verbose)
self.check_freq = check_freq
self.log_dir = log_dir
self.save_path = os.path.join(log_dir, 'best_model')
self.best_mean_reward = -np.inf
def _init_callback(self) -> None:
# Create folder if needed
if self.save_path is not None:
os.makedirs(self.save_path, exist_ok=True)
def _on_step(self) -> bool:
if self.n_calls % self.check_freq == 0:
# Retrieve training reward
x, y = ts2xy(load_results(self.log_dir), 'timesteps')
if len(x) > 0:
# Mean training reward over the last 100 episodes
mean_reward = np.mean(y[-100:])
if self.verbose > 0:
print("Num timesteps: {}".format(self.num_timesteps))
print("Best mean reward: {:.2f} - Last mean reward per episode: {:.2f}".format(self.best_mean_reward, mean_reward))
# New best model, you could save the agent here
if mean_reward > self.best_mean_reward:
self.best_mean_reward = mean_reward
# Example for saving best model
if self.verbose > 0:
print("Saving new best model to {}".format(self.save_path))
self.model.save(self.save_path)
return True
# Create log dir
log_dir = "tmp/"
os.makedirs(log_dir, exist_ok=True)
# Create and wrap the environment
env = gym.make('LunarLanderContinuous-v2')
env = Monitor(env, log_dir)
# Add some param noise for exploration
param_noise = AdaptiveParamNoiseSpec(initial_stddev=0.1, desired_action_stddev=0.1)
# Because we use parameter noise, we should use a MlpPolicy with layer normalization
model = DDPG(LnMlpPolicy, env, param_noise=param_noise, verbose=0)
# Create the callback: check every 1000 steps
callback = SaveOnBestTrainingRewardCallback(check_freq=1000, log_dir=log_dir)
# Train the agent
time_steps = 1e5
model.learn(total_timesteps=int(time_steps), callback=callback)
results_plotter.plot_results([log_dir], time_steps, results_plotter.X_TIMESTEPS, "DDPG LunarLander")
plt.show()
Atari Games¶
Trained A2C agent on Breakout
Pong Environment
Training a RL agent on Atari games is straightforward thanks to make_atari_env helper function.
It will do all the preprocessing
and multiprocessing for you.
from stable_baselines.common.cmd_util import make_atari_env
from stable_baselines.common.vec_env import VecFrameStack
from stable_baselines import ACER
# There already exists an environment generator
# that will make and wrap atari environments correctly.
# Here we are also multiprocessing training (num_env=4 => 4 processes)
env = make_atari_env('PongNoFrameskip-v4', num_env=4, seed=0)
# Frame-stacking with 4 frames
env = VecFrameStack(env, n_stack=4)
model = ACER('CnnPolicy', env, verbose=1)
model.learn(total_timesteps=25000)
obs = env.reset()
while True:
action, _states = model.predict(obs)
obs, rewards, dones, info = env.step(action)
env.render()
Mujoco: Normalizing input features¶
Normalizing input features may be essential to successful training of an RL agent (by default, images are scaled but not other types of input), for instance when training on Mujoco. For that, a wrapper exists and will compute a running average and standard deviation of input features (it can do the same for rewards).
Note
We cannot provide a notebook for this example because Mujoco is a proprietary engine and requires a license.
import gym
from stable_baselines.common.policies import MlpPolicy
from stable_baselines.common.vec_env import DummyVecEnv, VecNormalize
from stable_baselines import PPO2
env = DummyVecEnv([lambda: gym.make("Reacher-v2")])
# Automatically normalize the input features
env = VecNormalize(env, norm_obs=True, norm_reward=False,
clip_obs=10.)
model = PPO2(MlpPolicy, env)
model.learn(total_timesteps=2000)
# Don't forget to save the VecNormalize statistics when saving the agent
log_dir = "/tmp/"
model.save(log_dir + "ppo_reacher")
env.save(os.path.join(log_dir, "vec_normalize.pkl"))
Custom Policy Network¶
Stable baselines provides default policy networks for images (CNNPolicies) and other type of inputs (MlpPolicies). However, you can also easily define a custom architecture for the policy network (see custom policy section):
import gym
from stable_baselines.common.policies import FeedForwardPolicy
from stable_baselines.common.vec_env import DummyVecEnv
from stable_baselines import A2C
# Custom MLP policy of three layers of size 128 each
class CustomPolicy(FeedForwardPolicy):
def __init__(self, *args, **kwargs):
super(CustomPolicy, self).__init__(*args, **kwargs,
net_arch=[dict(pi=[128, 128, 128], vf=[128, 128, 128])],
feature_extraction="mlp")
model = A2C(CustomPolicy, 'LunarLander-v2', verbose=1)
# Train the agent
model.learn(total_timesteps=100000)
Accessing and modifying model parameters¶
You can access model’s parameters via load_parameters and get_parameters functions, which
use dictionaries that map variable names to NumPy arrays.
These functions are useful when you need to e.g. evaluate large set of models with same network structure, visualize different layers of the network or modify parameters manually.
You can access original Tensorflow Variables with function get_parameter_list.
Following example demonstrates reading parameters, modifying some of them and loading them to model
by implementing evolution strategy
for solving CartPole-v1 environment. The initial guess for parameters is obtained by running
A2C policy gradient updates on the model.
import gym
import numpy as np
from stable_baselines import A2C
def mutate(params):
"""Mutate parameters by adding normal noise to them"""
return dict((name, param + np.random.normal(size=param.shape))
for name, param in params.items())
def evaluate(env, model):
"""Return mean fitness (sum of episodic rewards) for given model"""
episode_rewards = []
for _ in range(10):
reward_sum = 0
done = False
obs = env.reset()
while not done:
action, _states = model.predict(obs)
obs, reward, done, info = env.step(action)
reward_sum += reward
episode_rewards.append(reward_sum)
return np.mean(episode_rewards)
# Create env
env = gym.make('CartPole-v1')
# Create policy with a small network
model = A2C('MlpPolicy', env, ent_coef=0.0, learning_rate=0.1,
policy_kwargs={'net_arch': [8, ]})
# Use traditional actor-critic policy gradient updates to
# find good initial parameters
model.learn(total_timesteps=5000)
# Get the parameters as the starting point for ES
mean_params = model.get_parameters()
# Include only variables with "/pi/" (policy) or "/shared" (shared layers)
# in their name: Only these ones affect the action.
mean_params = dict((key, value) for key, value in mean_params.items()
if ("/pi/" in key or "/shared" in key))
for iteration in range(10):
# Create population of candidates and evaluate them
population = []
for population_i in range(100):
candidate = mutate(mean_params)
# Load new policy parameters to agent.
# Tell function that it should only update parameters
# we give it (policy parameters)
model.load_parameters(candidate, exact_match=False)
fitness = evaluate(env, model)
population.append((candidate, fitness))
# Take top 10% and use average over their parameters as next mean parameter
top_candidates = sorted(population, key=lambda x: x[1], reverse=True)[:10]
mean_params = dict(
(name, np.stack([top_candidate[0][name] for top_candidate in top_candidates]).mean(0))
for name in mean_params.keys()
)
mean_fitness = sum(top_candidate[1] for top_candidate in top_candidates) / 10.0
print("Iteration {:<3} Mean top fitness: {:.2f}".format(iteration, mean_fitness))
Recurrent Policies¶
This example demonstrate how to train a recurrent policy and how to test it properly.
Warning
One current limitation of recurrent policies is that you must test them with the same number of environments they have been trained on.
from stable_baselines import PPO2
# For recurrent policies, with PPO2, the number of environments run in parallel
# should be a multiple of nminibatches.
model = PPO2('MlpLstmPolicy', 'CartPole-v1', nminibatches=1, verbose=1)
model.learn(50000)
# Retrieve the env
env = model.get_env()
obs = env.reset()
# Passing state=None to the predict function means
# it is the initial state
state = None
# When using VecEnv, done is a vector
done = [False for _ in range(env.num_envs)]
for _ in range(1000):
# We need to pass the previous state and a mask for recurrent policies
# to reset lstm state when a new episode begin
action, state = model.predict(obs, state=state, mask=done)
obs, reward , done, _ = env.step(action)
# Note: with VecEnv, env.reset() is automatically called
# Show the env
env.render()
Hindsight Experience Replay (HER)¶
For this example, we are using Highway-Env by @eleurent.
The highway-parking-v0 environment.
The parking env is a goal-conditioned continuous control task, in which the vehicle must park in a given space with the appropriate heading.
Note
the hyperparameters in the following example were optimized for that environment.
import gym
import highway_env
import numpy as np
from stable_baselines import HER, SAC, DDPG, TD3
from stable_baselines.ddpg import NormalActionNoise
env = gym.make("parking-v0")
# Create 4 artificial transitions per real transition
n_sampled_goal = 4
# SAC hyperparams:
model = HER('MlpPolicy', env, SAC, n_sampled_goal=n_sampled_goal,
goal_selection_strategy='future',
verbose=1, buffer_size=int(1e6),
learning_rate=1e-3,
gamma=0.95, batch_size=256,
policy_kwargs=dict(layers=[256, 256, 256]))
# DDPG Hyperparams:
# NOTE: it works even without action noise
# n_actions = env.action_space.shape[0]
# noise_std = 0.2
# action_noise = NormalActionNoise(mean=np.zeros(n_actions), sigma=noise_std * np.ones(n_actions))
# model = HER('MlpPolicy', env, DDPG, n_sampled_goal=n_sampled_goal,
# goal_selection_strategy='future',
# verbose=1, buffer_size=int(1e6),
# actor_lr=1e-3, critic_lr=1e-3, action_noise=action_noise,
# gamma=0.95, batch_size=256,
# policy_kwargs=dict(layers=[256, 256, 256]))
model.learn(int(2e5))
model.save('her_sac_highway')
# Load saved model
model = HER.load('her_sac_highway', env=env)
obs = env.reset()
# Evaluate the agent
episode_reward = 0
for _ in range(100):
action, _ = model.predict(obs)
obs, reward, done, info = env.step(action)
env.render()
episode_reward += reward
if done or info.get('is_success', False):
print("Reward:", episode_reward, "Success?", info.get('is_success', False))
episode_reward = 0.0
obs = env.reset()
Continual Learning¶
You can also move from learning on one environment to another for continual learning
(PPO2 on DemonAttack-v0, then transferred on SpaceInvaders-v0):
from stable_baselines.common.cmd_util import make_atari_env
from stable_baselines import PPO2
# There already exists an environment generator
# that will make and wrap atari environments correctly
env = make_atari_env('DemonAttackNoFrameskip-v4', num_env=8, seed=0)
model = PPO2('CnnPolicy', env, verbose=1)
model.learn(total_timesteps=10000)
obs = env.reset()
for i in range(1000):
action, _states = model.predict(obs)
obs, rewards, dones, info = env.step(action)
env.render()
# Close the processes
env.close()
# The number of environments must be identical when changing environments
env = make_atari_env('SpaceInvadersNoFrameskip-v4', num_env=8, seed=0)
# change env
model.set_env(env)
model.learn(total_timesteps=10000)
obs = env.reset()
while True:
action, _states = model.predict(obs)
obs, rewards, dones, info = env.step(action)
env.render()
env.close()
Record a Video¶
Record a mp4 video (here using a random agent).
Note
It requires ffmpeg or avconv to be installed on the machine.
import gym
from stable_baselines.common.vec_env import VecVideoRecorder, DummyVecEnv
env_id = 'CartPole-v1'
video_folder = 'logs/videos/'
video_length = 100
env = DummyVecEnv([lambda: gym.make(env_id)])
obs = env.reset()
# Record the video starting at the first step
env = VecVideoRecorder(env, video_folder,
record_video_trigger=lambda x: x == 0, video_length=video_length,
name_prefix="random-agent-{}".format(env_id))
env.reset()
for _ in range(video_length + 1):
action = [env.action_space.sample()]
obs, _, _, _ = env.step(action)
# Save the video
env.close()
Bonus: Make a GIF of a Trained Agent¶
Note
For Atari games, you need to use a screen recorder such as Kazam. And then convert the video using ffmpeg
import imageio
import numpy as np
from stable_baselines import A2C
model = A2C("MlpPolicy", "LunarLander-v2").learn(100000)
images = []
obs = model.env.reset()
img = model.env.render(mode='rgb_array')
for i in range(350):
images.append(img)
action, _ = model.predict(obs)
obs, _, _ ,_ = model.env.step(action)
img = model.env.render(mode='rgb_array')
imageio.mimsave('lander_a2c.gif', [np.array(img) for i, img in enumerate(images) if i%2 == 0], fps=29)
Vectorized Environments¶
Vectorized Environments are a method for stacking multiple independent environments into a single environment.
Instead of training an RL agent on 1 environment per step, it allows us to train it on n environments per step.
Because of this, actions passed to the environment are now a vector (of dimension n).
It is the same for observations, rewards and end of episode signals (dones).
In the case of non-array observation spaces such as Dict or Tuple, where different sub-spaces
may have different shapes, the sub-observations are vectors (of dimension n).
| Name | Box |
Discrete |
Dict |
Tuple |
Multi Processing |
|---|---|---|---|---|---|
| DummyVecEnv | ✔️ | ✔️ | ✔️ | ✔️ | ❌️ |
| SubprocVecEnv | ✔️ | ✔️ | ✔️ | ✔️ | ✔️ |
Note
Vectorized environments are required when using wrappers for frame-stacking or normalization.
Note
When using vectorized environments, the environments are automatically reset at the end of each episode.
Thus, the observation returned for the i-th environment when done[i] is true will in fact be the first observation of the next episode, not the last observation of the episode that has just terminated.
You can access the “real” final observation of the terminated episode—that is, the one that accompanied the done event provided by the underlying environment—using the terminal_observation keys in the info dicts returned by the vecenv.
Warning
When using SubprocVecEnv, users must wrap the code in an if __name__ == "__main__": if using the forkserver or spawn start method (default on Windows).
On Linux, the default start method is fork which is not thread safe and can create deadlocks.
For more information, see Python’s multiprocessing guidelines.
VecEnv¶
-
class
stable_baselines.common.vec_env.VecEnv(num_envs, observation_space, action_space)[source]¶ An abstract asynchronous, vectorized environment.
Parameters: - num_envs – (int) the number of environments
- observation_space – (Gym Space) the observation space
- action_space – (Gym Space) the action space
-
env_method(method_name, *method_args, indices=None, **method_kwargs)[source]¶ Call instance methods of vectorized environments.
Parameters: - method_name – (str) The name of the environment method to invoke.
- indices – (list,int) Indices of envs whose method to call
- method_args – (tuple) Any positional arguments to provide in the call
- method_kwargs – (dict) Any keyword arguments to provide in the call
Returns: (list) List of items returned by the environment’s method call
-
get_attr(attr_name, indices=None)[source]¶ Return attribute from vectorized environment.
Parameters: - attr_name – (str) The name of the attribute whose value to return
- indices – (list,int) Indices of envs to get attribute from
Returns: (list) List of values of ‘attr_name’ in all environments
-
get_images(*args, **kwargs) → Sequence[numpy.ndarray][source]¶ Return RGB images from each environment
-
getattr_depth_check(name, already_found)[source]¶ Check if an attribute reference is being hidden in a recursive call to __getattr__
Parameters: - name – (str) name of attribute to check for
- already_found – (bool) whether this attribute has already been found in a wrapper
Returns: (str or None) name of module whose attribute is being shadowed, if any.
-
render(mode: str, *args, **kwargs)[source]¶ Gym environment rendering
Parameters: mode – the rendering type
-
reset()[source]¶ Reset all the environments and return an array of observations, or a tuple of observation arrays.
If step_async is still doing work, that work will be cancelled and step_wait() should not be called until step_async() is invoked again.
Returns: ([int] or [float]) observation
-
seed(seed: Optional[int] = None) → List[Union[None, int]][source]¶ Sets the random seeds for all environments, based on a given seed. Each individual environment will still get its own seed, by incrementing the given seed.
Parameters: seed – (Optional[int]) The random seed. May be None for completely random seeding. Returns: (List[Union[None, int]]) Returns a list containing the seeds for each individual env. Note that all list elements may be None, if the env does not return anything when being seeded.
-
set_attr(attr_name, value, indices=None)[source]¶ Set attribute inside vectorized environments.
Parameters: - attr_name – (str) The name of attribute to assign new value
- value – (obj) Value to assign to attr_name
- indices – (list,int) Indices of envs to assign value
Returns: (NoneType)
-
step(actions)[source]¶ Step the environments with the given action
Parameters: actions – ([int] or [float]) the action Returns: ([int] or [float], [float], [bool], dict) observation, reward, done, information
DummyVecEnv¶
-
class
stable_baselines.common.vec_env.DummyVecEnv(env_fns)[source]¶ Creates a simple vectorized wrapper for multiple environments, calling each environment in sequence on the current Python process. This is useful for computationally simple environment such as
cartpole-v1, as the overhead of multiprocess or multithread outweighs the environment computation time. This can also be used for RL methods that require a vectorized environment, but that you want a single environments to train with.Parameters: env_fns – ([callable]) A list of functions that will create the environments (each callable returns a Gym.Env instance when called). -
env_method(method_name, *method_args, indices=None, **method_kwargs)[source]¶ Call instance methods of vectorized environments.
-
get_attr(attr_name, indices=None)[source]¶ Return attribute from vectorized environment (see base class).
-
get_images(*args, **kwargs) → Sequence[numpy.ndarray][source]¶ Return RGB images from each environment
-
render(*args, **kwargs)[source]¶ Gym environment rendering. If there are multiple environments then they are tiled together in one image via BaseVecEnv.render(). Otherwise (if self.num_envs == 1), we pass the render call directly to the underlying environment.
Therefore, some arguments such as mode will have values that are valid only when num_envs == 1.
Parameters: mode – The rendering type.
-
reset()[source]¶ Reset all the environments and return an array of observations, or a tuple of observation arrays.
If step_async is still doing work, that work will be cancelled and step_wait() should not be called until step_async() is invoked again.
Returns: ([int] or [float]) observation
-
seed(seed=None)[source]¶ Sets the random seeds for all environments, based on a given seed. Each individual environment will still get its own seed, by incrementing the given seed.
Parameters: seed – (Optional[int]) The random seed. May be None for completely random seeding. Returns: (List[Union[None, int]]) Returns a list containing the seeds for each individual env. Note that all list elements may be None, if the env does not return anything when being seeded.
-
set_attr(attr_name, value, indices=None)[source]¶ Set attribute inside vectorized environments (see base class).
-
SubprocVecEnv¶
-
class
stable_baselines.common.vec_env.SubprocVecEnv(env_fns, start_method=None)[source]¶ Creates a multiprocess vectorized wrapper for multiple environments, distributing each environment to its own process, allowing significant speed up when the environment is computationally complex.
For performance reasons, if your environment is not IO bound, the number of environments should not exceed the number of logical cores on your CPU.
Warning
Only ‘forkserver’ and ‘spawn’ start methods are thread-safe, which is important when TensorFlow sessions or other non thread-safe libraries are used in the parent (see issue #217). However, compared to ‘fork’ they incur a small start-up cost and have restrictions on global variables. With those methods, users must wrap the code in an
if __name__ == "__main__":block. For more information, see the multiprocessing documentation.Parameters: - env_fns – ([callable]) A list of functions that will create the environments (each callable returns a Gym.Env instance when called).
- start_method – (str) method used to start the subprocesses. Must be one of the methods returned by multiprocessing.get_all_start_methods(). Defaults to ‘forkserver’ on available platforms, and ‘spawn’ otherwise.
-
env_method(method_name, *method_args, indices=None, **method_kwargs)[source]¶ Call instance methods of vectorized environments.
-
get_attr(attr_name, indices=None)[source]¶ Return attribute from vectorized environment (see base class).
-
get_images(*args, **kwargs) → Sequence[numpy.ndarray][source]¶ Return RGB images from each environment
-
reset()[source]¶ Reset all the environments and return an array of observations, or a tuple of observation arrays.
If step_async is still doing work, that work will be cancelled and step_wait() should not be called until step_async() is invoked again.
Returns: ([int] or [float]) observation
-
seed(seed=None)[source]¶ Sets the random seeds for all environments, based on a given seed. Each individual environment will still get its own seed, by incrementing the given seed.
Parameters: seed – (Optional[int]) The random seed. May be None for completely random seeding. Returns: (List[Union[None, int]]) Returns a list containing the seeds for each individual env. Note that all list elements may be None, if the env does not return anything when being seeded.
-
set_attr(attr_name, value, indices=None)[source]¶ Set attribute inside vectorized environments (see base class).
Wrappers¶
VecFrameStack¶
VecNormalize¶
-
class
stable_baselines.common.vec_env.VecNormalize(venv, training=True, norm_obs=True, norm_reward=True, clip_obs=10.0, clip_reward=10.0, gamma=0.99, epsilon=1e-08)[source]¶ A moving average, normalizing wrapper for vectorized environment.
It is pickleable which will save moving averages and configuration parameters. The wrapped environment venv is not saved, and must be restored manually with set_venv after being unpickled.
Parameters: - venv – (VecEnv) the vectorized environment to wrap
- training – (bool) Whether to update or not the moving average
- norm_obs – (bool) Whether to normalize observation or not (default: True)
- norm_reward – (bool) Whether to normalize rewards or not (default: True)
- clip_obs – (float) Max absolute value for observation
- clip_reward – (float) Max value absolute for discounted reward
- gamma – (float) discount factor
- epsilon – (float) To avoid division by zero
-
get_original_obs() → numpy.ndarray[source]¶ Returns an unnormalized version of the observations from the most recent step or reset.
-
get_original_reward() → numpy.ndarray[source]¶ Returns an unnormalized version of the rewards from the most recent step.
-
static
load(load_path, venv)[source]¶ Loads a saved VecNormalize object.
Parameters: - load_path – the path to load from.
- venv – the VecEnv to wrap.
Returns: (VecNormalize)
-
load_running_average(path)[source]¶ Parameters: path – (str) path to log dir Deprecated since version 2.9.0: This function will be removed in a future version
-
normalize_obs(obs: numpy.ndarray) → numpy.ndarray[source]¶ Normalize observations using this VecNormalize’s observations statistics. Calling this method does not update statistics.
-
normalize_reward(reward: numpy.ndarray) → numpy.ndarray[source]¶ Normalize rewards using this VecNormalize’s rewards statistics. Calling this method does not update statistics.
-
save_running_average(path)[source]¶ Parameters: path – (str) path to log dir Deprecated since version 2.9.0: This function will be removed in a future version
VecVideoRecorder¶
-
class
stable_baselines.common.vec_env.VecVideoRecorder(venv, video_folder, record_video_trigger, video_length=200, name_prefix='rl-video')[source]¶ Wraps a VecEnv or VecEnvWrapper object to record rendered image as mp4 video. It requires ffmpeg or avconv to be installed on the machine.
Parameters: - venv – (VecEnv or VecEnvWrapper)
- video_folder – (str) Where to save videos
- record_video_trigger – (func) Function that defines when to start recording. The function takes the current number of step, and returns whether we should start recording or not.
- video_length – (int) Length of recorded videos
- name_prefix – (str) Prefix to the video name
VecCheckNan¶
-
class
stable_baselines.common.vec_env.VecCheckNan(venv, raise_exception=False, warn_once=True, check_inf=True)[source]¶ NaN and inf checking wrapper for vectorized environment, will raise a warning by default, allowing you to know from what the NaN of inf originated from.
Parameters: - venv – (VecEnv) the vectorized environment to wrap
- raise_exception – (bool) Whether or not to raise a ValueError, instead of a UserWarning
- warn_once – (bool) Whether or not to only warn once.
- check_inf – (bool) Whether or not to check for +inf or -inf as well
-
reset()[source]¶ Reset all the environments and return an array of observations, or a tuple of observation arrays.
If step_async is still doing work, that work will be cancelled and step_wait() should not be called until step_async() is invoked again.
Returns: ([int] or [float]) observation
Using Custom Environments¶
To use the rl baselines with custom environments, they just need to follow the gym interface. That is to say, your environment must implement the following methods (and inherits from OpenAI Gym Class):
Note
If you are using images as input, the input values must be in [0, 255] as the observation is normalized (dividing by 255 to have values in [0, 1]) when using CNN policies.
import gym
from gym import spaces
class CustomEnv(gym.Env):
"""Custom Environment that follows gym interface"""
metadata = {'render.modes': ['human']}
def __init__(self, arg1, arg2, ...):
super(CustomEnv, self).__init__()
# Define action and observation space
# They must be gym.spaces objects
# Example when using discrete actions:
self.action_space = spaces.Discrete(N_DISCRETE_ACTIONS)
# Example for using image as input:
self.observation_space = spaces.Box(low=0, high=255,
shape=(HEIGHT, WIDTH, N_CHANNELS), dtype=np.uint8)
def step(self, action):
...
return observation, reward, done, info
def reset(self):
...
return observation # reward, done, info can't be included
def render(self, mode='human'):
...
def close (self):
...
Then you can define and train a RL agent with:
# Instantiate the env
env = CustomEnv(arg1, ...)
# Define and Train the agent
model = A2C('CnnPolicy', env).learn(total_timesteps=1000)
To check that your environment follows the gym interface, please use:
from stable_baselines.common.env_checker import check_env
env = CustomEnv(arg1, ...)
# It will check your custom environment and output additional warnings if needed
check_env(env)
We have created a colab notebook for a concrete example of creating a custom environment.
You can also find a complete guide online on creating a custom Gym environment.
Optionally, you can also register the environment with gym,
that will allow you to create the RL agent in one line (and use gym.make() to instantiate the env).
In the project, for testing purposes, we use a custom environment named IdentityEnv
defined in this file.
An example of how to use it can be found here.
Custom Policy Network¶
Stable baselines provides default policy networks (see Policies ) for images (CNNPolicies) and other type of input features (MlpPolicies).
One way of customising the policy network architecture is to pass arguments when creating the model,
using policy_kwargs parameter:
import gym
import tensorflow as tf
from stable_baselines import PPO2
# Custom MLP policy of two layers of size 32 each with tanh activation function
policy_kwargs = dict(act_fun=tf.nn.tanh, net_arch=[32, 32])
# Create the agent
model = PPO2("MlpPolicy", "CartPole-v1", policy_kwargs=policy_kwargs, verbose=1)
# Retrieve the environment
env = model.get_env()
# Train the agent
model.learn(total_timesteps=100000)
# Save the agent
model.save("ppo2-cartpole")
del model
# the policy_kwargs are automatically loaded
model = PPO2.load("ppo2-cartpole")
You can also easily define a custom architecture for the policy (or value) network:
Note
Defining a custom policy class is equivalent to passing policy_kwargs.
However, it lets you name the policy and so makes usually the code clearer.
policy_kwargs should be rather used when doing hyperparameter search.
import gym
from stable_baselines.common.policies import FeedForwardPolicy, register_policy
from stable_baselines.common.vec_env import DummyVecEnv
from stable_baselines import A2C
# Custom MLP policy of three layers of size 128 each
class CustomPolicy(FeedForwardPolicy):
def __init__(self, *args, **kwargs):
super(CustomPolicy, self).__init__(*args, **kwargs,
net_arch=[dict(pi=[128, 128, 128],
vf=[128, 128, 128])],
feature_extraction="mlp")
# Create and wrap the environment
env = gym.make('LunarLander-v2')
env = DummyVecEnv([lambda: env])
model = A2C(CustomPolicy, env, verbose=1)
# Train the agent
model.learn(total_timesteps=100000)
# Save the agent
model.save("a2c-lunar")
del model
# When loading a model with a custom policy
# you MUST pass explicitly the policy when loading the saved model
model = A2C.load("a2c-lunar", policy=CustomPolicy)
Warning
When loading a model with a custom policy, you must pass the custom policy explicitly when loading the model. (cf previous example)
You can also register your policy, to help with code simplicity: you can refer to your custom policy using a string.
import gym
from stable_baselines.common.policies import FeedForwardPolicy, register_policy
from stable_baselines.common.vec_env import DummyVecEnv
from stable_baselines import A2C
# Custom MLP policy of three layers of size 128 each
class CustomPolicy(FeedForwardPolicy):
def __init__(self, *args, **kwargs):
super(CustomPolicy, self).__init__(*args, **kwargs,
net_arch=[dict(pi=[128, 128, 128],
vf=[128, 128, 128])],
feature_extraction="mlp")
# Register the policy, it will check that the name is not already taken
register_policy('CustomPolicy', CustomPolicy)
# Because the policy is now registered, you can pass
# a string to the agent constructor instead of passing a class
model = A2C(policy='CustomPolicy', env='LunarLander-v2', verbose=1).learn(total_timesteps=100000)
Deprecated since version 2.3.0: Use net_arch instead of layers parameter to define the network architecture. It allows to have a greater control.
The net_arch parameter of FeedForwardPolicy allows to specify the amount and size of the hidden layers and how many
of them are shared between the policy network and the value network. It is assumed to be a list with the following
structure:
- An arbitrary length (zero allowed) number of integers each specifying the number of units in a shared layer. If the number of ints is zero, there will be no shared layers.
- An optional dict, to specify the following non-shared layers for the value network and the policy network.
It is formatted like
dict(vf=[<value layer sizes>], pi=[<policy layer sizes>]). If it is missing any of the keys (pi or vf), no non-shared layers (empty list) is assumed.
In short: [<shared layers>, dict(vf=[<non-shared value network layers>], pi=[<non-shared policy network layers>])].
Examples¶
Two shared layers of size 128: net_arch=[128, 128]
obs
|
<128>
|
<128>
/ \
action value
Value network deeper than policy network, first layer shared: net_arch=[128, dict(vf=[256, 256])]
obs
|
<128>
/ \
action <256>
|
<256>
|
value
Initially shared then diverging: [128, dict(vf=[256], pi=[16])]
obs
|
<128>
/ \
<16> <256>
| |
action value
The LstmPolicy can be used to construct recurrent policies in a similar way:
class CustomLSTMPolicy(LstmPolicy):
def __init__(self, sess, ob_space, ac_space, n_env, n_steps, n_batch, n_lstm=64, reuse=False, **_kwargs):
super().__init__(sess, ob_space, ac_space, n_env, n_steps, n_batch, n_lstm, reuse,
net_arch=[8, 'lstm', dict(vf=[5, 10], pi=[10])],
layer_norm=True, feature_extraction="mlp", **_kwargs)
Here the net_arch parameter takes an additional (mandatory) ‘lstm’ entry within the shared network section.
The LSTM is shared between value network and policy network.
If your task requires even more granular control over the policy architecture, you can redefine the policy directly:
import gym
import tensorflow as tf
from stable_baselines.common.policies import ActorCriticPolicy, register_policy, nature_cnn
from stable_baselines.common.vec_env import DummyVecEnv
from stable_baselines import A2C
# Custom MLP policy of three layers of size 128 each for the actor and 2 layers of 32 for the critic,
# with a nature_cnn feature extractor
class CustomPolicy(ActorCriticPolicy):
def __init__(self, sess, ob_space, ac_space, n_env, n_steps, n_batch, reuse=False, **kwargs):
super(CustomPolicy, self).__init__(sess, ob_space, ac_space, n_env, n_steps, n_batch, reuse=reuse, scale=True)
with tf.variable_scope("model", reuse=reuse):
activ = tf.nn.relu
extracted_features = nature_cnn(self.processed_obs, **kwargs)
extracted_features = tf.layers.flatten(extracted_features)
pi_h = extracted_features
for i, layer_size in enumerate([128, 128, 128]):
pi_h = activ(tf.layers.dense(pi_h, layer_size, name='pi_fc' + str(i)))
pi_latent = pi_h
vf_h = extracted_features
for i, layer_size in enumerate([32, 32]):
vf_h = activ(tf.layers.dense(vf_h, layer_size, name='vf_fc' + str(i)))
value_fn = tf.layers.dense(vf_h, 1, name='vf')
vf_latent = vf_h
self._proba_distribution, self._policy, self.q_value = \
self.pdtype.proba_distribution_from_latent(pi_latent, vf_latent, init_scale=0.01)
self._value_fn = value_fn
self._setup_init()
def step(self, obs, state=None, mask=None, deterministic=False):
if deterministic:
action, value, neglogp = self.sess.run([self.deterministic_action, self.value_flat, self.neglogp],
{self.obs_ph: obs})
else:
action, value, neglogp = self.sess.run([self.action, self.value_flat, self.neglogp],
{self.obs_ph: obs})
return action, value, self.initial_state, neglogp
def proba_step(self, obs, state=None, mask=None):
return self.sess.run(self.policy_proba, {self.obs_ph: obs})
def value(self, obs, state=None, mask=None):
return self.sess.run(self.value_flat, {self.obs_ph: obs})
# Create and wrap the environment
env = DummyVecEnv([lambda: gym.make('Breakout-v0')])
model = A2C(CustomPolicy, env, verbose=1)
# Train the agent
model.learn(total_timesteps=100000)
Callbacks¶
A callback is a set of functions that will be called at given stages of the training procedure. You can use callbacks to access internal state of the RL model during training. It allows one to do monitoring, auto saving, model manipulation, progress bars, …
Custom Callback¶
To build a custom callback, you need to create a class that derives from BaseCallback.
This will give you access to events (_on_training_start, _on_step) and useful variables (like self.model for the RL model).
You can find two examples of custom callbacks in the documentation: one for saving the best model according to the training reward (see Examples), and one for logging additional values with Tensorboard (see Tensorboard section).
from stable_baselines.common.callbacks import BaseCallback
class CustomCallback(BaseCallback):
"""
A custom callback that derives from ``BaseCallback``.
:param verbose: (int) Verbosity level 0: not output 1: info 2: debug
"""
def __init__(self, verbose=0):
super(CustomCallback, self).__init__(verbose)
# Those variables will be accessible in the callback
# (they are defined in the base class)
# The RL model
# self.model = None # type: BaseRLModel
# An alias for self.model.get_env(), the environment used for training
# self.training_env = None # type: Union[gym.Env, VecEnv, None]
# Number of time the callback was called
# self.n_calls = 0 # type: int
# self.num_timesteps = 0 # type: int
# local and global variables
# self.locals = None # type: Dict[str, Any]
# self.globals = None # type: Dict[str, Any]
# The logger object, used to report things in the terminal
# self.logger = None # type: logger.Logger
# # Sometimes, for event callback, it is useful
# # to have access to the parent object
# self.parent = None # type: Optional[BaseCallback]
def _on_training_start(self) -> None:
"""
This method is called before the first rollout starts.
"""
pass
def _on_rollout_start(self) -> None:
"""
A rollout is the collection of environment interaction
using the current policy.
This event is triggered before collecting new samples.
"""
pass
def _on_step(self) -> bool:
"""
This method will be called by the model after each call to `env.step()`.
For child callback (of an `EventCallback`), this will be called
when the event is triggered.
:return: (bool) If the callback returns False, training is aborted early.
"""
return True
def _on_rollout_end(self) -> None:
"""
This event is triggered before updating the policy.
"""
pass
def _on_training_end(self) -> None:
"""
This event is triggered before exiting the `learn()` method.
"""
pass
Note
self.num_timesteps corresponds to the total number of steps taken in the environment, i.e., it is the number of environments multiplied by the number of time env.step() was called
You should know that PPO1 and TRPO update self.num_timesteps after each rollout (and not each step) because they rely on MPI.
For the other algorithms, self.num_timesteps is incremented by n_envs (number of environments) after each call to env.step()
Note
For off-policy algorithms like SAC, DDPG, TD3 or DQN, the notion of rollout corresponds to the steps taken in the environment between two updates.
Event Callback¶
Compared to Keras, Stable Baselines provides a second type of BaseCallback, named EventCallback that is meant to trigger events. When an event is triggered, then a child callback is called.
As an example, EvalCallback is an EventCallback that will trigger its child callback when there is a new best model.
A child callback is for instance StopTrainingOnRewardThreshold that stops the training if the mean reward achieved by the RL model is above a threshold.
Note
We recommend to take a look at the source code of EvalCallback and StopTrainingOnRewardThreshold to have a better overview of what can be achieved with this kind of callbacks.
class EventCallback(BaseCallback):
"""
Base class for triggering callback on event.
:param callback: (Optional[BaseCallback]) Callback that will be called
when an event is triggered.
:param verbose: (int)
"""
def __init__(self, callback: Optional[BaseCallback] = None, verbose: int = 0):
super(EventCallback, self).__init__(verbose=verbose)
self.callback = callback
# Give access to the parent
if callback is not None:
self.callback.parent = self
...
def _on_event(self) -> bool:
if self.callback is not None:
return self.callback()
return True
Callback Collection¶
Stable Baselines provides you with a set of common callbacks for:
- saving the model periodically (CheckpointCallback)
- evaluating the model periodically and saving the best one (EvalCallback)
- chaining callbacks (CallbackList)
- triggering callback on events (Event Callback, EveryNTimesteps)
- stopping the training early based on a reward threshold (StopTrainingOnRewardThreshold)
CheckpointCallback¶
Callback for saving a model every save_freq steps, you must specify a log folder (save_path)
and optionally a prefix for the checkpoints (rl_model by default).
from stable_baselines import SAC
from stable_baselines.common.callbacks import CheckpointCallback
# Save a checkpoint every 1000 steps
checkpoint_callback = CheckpointCallback(save_freq=1000, save_path='./logs/',
name_prefix='rl_model')
model = SAC('MlpPolicy', 'Pendulum-v0')
model.learn(2000, callback=checkpoint_callback)
EvalCallback¶
Evaluate periodically the performance of an agent, using a separate test environment.
It will save the best model if best_model_save_path folder is specified and save the evaluations results in a numpy archive (evaluations.npz) if log_path folder is specified.
Note
You can pass a child callback via the callback_on_new_best argument. It will be triggered each time there is a new best model.
import gym
from stable_baselines import SAC
from stable_baselines.common.callbacks import EvalCallback
# Separate evaluation env
eval_env = gym.make('Pendulum-v0')
# Use deterministic actions for evaluation
eval_callback = EvalCallback(eval_env, best_model_save_path='./logs/',
log_path='./logs/', eval_freq=500,
deterministic=True, render=False)
model = SAC('MlpPolicy', 'Pendulum-v0')
model.learn(5000, callback=eval_callback)
CallbackList¶
Class for chaining callbacks, they will be called sequentially.
Alternatively, you can pass directly a list of callbacks to the learn() method, it will be converted automatically to a CallbackList.
import gym
from stable_baselines import SAC
from stable_baselines.common.callbacks import CallbackList, CheckpointCallback, EvalCallback
checkpoint_callback = CheckpointCallback(save_freq=1000, save_path='./logs/')
# Separate evaluation env
eval_env = gym.make('Pendulum-v0')
eval_callback = EvalCallback(eval_env, best_model_save_path='./logs/best_model',
log_path='./logs/results', eval_freq=500)
# Create the callback list
callback = CallbackList([checkpoint_callback, eval_callback])
model = SAC('MlpPolicy', 'Pendulum-v0')
# Equivalent to:
# model.learn(5000, callback=[checkpoint_callback, eval_callback])
model.learn(5000, callback=callback)
StopTrainingOnRewardThreshold¶
Stop the training once a threshold in episodic reward (mean episode reward over the evaluations) has been reached (i.e., when the model is good enough). It must be used with the EvalCallback and use the event triggered by a new best model.
import gym
from stable_baselines import SAC
from stable_baselines.common.callbacks import EvalCallback, StopTrainingOnRewardThreshold
# Separate evaluation env
eval_env = gym.make('Pendulum-v0')
# Stop training when the model reaches the reward threshold
callback_on_best = StopTrainingOnRewardThreshold(reward_threshold=-200, verbose=1)
eval_callback = EvalCallback(eval_env, callback_on_new_best=callback_on_best, verbose=1)
model = SAC('MlpPolicy', 'Pendulum-v0', verbose=1)
# Almost infinite number of timesteps, but the training will stop
# early as soon as the reward threshold is reached
model.learn(int(1e10), callback=eval_callback)
EveryNTimesteps¶
An Event Callback that will trigger its child callback every n_steps timesteps.
Note
Because of the way PPO1 and TRPO work (they rely on MPI), n_steps is a lower bound between two events.
import gym
from stable_baselines import PPO2
from stable_baselines.common.callbacks import CheckpointCallback, EveryNTimesteps
# this is equivalent to defining CheckpointCallback(save_freq=500)
# checkpoint_callback will be triggered every 500 steps
checkpoint_on_event = CheckpointCallback(save_freq=1, save_path='./logs/')
event_callback = EveryNTimesteps(n_steps=500, callback=checkpoint_on_event)
model = PPO2('MlpPolicy', 'Pendulum-v0', verbose=1)
model.learn(int(2e4), callback=event_callback)
Legacy: A functional approach¶
Warning
This way of doing callbacks is deprecated in favor of the object oriented approach.
A callback function takes the locals() variables and the globals() variables from the model, then returns a boolean value for whether or not the training should continue.
Thanks to the access to the models variables, in particular _locals["self"], we are able to even change the parameters of the model without halting the training, or changing the model’s code.
from typing import Dict, Any
from stable_baselines import PPO2
def simple_callback(_locals: Dict[str, Any], _globals: Dict[str, Any]) -> bool:
"""
Callback called at each step (for DQN and others) or after n steps (see ACER or PPO2).
This callback will save the model and stop the training after the first call.
:param _locals: (Dict[str, Any])
:param _globals: (Dict[str, Any])
:return: (bool) If your callback returns False, training is aborted early.
"""
print("callback called")
# Save the model
_locals["self"].save("saved_model")
# If you want to continue training, the callback must return True.
# return True # returns True, training continues.
print("stop training")
return False # returns False, training stops.
model = PPO2('MlpPolicy', 'CartPole-v1')
model.learn(2000, callback=simple_callback)
-
class
stable_baselines.common.callbacks.BaseCallback(verbose: int = 0)[source]¶ Base class for callback.
Parameters: verbose – (int)
-
class
stable_baselines.common.callbacks.CallbackList(callbacks: List[stable_baselines.common.callbacks.BaseCallback])[source]¶ Class for chaining callbacks.
Parameters: callbacks – (List[BaseCallback]) A list of callbacks that will be called sequentially.
-
class
stable_baselines.common.callbacks.CheckpointCallback(save_freq: int, save_path: str, name_prefix='rl_model', verbose=0)[source]¶ Callback for saving a model every save_freq steps
Parameters: - save_freq – (int)
- save_path – (str) Path to the folder where the model will be saved.
- name_prefix – (str) Common prefix to the saved models
-
class
stable_baselines.common.callbacks.ConvertCallback(callback, verbose=0)[source]¶ Convert functional callback (old-style) to object.
Parameters: - callback – (Callable)
- verbose – (int)
-
class
stable_baselines.common.callbacks.EvalCallback(eval_env: Union[<MagicMock id='140553671154432'>, stable_baselines.common.vec_env.base_vec_env.VecEnv], callback_on_new_best: Optional[stable_baselines.common.callbacks.BaseCallback] = None, n_eval_episodes: int = 5, eval_freq: int = 10000, log_path: str = None, best_model_save_path: str = None, deterministic: bool = True, render: bool = False, verbose: int = 1)[source]¶ Callback for evaluating an agent.
Parameters: - eval_env – (Union[gym.Env, VecEnv]) The environment used for initialization
- callback_on_new_best – (Optional[BaseCallback]) Callback to trigger when there is a new best model according to the mean_reward
- n_eval_episodes – (int) The number of episodes to test the agent
- eval_freq – (int) Evaluate the agent every eval_freq call of the callback.
- log_path – (str) Path to a folder where the evaluations (evaluations.npz) will be saved. It will be updated at each evaluation.
- best_model_save_path – (str) Path to a folder where the best model according to performance on the eval env will be saved.
- deterministic – (bool) Whether the evaluation should use a stochastic or deterministic actions.
- render – (bool) Whether to render or not the environment during evaluation
- verbose – (int)
-
class
stable_baselines.common.callbacks.EventCallback(callback: Optional[stable_baselines.common.callbacks.BaseCallback] = None, verbose: int = 0)[source]¶ Base class for triggering callback on event.
Parameters: - callback – (Optional[BaseCallback]) Callback that will be called when an event is triggered.
- verbose – (int)
-
class
stable_baselines.common.callbacks.EveryNTimesteps(n_steps: int, callback: stable_baselines.common.callbacks.BaseCallback)[source]¶ Trigger a callback every n_steps timesteps
Parameters: - n_steps – (int) Number of timesteps between two trigger.
- callback – (BaseCallback) Callback that will be called when the event is triggered.
-
class
stable_baselines.common.callbacks.StopTrainingOnRewardThreshold(reward_threshold: float, verbose: int = 0)[source]¶ Stop the training once a threshold in episodic reward has been reached (i.e. when the model is good enough).
It must be used with the EvalCallback.
Parameters: - reward_threshold – (float) Minimum expected reward per episode to stop training.
- verbose – (int)
Tensorboard Integration¶
Basic Usage¶
To use Tensorboard with the rl baselines, you simply need to define a log location for the RL agent:
import gym
from stable_baselines import A2C
model = A2C('MlpPolicy', 'CartPole-v1', verbose=1, tensorboard_log="./a2c_cartpole_tensorboard/")
model.learn(total_timesteps=10000)
Or after loading an existing model (by default the log path is not saved):
import gym
from stable_baselines.common.vec_env import DummyVecEnv
from stable_baselines import A2C
env = gym.make('CartPole-v1')
env = DummyVecEnv([lambda: env]) # The algorithms require a vectorized environment to run
model = A2C.load("./a2c_cartpole.pkl", env=env, tensorboard_log="./a2c_cartpole_tensorboard/")
model.learn(total_timesteps=10000)
You can also define custom logging name when training (by default it is the algorithm name)
import gym
from stable_baselines import A2C
model = A2C('MlpPolicy', 'CartPole-v1', verbose=1, tensorboard_log="./a2c_cartpole_tensorboard/")
model.learn(total_timesteps=10000, tb_log_name="first_run")
# Pass reset_num_timesteps=False to continue the training curve in tensorboard
# By default, it will create a new curve
model.learn(total_timesteps=10000, tb_log_name="second_run", reset_num_timesteps=False)
model.learn(total_timesteps=10000, tb_log_name="thrid_run", reset_num_timesteps=False)
Once the learn function is called, you can monitor the RL agent during or after the training, with the following bash command:
tensorboard --logdir ./a2c_cartpole_tensorboard/
you can also add past logging folders:
tensorboard --logdir ./a2c_cartpole_tensorboard/;./ppo2_cartpole_tensorboard/
It will display information such as the model graph, the episode reward, the model losses, the observation and other parameter unique to some models.
Logging More Values¶
Using a callback, you can easily log more values with TensorBoard. Here is a simple example on how to log both additional tensor or arbitrary scalar value:
import tensorflow as tf
import numpy as np
from stable_baselines import SAC
from stable_baselines.common.callbacks import BaseCallback
model = SAC("MlpPolicy", "Pendulum-v0", tensorboard_log="/tmp/sac/", verbose=1)
class TensorboardCallback(BaseCallback):
"""
Custom callback for plotting additional values in tensorboard.
"""
def __init__(self, verbose=0):
self.is_tb_set = False
super(TensorboardCallback, self).__init__(verbose)
def _on_step(self) -> bool:
# Log additional tensor
if not self.is_tb_set:
with self.model.graph.as_default():
tf.summary.scalar('value_target', tf.reduce_mean(self.model.value_target))
self.model.summary = tf.summary.merge_all()
self.is_tb_set = True
# Log scalar value (here a random variable)
value = np.random.random()
summary = tf.Summary(value=[tf.Summary.Value(tag='random_value', simple_value=value)])
self.locals['writer'].add_summary(summary, self.num_timesteps)
return True
model.learn(50000, callback=TensorboardCallback())
Legacy Integration¶
All the information displayed in the terminal (default logging) can be also logged in tensorboard. For that, you need to define several environment variables:
# formats are comma-separated, but for tensorboard you only need the last one
# stdout -> terminal
export OPENAI_LOG_FORMAT='stdout,log,csv,tensorboard'
export OPENAI_LOGDIR=path/to/tensorboard/data
and to configure the logger using:
from stable_baselines.logger import configure
configure()
Then start tensorboard with:
tensorboard --logdir=$OPENAI_LOGDIR
RL Baselines Zoo¶
RL Baselines Zoo. is a collection of pre-trained Reinforcement Learning agents using Stable-Baselines. It also provides basic scripts for training, evaluating agents, tuning hyperparameters and recording videos.
Goals of this repository:
- Provide a simple interface to train and enjoy RL agents
- Benchmark the different Reinforcement Learning algorithms
- Provide tuned hyperparameters for each environment and RL algorithm
- Have fun with the trained agents!
Installation¶
1. Install dependencies
apt-get install swig cmake libopenmpi-dev zlib1g-dev ffmpeg
pip install stable-baselines box2d box2d-kengz pyyaml pybullet optuna pytablewriter
- Clone the repository:
git clone https://github.com/araffin/rl-baselines-zoo
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/
Train for multiple environments (with one call) and with tensorboard logging:
python train.py --algo a2c --env MountainCar-v0 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
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
Hyperparameter Optimization¶
We use Optuna for optimizing the hyperparameters.
Tune the hyperparameters for PPO2, using a random sampler and median pruner, 2 parallels jobs, with a budget of 1000 trials and a maximum of 50000 steps:
python train.py --algo ppo2 --env MountainCar-v0 -n 50000 -optimize --n-trials 1000 --n-jobs 2 \
--sampler random --pruner median
Colab Notebook: Try it Online!¶
You can train agents online using Google colab notebook.
Note
You can find more information about the rl baselines zoo in the repo README. For instance, how to record a video of a trained agent.
Pre-Training (Behavior Cloning)¶
With the .pretrain() method, you can pre-train RL policies using trajectories from an expert, and therefore accelerate training.
Behavior Cloning (BC) treats the problem of imitation learning, i.e., using expert demonstrations, as a supervised learning problem. That is to say, given expert trajectories (observations-actions pairs), the policy network is trained to reproduce the expert behavior: for a given observation, the action taken by the policy must be the one taken by the expert.
Expert trajectories can be human demonstrations, trajectories from another controller (e.g. a PID controller) or trajectories from a trained RL agent.
Note
Only Box and Discrete spaces are supported for now for pre-training a model.
Note
Images datasets are treated a bit differently as other datasets to avoid memory issues. The images from the expert demonstrations must be located in a folder, not in the expert numpy archive.
Generate Expert Trajectories¶
Here, we are going to train a RL model and then generate expert trajectories using this agent.
Note that in practice, generating expert trajectories usually does not require training an RL agent.
The following example is only meant to demonstrate the pretrain() feature.
However, we recommend users to take a look at the code of the generate_expert_traj() function (located in gail/dataset/ folder)
to learn about the data structure of the expert dataset (see below for an overview) and how to record trajectories.
from stable_baselines import DQN
from stable_baselines.gail import generate_expert_traj
model = DQN('MlpPolicy', 'CartPole-v1', verbose=1)
# Train a DQN agent for 1e5 timesteps and generate 10 trajectories
# data will be saved in a numpy archive named `expert_cartpole.npz`
generate_expert_traj(model, 'expert_cartpole', n_timesteps=int(1e5), n_episodes=10)
Here is an additional example when the expert controller is a callable, that is passed to the function instead of a RL model. The idea is that this callable can be a PID controller, asking a human player, …
import gym
from stable_baselines.gail import generate_expert_traj
env = gym.make("CartPole-v1")
# Here the expert is a random agent
# but it can be any python function, e.g. a PID controller
def dummy_expert(_obs):
"""
Random agent. It samples actions randomly
from the action space of the environment.
:param _obs: (np.ndarray) Current observation
:return: (np.ndarray) action taken by the expert
"""
return env.action_space.sample()
# Data will be saved in a numpy archive named `expert_cartpole.npz`
# when using something different than an RL expert,
# you must pass the environment object explicitly
generate_expert_traj(dummy_expert, 'dummy_expert_cartpole', env, n_episodes=10)
Pre-Train a Model using Behavior Cloning¶
Using the expert_cartpole.npz dataset generated with the previous script.
from stable_baselines import PPO2
from stable_baselines.gail import ExpertDataset
# Using only one expert trajectory
# you can specify `traj_limitation=-1` for using the whole dataset
dataset = ExpertDataset(expert_path='expert_cartpole.npz',
traj_limitation=1, batch_size=128)
model = PPO2('MlpPolicy', 'CartPole-v1', verbose=1)
# Pretrain the PPO2 model
model.pretrain(dataset, n_epochs=1000)
# As an option, you can train the RL agent
# model.learn(int(1e5))
# Test the pre-trained model
env = model.get_env()
obs = env.reset()
reward_sum = 0.0
for _ in range(1000):
action, _ = model.predict(obs)
obs, reward, done, _ = env.step(action)
reward_sum += reward
env.render()
if done:
print(reward_sum)
reward_sum = 0.0
obs = env.reset()
env.close()
Data Structure of the Expert Dataset¶
The expert dataset is a .npz archive. The data is saved in python dictionary format with keys: actions, episode_returns, rewards, obs, episode_starts.
In case of images, obs contains the relative path to the images.
obs, actions: shape (N * L, ) + S
where N = # episodes, L = episode length and S is the environment observation/action space.
S = (1, ) for discrete space
-
class
stable_baselines.gail.ExpertDataset(expert_path=None, traj_data=None, train_fraction=0.7, batch_size=64, traj_limitation=-1, randomize=True, verbose=1, sequential_preprocessing=False)[source]¶ Dataset for using behavior cloning or GAIL.
The structure of the expert dataset is a dict, saved as an “.npz” archive. The dictionary contains the keys ‘actions’, ‘episode_returns’, ‘rewards’, ‘obs’ and ‘episode_starts’. The corresponding values have data concatenated across episode: the first axis is the timestep, the remaining axes index into the data. In case of images, ‘obs’ contains the relative path to the images, to enable space saving from image compression.
Parameters: - expert_path – (str) The path to trajectory data (.npz file). Mutually exclusive with traj_data.
- traj_data – (dict) Trajectory data, in format described above. Mutually exclusive with expert_path.
- train_fraction – (float) the train validation split (0 to 1) for pre-training using behavior cloning (BC)
- batch_size – (int) the minibatch size for behavior cloning
- traj_limitation – (int) the number of trajectory to use (if -1, load all)
- randomize – (bool) if the dataset should be shuffled
- verbose – (int) Verbosity
- sequential_preprocessing – (bool) Do not use subprocess to preprocess the data (slower but use less memory for the CI)
-
get_next_batch(split=None)[source]¶ Get the batch from the dataset.
Parameters: split – (str) the type of data split (can be None, ‘train’, ‘val’) Returns: (np.ndarray, np.ndarray) inputs and labels
-
class
stable_baselines.gail.DataLoader(indices, observations, actions, batch_size, n_workers=1, infinite_loop=True, max_queue_len=1, shuffle=False, start_process=True, backend='threading', sequential=False, partial_minibatch=True)[source]¶ A custom dataloader to preprocessing observations (including images) and feed them to the network.
Original code for the dataloader from https://github.com/araffin/robotics-rl-srl (MIT licence) Authors: Antonin Raffin, René Traoré, Ashley Hill
Parameters: - indices – ([int]) list of observations indices
- observations – (np.ndarray) observations or images path
- actions – (np.ndarray) actions
- batch_size – (int) Number of samples per minibatch
- n_workers – (int) number of preprocessing worker (for loading the images)
- infinite_loop – (bool) whether to have an iterator that can be reset
- max_queue_len – (int) Max number of minibatches that can be preprocessed at the same time
- shuffle – (bool) Shuffle the minibatch after each epoch
- start_process – (bool) Start the preprocessing process (default: True)
- backend – (str) joblib backend (one of ‘multiprocessing’, ‘sequential’, ‘threading’ or ‘loky’ in newest versions)
- sequential – (bool) Do not use subprocess to preprocess the data (slower but use less memory for the CI)
- partial_minibatch – (bool) Allow partial minibatches (minibatches with a number of element lesser than the batch_size)
-
stable_baselines.gail.generate_expert_traj(model, save_path=None, env=None, n_timesteps=0, n_episodes=100, image_folder='recorded_images')[source]¶ Train expert controller (if needed) and record expert trajectories.
Note
only Box and Discrete spaces are supported for now.
Parameters: - model – (RL model or callable) The expert model, if it needs to be trained,
then you need to pass
n_timesteps > 0. - save_path – (str) Path without the extension where the expert dataset will be saved (ex: ‘expert_cartpole’ -> creates ‘expert_cartpole.npz’). If not specified, it will not save, and just return the generated expert trajectories. This parameter must be specified for image-based environments.
- env – (gym.Env) The environment, if not defined then it tries to use the model environment.
- n_timesteps – (int) Number of training timesteps
- n_episodes – (int) Number of trajectories (episodes) to record
- image_folder – (str) When using images, folder that will be used to record images.
Returns: (dict) the generated expert trajectories.
- model – (RL model or callable) The expert model, if it needs to be trained,
then you need to pass
Dealing with NaNs and infs¶
During the training of a model on a given environment, it is possible that the RL model becomes completely corrupted when a NaN or an inf is given or returned from the RL model.
How and why?¶
The issue arises then NaNs or infs do not crash, but simply get propagated through the training, until all the floating point number converge to NaN or inf. This is in line with the IEEE Standard for Floating-Point Arithmetic (IEEE 754) standard, as it says:
Note
- Five possible exceptions can occur:
- Invalid operation (\(\sqrt{-1}\), \(\inf \times 1\), \(\text{NaN}\ \mathrm{mod}\ 1\), …) return NaN
- Division by zero:
- if the operand is not zero (\(1/0\), \(-2/0\), …) returns \(\pm\inf\)
- if the operand is zero (\(0/0\)) returns signaling NaN
- Overflow (exponent too high to represent) returns \(\pm\inf\)
- Underflow (exponent too low to represent) returns \(0\)
- Inexact (not representable exactly in base 2, eg: \(1/5\)) returns the rounded value (ex:
assert (1/5) * 3 == 0.6000000000000001)
And of these, only Division by zero will signal an exception, the rest will propagate invalid values quietly.
In python, dividing by zero will indeed raise the exception: ZeroDivisionError: float division by zero,
but ignores the rest.
The default in numpy, will warn: RuntimeWarning: invalid value encountered
but will not halt the code.
And the worst of all, Tensorflow will not signal anything
import tensorflow as tf
import numpy as np
print("tensorflow test:")
a = tf.constant(1.0)
b = tf.constant(0.0)
c = a / b
sess = tf.Session()
val = sess.run(c) # this will be quiet
print(val)
sess.close()
print("\r\nnumpy test:")
a = np.float64(1.0)
b = np.float64(0.0)
val = a / b # this will warn
print(val)
print("\r\npure python test:")
a = 1.0
b = 0.0
val = a / b # this will raise an exception and halt.
print(val)
Unfortunately, most of the floating point operations are handled by Tensorflow and numpy, meaning you might get little to no warning when a invalid value occurs.
Numpy parameters¶
Numpy has a convenient way of dealing with invalid value: numpy.seterr, which defines for the python process, how it should handle floating point error.
import numpy as np
np.seterr(all='raise') # define before your code.
print("numpy test:")
a = np.float64(1.0)
b = np.float64(0.0)
val = a / b # this will now raise an exception instead of a warning.
print(val)
but this will also avoid overflow issues on floating point numbers:
import numpy as np
np.seterr(all='raise') # define before your code.
print("numpy overflow test:")
a = np.float64(10)
b = np.float64(1000)
val = a ** b # this will now raise an exception
print(val)
but will not avoid the propagation issues:
import numpy as np
np.seterr(all='raise') # define before your code.
print("numpy propagation test:")
a = np.float64('NaN')
b = np.float64(1.0)
val = a + b # this will neither warn nor raise anything
print(val)
Tensorflow parameters¶
Tensorflow can add checks for detecting and dealing with invalid value: tf.add_check_numerics_ops and tf.check_numerics, however they will add operations to the Tensorflow graph and raise the computation time.
import tensorflow as tf
print("tensorflow test:")
a = tf.constant(1.0)
b = tf.constant(0.0)
c = a / b
check_nan = tf.add_check_numerics_ops() # add after your graph definition.
sess = tf.Session()
val, _ = sess.run([c, check_nan]) # this will now raise an exception
print(val)
sess.close()
but this will also avoid overflow issues on floating point numbers:
import tensorflow as tf
print("tensorflow overflow test:")
check_nan = [] # the list of check_numerics operations
a = tf.constant(10)
b = tf.constant(1000)
c = a ** b
check_nan.append(tf.check_numerics(c, "")) # check the 'c' operations
sess = tf.Session()
val, _ = sess.run([c] + check_nan) # this will now raise an exception
print(val)
sess.close()
and catch propagation issues:
import tensorflow as tf
print("tensorflow propagation test:")
check_nan = [] # the list of check_numerics operations
a = tf.constant('NaN')
b = tf.constant(1.0)
c = a + b
check_nan.append(tf.check_numerics(c, "")) # check the 'c' operations
sess = tf.Session()
val, _ = sess.run([c] + check_nan) # this will now raise an exception
print(val)
sess.close()
VecCheckNan Wrapper¶
In order to find when and from where the invalid value originated from, stable-baselines comes with a VecCheckNan wrapper.
It will monitor the actions, observations, and rewards, indicating what action or observation caused it and from what.
import gym
from gym import spaces
import numpy as np
from stable_baselines import PPO2
from stable_baselines.common.vec_env import DummyVecEnv, VecCheckNan
class NanAndInfEnv(gym.Env):
"""Custom Environment that raised NaNs and Infs"""
metadata = {'render.modes': ['human']}
def __init__(self):
super(NanAndInfEnv, self).__init__()
self.action_space = spaces.Box(low=-np.inf, high=np.inf, shape=(1,), dtype=np.float64)
self.observation_space = spaces.Box(low=-np.inf, high=np.inf, shape=(1,), dtype=np.float64)
def step(self, _action):
randf = np.random.rand()
if randf > 0.99:
obs = float('NaN')
elif randf > 0.98:
obs = float('inf')
else:
obs = randf
return [obs], 0.0, False, {}
def reset(self):
return [0.0]
def render(self, mode='human', close=False):
pass
# Create environment
env = DummyVecEnv([lambda: NanAndInfEnv()])
env = VecCheckNan(env, raise_exception=True)
# Instantiate the agent
model = PPO2('MlpPolicy', env)
# Train the agent
model.learn(total_timesteps=int(2e5)) # this will crash explaining that the invalid value originated from the environment.
RL Model hyperparameters¶
Depending on your hyperparameters, NaN can occurs much more often. A great example of this: https://github.com/hill-a/stable-baselines/issues/340
Be aware, the hyperparameters given by default seem to work in most cases, however your environment might not play nice with them. If this is the case, try to read up on the effect each hyperparameters has on the model, so that you can try and tune them to get a stable model. Alternatively, you can try automatic hyperparameter tuning (included in the rl zoo).
Missing values from datasets¶
If your environment is generated from an external dataset, do not forget to make sure your dataset does not contain NaNs. As some datasets will sometimes fill missing values with NaNs as a surrogate value.
Here is some reading material about finding NaNs: https://pandas.pydata.org/pandas-docs/stable/user_guide/missing_data.html
And filling the missing values with something else (imputation): https://towardsdatascience.com/how-to-handle-missing-data-8646b18db0d4
On saving and loading¶
Stable baselines stores both neural network parameters and algorithm-related parameters such as exploration schedule, number of environments and observation/action space. This allows continual learning and easy use of trained agents without training, but it is not without its issues. Following describes two formats used to save agents in stable baselines, their pros and shortcomings.
Terminology used in this page:
- parameters refer to neural network parameters (also called “weights”). This is a dictionary mapping Tensorflow variable name to a NumPy array.
- data refers to RL algorithm parameters, e.g. learning rate, exploration schedule, action/observation space. These depend on the algorithm used. This is a dictionary mapping classes variable names their values.
Cloudpickle (stable-baselines<=2.7.0)¶
Original stable baselines save format. Data and parameters are bundled up into a tuple (data, parameters)
and then serialized with cloudpickle library (essentially the same as pickle).
This save format is still available via an argument in model save function in stable-baselines versions above v2.7.0 for backwards compatibility reasons, but its usage is discouraged.
Pros:
- Easy to implement and use.
- Works with almost any type of Python object, including functions.
Cons:
- Pickle/Cloudpickle is not designed for long-term storage or sharing between Python version.
- If one object in file is not readable (e.g. wrong library version), then reading the rest of the file is difficult.
- Python-specific format, hard to read stored files from other languages.
If part of a saved model becomes unreadable for any reason (e.g. different Tensorflow versions), then it may be tricky to restore any of the model. For this reason another save format was designed.
Zip-archive (stable-baselines>2.7.0)¶
A zip-archived JSON dump and NumPy zip archive of the arrays. The data dictionary (class parameters)
is stored as a JSON file, model parameters are serialized with numpy.savez function and these two files
are stored under a single .zip archive.
Any objects that are not JSON serializable are serialized with cloudpickle and stored as base64-encoded string in the JSON file, along with some information that was stored in the serialization. This allows inspecting stored objects without deserializing the object itself.
This format allows skipping elements in the file, i.e. we can skip deserializing objects that are
broken/non-serializable. This can be done via custom_objects argument to load functions.
This is the default save format in stable baselines versions after v2.7.0.
File structure:
saved_model.zip/
├── data JSON file of class-parameters (dictionary)
├── parameter_list JSON file of model parameters and their ordering (list)
├── parameters Bytes from numpy.savez (a zip file of the numpy arrays). ...
├── ... Being a zip-archive itself, this object can also be opened ...
├── ... as a zip-archive and browsed.
Pros:
- More robust to unserializable objects (one bad object does not break everything).
- Saved file can be inspected/extracted with zip-archive explorers and by other languages.
Cons:
- More complex implementation.
- Still relies partly on cloudpickle for complex objects (e.g. custom functions).
Exporting models¶
After training an agent, you may want to deploy/use it in an other language or framework, like PyTorch or tensorflowjs. Stable Baselines does not include tools to export models to other frameworks, but this document aims to cover parts that are required for exporting along with more detailed stories from users of Stable Baselines.
Background¶
In Stable Baselines, the controller is stored inside policies which convert
observations into actions. Each learning algorithm (e.g. DQN, A2C, SAC) contains
one or more policies, some of which are only used for training. An easy way to find
the policy is to check the code for the predict function of the agent:
This function should only call one policy with simple arguments.
Policies hold the necessary Tensorflow placeholders and tensors to do the inference (i.e. predict actions), so it is enough to export these policies to do inference in an another framework.
Note
Learning algorithms also may contain other Tensorflow placeholders, that are used for training only and are not required for inference.
Warning
When using CNN policies, the observation is normalized internally (dividing by 255 to have values in [0, 1])
Export to PyTorch¶
A known working solution is to use get_parameters
function to obtain model parameters, construct the network manually in PyTorch and assign parameters correctly.
Warning
PyTorch and Tensorflow have internal differences with e.g. 2D convolutions (see discussion linked below).
See discussion #372 for details.
Export to C++¶
Tensorflow, which is the backbone of Stable Baselines, is fundamentally a C/C++ library despite being most commonly accessed through the Python frontend layer. This design choice means that the models created at Python level should generally be fully compliant with the respective C++ version of Tensorflow.
Warning
It is advisable not to mix-and-match different versions of Tensorflow libraries, particularly in terms of the state. Moving computational graphs is generally more forgiving. As a matter of fact, mentioned below PPO_CPP project uses graphs generated with Python Tensorflow 1.x in C++ Tensorflow 2 version.
Stable Baselines comes very handily when hoping to migrate a computational graph and/or a state (weights) as
the existing algorithms define most of the necessary computations for you so you don’t need to recreate the core of the algorithms again.
This is exactly the idea that has been used in the PPO_CPP project, which executes the training at the C++ level for the sake of
computational efficiency. The graphs are exported from Stable Baselines’ PPO2 implementation through tf.train.export_meta_graph
function. Alternatively, and perhaps more commonly, you could use the C++ layer only for inference. That could be useful
as a deployment step of server backends or optimization for more limited devices.
Warning
As a word of caution, C++-level APIs are more imperative than their Python counterparts or more plainly speaking: cruder.
This is particularly apparent in Tensorflow 2.0 where the declarativeness of Autograph exists only at Python level. The
C++ counterpart still operates on Session objects’ use, which are known from earlier versions of Tensorflow. In our use case,
availability of graphs utilized by Session depends on the use of tf.function decorators. However, as of November 2019, Stable Baselines still
uses Tensorflow 1.x in the main version which is slightly easier to use in the context of the C++ portability.
Export to tensorflowjs / tfjs¶
Can be done via Tensorflow’s simple_save function and tensorflowjs_converter.
See discussion #474 for details.
Manual export¶
You can also manually export required parameters (weights) and construct the network in your desired framework, as done with the PyTorch example above.
You can access parameters of the model via agents’
get_parameters
function. If you use default policies, you can find the architecture of the networks in
source for policies. Otherwise, for DQN/SAC/DDPG or TD3 you need to check the policies.py file located
in their respective folders.
Base RL Class¶
Common interface for all the RL algorithms
-
class
stable_baselines.common.base_class.BaseRLModel(policy, env, verbose=0, *, requires_vec_env, policy_base, policy_kwargs=None, seed=None, n_cpu_tf_sess=None)[source]¶ The base RL model
Parameters: - policy – (BasePolicy) Policy object
- env – (Gym environment) The environment to learn from (if registered in Gym, can be str. Can be None for loading trained models)
- verbose – (int) the verbosity level: 0 none, 1 training information, 2 tensorflow debug
- requires_vec_env – (bool) Does this model require a vectorized environment
- policy_base – (BasePolicy) the base policy used by this method
- policy_kwargs – (dict) additional arguments to be passed to the policy on creation
- seed – (int) Seed for the pseudo-random generators (python, numpy, tensorflow). If None (default), use random seed. Note that if you want completely deterministic results, you must set n_cpu_tf_sess to 1.
- n_cpu_tf_sess – (int) The number of threads for TensorFlow operations If None, the number of cpu of the current machine will be used.
-
action_probability(observation, state=None, mask=None, actions=None, logp=False)[source]¶ If
actionsisNone, then get the model’s action probability distribution from a given observation.- Depending on the action space the output is:
- Discrete: probability for each possible action
- Box: mean and standard deviation of the action output
However if
actionsis notNone, this function will return the probability that the given actions are taken with the given parameters (observation, state, …) on this model. For discrete action spaces, it returns the probability mass; for continuous action spaces, the probability density. This is since the probability mass will always be zero in continuous spaces, see http://blog.christianperone.com/2019/01/ for a good explanationParameters: - observation – (np.ndarray) the input observation
- state – (np.ndarray) The last states (can be None, used in recurrent policies)
- mask – (np.ndarray) The last masks (can be None, used in recurrent policies)
- actions – (np.ndarray) (OPTIONAL) For calculating the likelihood that the given actions are chosen by the model for each of the given parameters. Must have the same number of actions and observations. (set to None to return the complete action probability distribution)
- logp – (bool) (OPTIONAL) When specified with actions, returns probability in log-space. This has no effect if actions is None.
Returns: (np.ndarray) the model’s (log) action probability
-
get_env()[source]¶ returns the current environment (can be None if not defined)
Returns: (Gym Environment) The current environment
-
get_parameter_list()[source]¶ Get tensorflow Variables of model’s parameters
This includes all variables necessary for continuing training (saving / loading).
Returns: (list) List of tensorflow Variables
-
get_parameters()[source]¶ Get current model parameters as dictionary of variable name -> ndarray.
Returns: (OrderedDict) Dictionary of variable name -> ndarray of model’s parameters.
-
get_vec_normalize_env() → Optional[stable_baselines.common.vec_env.vec_normalize.VecNormalize][source]¶ Return the
VecNormalizewrapper of the training env if it exists.Returns: Optional[VecNormalize] The VecNormalizeenv.
-
learn(total_timesteps, callback=None, log_interval=100, tb_log_name='run', reset_num_timesteps=True)[source]¶ Return a trained model.
Parameters: - total_timesteps – (int) The total number of samples to train on
- callback – (Union[callable, [callable], BaseCallback]) function called at every steps with state of the algorithm. It takes the local and global variables. If it returns False, training is aborted. When the callback inherits from BaseCallback, you will have access to additional stages of the training (training start/end), please read the documentation for more details.
- log_interval – (int) The number of timesteps before logging.
- tb_log_name – (str) the name of the run for tensorboard log
- reset_num_timesteps – (bool) whether or not to reset the current timestep number (used in logging)
Returns: (BaseRLModel) the trained model
-
classmethod
load(load_path, env=None, custom_objects=None, **kwargs)[source]¶ Load the model from file
Parameters: - load_path – (str or file-like) the saved parameter location
- env – (Gym Environment) the new environment to run the loaded model on (can be None if you only need prediction from a trained model)
- custom_objects – (dict) Dictionary of objects to replace upon loading. If a variable is present in this dictionary as a key, it will not be deserialized and the corresponding item will be used instead. Similar to custom_objects in keras.models.load_model. Useful when you have an object in file that can not be deserialized.
- kwargs – extra arguments to change the model when loading
-
load_parameters(load_path_or_dict, exact_match=True)[source]¶ Load model parameters from a file or a dictionary
Dictionary keys should be tensorflow variable names, which can be obtained with
get_parametersfunction. Ifexact_matchis True, dictionary should contain keys for all model’s parameters, otherwise RunTimeError is raised. If False, only variables included in the dictionary will be updated.This does not load agent’s hyper-parameters.
Warning
This function does not update trainer/optimizer variables (e.g. momentum). As such training after using this function may lead to less-than-optimal results.
Parameters: - load_path_or_dict – (str or file-like or dict) Save parameter location or dict of parameters as variable.name -> ndarrays to be loaded.
- exact_match – (bool) If True, expects load dictionary to contain keys for all variables in the model. If False, loads parameters only for variables mentioned in the dictionary. Defaults to True.
-
predict(observation, state=None, mask=None, deterministic=False)[source]¶ Get the model’s action from an observation
Parameters: - observation – (np.ndarray) the input observation
- state – (np.ndarray) The last states (can be None, used in recurrent policies)
- mask – (np.ndarray) The last masks (can be None, used in recurrent policies)
- deterministic – (bool) Whether or not to return deterministic actions.
Returns: (np.ndarray, np.ndarray) the model’s action and the next state (used in recurrent policies)
-
pretrain(dataset, n_epochs=10, learning_rate=0.0001, adam_epsilon=1e-08, val_interval=None)[source]¶ Pretrain a model using behavior cloning: supervised learning given an expert dataset.
NOTE: only Box and Discrete spaces are supported for now.
Parameters: - dataset – (ExpertDataset) Dataset manager
- n_epochs – (int) Number of iterations on the training set
- learning_rate – (float) Learning rate
- adam_epsilon – (float) the epsilon value for the adam optimizer
- val_interval – (int) Report training and validation losses every n epochs. By default, every 10th of the maximum number of epochs.
Returns: (BaseRLModel) the pretrained model
-
save(save_path, cloudpickle=False)[source]¶ Save the current parameters to file
Parameters: - save_path – (str or file-like) The save location
- cloudpickle – (bool) Use older cloudpickle format instead of zip-archives.
-
set_env(env)[source]¶ Checks the validity of the environment, and if it is coherent, set it as the current environment.
Parameters: env – (Gym Environment) The environment for learning a policy
Policy Networks¶
Stable-baselines provides a set of default policies, that can be used with most action spaces.
To customize the default policies, you can specify the policy_kwargs parameter to the model class you use.
Those kwargs are then passed to the policy on instantiation (see Custom Policy Network for an example).
If you need more control on the policy architecture, you can also create a custom policy (see Custom Policy Network).
Note
CnnPolicies are for images only. MlpPolicies are made for other type of features (e.g. robot joints)
Warning
For all algorithms (except DDPG, TD3 and SAC), continuous actions are clipped during training and testing (to avoid out of bound error).
Available Policies
MlpPolicy |
Policy object that implements actor critic, using a MLP (2 layers of 64) |
MlpLstmPolicy |
Policy object that implements actor critic, using LSTMs with a MLP feature extraction |
MlpLnLstmPolicy |
Policy object that implements actor critic, using a layer normalized LSTMs with a MLP feature extraction |
CnnPolicy |
Policy object that implements actor critic, using a CNN (the nature CNN) |
CnnLstmPolicy |
Policy object that implements actor critic, using LSTMs with a CNN feature extraction |
CnnLnLstmPolicy |
Policy object that implements actor critic, using a layer normalized LSTMs with a CNN feature extraction |
Base Classes¶
-
class
stable_baselines.common.policies.BasePolicy(sess, ob_space, ac_space, n_env, n_steps, n_batch, reuse=False, scale=False, obs_phs=None, add_action_ph=False)[source]¶ The base policy object
Parameters: - sess – (TensorFlow session) The current TensorFlow session
- ob_space – (Gym Space) The observation space of the environment
- ac_space – (Gym Space) The action space of the environment
- n_env – (int) The number of environments to run
- n_steps – (int) The number of steps to run for each environment
- n_batch – (int) The number of batches to run (n_envs * n_steps)
- reuse – (bool) If the policy is reusable or not
- scale – (bool) whether or not to scale the input
- obs_phs – (TensorFlow Tensor, TensorFlow Tensor) a tuple containing an override for observation placeholder and the processed observation placeholder respectively
- add_action_ph – (bool) whether or not to create an action placeholder
-
action_ph¶ tf.Tensor: placeholder for actions, shape (self.n_batch, ) + self.ac_space.shape.
-
initial_state¶ The initial state of the policy. For feedforward policies, None. For a recurrent policy, a NumPy array of shape (self.n_env, ) + state_shape.
-
is_discrete¶ bool: is action space discrete.
-
obs_ph¶ tf.Tensor: placeholder for observations, shape (self.n_batch, ) + self.ob_space.shape.
-
proba_step(obs, state=None, mask=None)[source]¶ Returns the action probability for a single step
Parameters: - obs – ([float] or [int]) The current observation of the environment
- state – ([float]) The last states (used in recurrent policies)
- mask – ([float]) The last masks (used in recurrent policies)
Returns: ([float]) the action probability
-
processed_obs¶ tf.Tensor: processed observations, shape (self.n_batch, ) + self.ob_space.shape.
The form of processing depends on the type of the observation space, and the parameters whether scale is passed to the constructor; see observation_input for more information.
-
step(obs, state=None, mask=None)[source]¶ Returns the policy for a single step
Parameters: - obs – ([float] or [int]) The current observation of the environment
- state – ([float]) The last states (used in recurrent policies)
- mask – ([float]) The last masks (used in recurrent policies)
Returns: ([float], [float], [float], [float]) actions, values, states, neglogp
-
class
stable_baselines.common.policies.ActorCriticPolicy(sess, ob_space, ac_space, n_env, n_steps, n_batch, reuse=False, scale=False)[source]¶ Policy object that implements actor critic
Parameters: - sess – (TensorFlow session) The current TensorFlow session
- ob_space – (Gym Space) The observation space of the environment
- ac_space – (Gym Space) The action space of the environment
- n_env – (int) The number of environments to run
- n_steps – (int) The number of steps to run for each environment
- n_batch – (int) The number of batch to run (n_envs * n_steps)
- reuse – (bool) If the policy is reusable or not
- scale – (bool) whether or not to scale the input
-
action¶ tf.Tensor: stochastic action, of shape (self.n_batch, ) + self.ac_space.shape.
-
deterministic_action¶ tf.Tensor: deterministic action, of shape (self.n_batch, ) + self.ac_space.shape.
-
neglogp¶ tf.Tensor: negative log likelihood of the action sampled by self.action.
-
pdtype¶ ProbabilityDistributionType: type of the distribution for stochastic actions.
-
policy¶ tf.Tensor: policy output, e.g. logits.
-
policy_proba¶ tf.Tensor: parameters of the probability distribution. Depends on pdtype.
-
proba_distribution¶ ProbabilityDistribution: distribution of stochastic actions.
-
step(obs, state=None, mask=None, deterministic=False)[source]¶ Returns the policy for a single step
Parameters: - obs – ([float] or [int]) The current observation of the environment
- state – ([float]) The last states (used in recurrent policies)
- mask – ([float]) The last masks (used in recurrent policies)
- deterministic – (bool) Whether or not to return deterministic actions.
Returns: ([float], [float], [float], [float]) actions, values, states, neglogp
-
value(obs, state=None, mask=None)[source]¶ Returns the value for a single step
Parameters: - obs – ([float] or [int]) The current observation of the environment
- state – ([float]) The last states (used in recurrent policies)
- mask – ([float]) The last masks (used in recurrent policies)
Returns: ([float]) The associated value of the action
-
value_flat¶ tf.Tensor: value estimate, of shape (self.n_batch, )
-
value_fn¶ tf.Tensor: value estimate, of shape (self.n_batch, 1)
-
class
stable_baselines.common.policies.FeedForwardPolicy(sess, ob_space, ac_space, n_env, n_steps, n_batch, reuse=False, layers=None, net_arch=None, act_fun=<MagicMock id='140553670725984'>, cnn_extractor=<function nature_cnn>, feature_extraction='cnn', **kwargs)[source]¶ Policy object that implements actor critic, using a feed forward neural network.
Parameters: - sess – (TensorFlow session) The current TensorFlow session
- ob_space – (Gym Space) The observation space of the environment
- ac_space – (Gym Space) The action space of the environment
- n_env – (int) The number of environments to run
- n_steps – (int) The number of steps to run for each environment
- n_batch – (int) The number of batch to run (n_envs * n_steps)
- reuse – (bool) If the policy is reusable or not
- layers – ([int]) (deprecated, use net_arch instead) The size of the Neural network for the policy (if None, default to [64, 64])
- net_arch – (list) Specification of the actor-critic policy network architecture (see mlp_extractor documentation for details).
- act_fun – (tf.func) the activation function to use in the neural network.
- cnn_extractor – (function (TensorFlow Tensor,
**kwargs): (TensorFlow Tensor)) the CNN feature extraction - feature_extraction – (str) The feature extraction type (“cnn” or “mlp”)
- kwargs – (dict) Extra keyword arguments for the nature CNN feature extraction
-
proba_step(obs, state=None, mask=None)[source]¶ Returns the action probability for a single step
Parameters: - obs – ([float] or [int]) The current observation of the environment
- state – ([float]) The last states (used in recurrent policies)
- mask – ([float]) The last masks (used in recurrent policies)
Returns: ([float]) the action probability
-
step(obs, state=None, mask=None, deterministic=False)[source]¶ Returns the policy for a single step
Parameters: - obs – ([float] or [int]) The current observation of the environment
- state – ([float]) The last states (used in recurrent policies)
- mask – ([float]) The last masks (used in recurrent policies)
- deterministic – (bool) Whether or not to return deterministic actions.
Returns: ([float], [float], [float], [float]) actions, values, states, neglogp
-
value(obs, state=None, mask=None)[source]¶ Returns the value for a single step
Parameters: - obs – ([float] or [int]) The current observation of the environment
- state – ([float]) The last states (used in recurrent policies)
- mask – ([float]) The last masks (used in recurrent policies)
Returns: ([float]) The associated value of the action
-
class
stable_baselines.common.policies.LstmPolicy(sess, ob_space, ac_space, n_env, n_steps, n_batch, n_lstm=256, reuse=False, layers=None, net_arch=None, act_fun=<MagicMock id='140553670564496'>, cnn_extractor=<function nature_cnn>, layer_norm=False, feature_extraction='cnn', **kwargs)[source]¶ Policy object that implements actor critic, using LSTMs.
Parameters: - sess – (TensorFlow session) The current TensorFlow session
- ob_space – (Gym Space) The observation space of the environment
- ac_space – (Gym Space) The action space of the environment
- n_env – (int) The number of environments to run
- n_steps – (int) The number of steps to run for each environment
- n_batch – (int) The number of batch to run (n_envs * n_steps)
- n_lstm – (int) The number of LSTM cells (for recurrent policies)
- reuse – (bool) If the policy is reusable or not
- layers – ([int]) The size of the Neural network before the LSTM layer (if None, default to [64, 64])
- net_arch – (list) Specification of the actor-critic policy network architecture. Notation similar to the format described in mlp_extractor but with additional support for a ‘lstm’ entry in the shared network part.
- act_fun – (tf.func) the activation function to use in the neural network.
- cnn_extractor – (function (TensorFlow Tensor,
**kwargs): (TensorFlow Tensor)) the CNN feature extraction - layer_norm – (bool) Whether or not to use layer normalizing LSTMs
- feature_extraction – (str) The feature extraction type (“cnn” or “mlp”)
- kwargs – (dict) Extra keyword arguments for the nature CNN feature extraction
-
proba_step(obs, state=None, mask=None)[source]¶ Returns the action probability for a single step
Parameters: - obs – ([float] or [int]) The current observation of the environment
- state – ([float]) The last states (used in recurrent policies)
- mask – ([float]) The last masks (used in recurrent policies)
Returns: ([float]) the action probability
-
step(obs, state=None, mask=None, deterministic=False)[source]¶ Returns the policy for a single step
Parameters: - obs – ([float] or [int]) The current observation of the environment
- state – ([float]) The last states (used in recurrent policies)
- mask – ([float]) The last masks (used in recurrent policies)
- deterministic – (bool) Whether or not to return deterministic actions.
Returns: ([float], [float], [float], [float]) actions, values, states, neglogp
MLP Policies¶
-
class
stable_baselines.common.policies.MlpPolicy(sess, ob_space, ac_space, n_env, n_steps, n_batch, reuse=False, **_kwargs)[source]¶ Policy object that implements actor critic, using a MLP (2 layers of 64)
Parameters: - sess – (TensorFlow session) The current TensorFlow session
- ob_space – (Gym Space) The observation space of the environment
- ac_space – (Gym Space) The action space of the environment
- n_env – (int) The number of environments to run
- n_steps – (int) The number of steps to run for each environment
- n_batch – (int) The number of batch to run (n_envs * n_steps)
- reuse – (bool) If the policy is reusable or not
- _kwargs – (dict) Extra keyword arguments for the nature CNN feature extraction
-
class
stable_baselines.common.policies.MlpLstmPolicy(sess, ob_space, ac_space, n_env, n_steps, n_batch, n_lstm=256, reuse=False, **_kwargs)[source]¶ Policy object that implements actor critic, using LSTMs with a MLP feature extraction
Parameters: - sess – (TensorFlow session) The current TensorFlow session
- ob_space – (Gym Space) The observation space of the environment
- ac_space – (Gym Space) The action space of the environment
- n_env – (int) The number of environments to run
- n_steps – (int) The number of steps to run for each environment
- n_batch – (int) The number of batch to run (n_envs * n_steps)
- n_lstm – (int) The number of LSTM cells (for recurrent policies)
- reuse – (bool) If the policy is reusable or not
- kwargs – (dict) Extra keyword arguments for the nature CNN feature extraction
-
class
stable_baselines.common.policies.MlpLnLstmPolicy(sess, ob_space, ac_space, n_env, n_steps, n_batch, n_lstm=256, reuse=False, **_kwargs)[source]¶ Policy object that implements actor critic, using a layer normalized LSTMs with a MLP feature extraction
Parameters: - sess – (TensorFlow session) The current TensorFlow session
- ob_space – (Gym Space) The observation space of the environment
- ac_space – (Gym Space) The action space of the environment
- n_env – (int) The number of environments to run
- n_steps – (int) The number of steps to run for each environment
- n_batch – (int) The number of batch to run (n_envs * n_steps)
- n_lstm – (int) The number of LSTM cells (for recurrent policies)
- reuse – (bool) If the policy is reusable or not
- kwargs – (dict) Extra keyword arguments for the nature CNN feature extraction
CNN Policies¶
-
class
stable_baselines.common.policies.CnnPolicy(sess, ob_space, ac_space, n_env, n_steps, n_batch, reuse=False, **_kwargs)[source]¶ Policy object that implements actor critic, using a CNN (the nature CNN)
Parameters: - sess – (TensorFlow session) The current TensorFlow session
- ob_space – (Gym Space) The observation space of the environment
- ac_space – (Gym Space) The action space of the environment
- n_env – (int) The number of environments to run
- n_steps – (int) The number of steps to run for each environment
- n_batch – (int) The number of batch to run (n_envs * n_steps)
- reuse – (bool) If the policy is reusable or not
- _kwargs – (dict) Extra keyword arguments for the nature CNN feature extraction
-
class
stable_baselines.common.policies.CnnLstmPolicy(sess, ob_space, ac_space, n_env, n_steps, n_batch, n_lstm=256, reuse=False, **_kwargs)[source]¶ Policy object that implements actor critic, using LSTMs with a CNN feature extraction
Parameters: - sess – (TensorFlow session) The current TensorFlow session
- ob_space – (Gym Space) The observation space of the environment
- ac_space – (Gym Space) The action space of the environment
- n_env – (int) The number of environments to run
- n_steps – (int) The number of steps to run for each environment
- n_batch – (int) The number of batch to run (n_envs * n_steps)
- n_lstm – (int) The number of LSTM cells (for recurrent policies)
- reuse – (bool) If the policy is reusable or not
- kwargs – (dict) Extra keyword arguments for the nature CNN feature extraction
-
class
stable_baselines.common.policies.CnnLnLstmPolicy(sess, ob_space, ac_space, n_env, n_steps, n_batch, n_lstm=256, reuse=False, **_kwargs)[source]¶ Policy object that implements actor critic, using a layer normalized LSTMs with a CNN feature extraction
Parameters: - sess – (TensorFlow session) The current TensorFlow session
- ob_space – (Gym Space) The observation space of the environment
- ac_space – (Gym Space) The action space of the environment
- n_env – (int) The number of environments to run
- n_steps – (int) The number of steps to run for each environment
- n_batch – (int) The number of batch to run (n_envs * n_steps)
- n_lstm – (int) The number of LSTM cells (for recurrent policies)
- reuse – (bool) If the policy is reusable or not
- kwargs – (dict) Extra keyword arguments for the nature CNN feature extraction
A2C¶
A synchronous, deterministic variant of Asynchronous Advantage Actor Critic (A3C). It uses multiple workers to avoid the use of a replay buffer.
Notes¶
- Original paper: https://arxiv.org/abs/1602.01783
- OpenAI blog post: https://openai.com/blog/baselines-acktr-a2c/
python -m stable_baselines.a2c.run_atariruns the algorithm for 40M frames = 10M timesteps on an Atari game. See help (-h) for more options.python -m stable_baselines.a2c.run_mujocoruns the algorithm for 1M frames on a Mujoco environment.
Can I use?¶
- Recurrent policies: ✔️
- Multi processing: ✔️
- Gym spaces:
| Space | Action | Observation |
|---|---|---|
| Discrete | ✔️ | ✔️ |
| Box | ✔️ | ✔️ |
| MultiDiscrete | ✔️ | ✔️ |
| MultiBinary | ✔️ | ✔️ |
Example¶
Train a A2C agent on CartPole-v1 using 4 processes.
import gym
from stable_baselines.common.policies import MlpPolicy
from stable_baselines.common import make_vec_env
from stable_baselines import A2C
# Parallel environments
env = make_vec_env('CartPole-v1', n_envs=4)
model = A2C(MlpPolicy, env, verbose=1)
model.learn(total_timesteps=25000)
model.save("a2c_cartpole")
del model # remove to demonstrate saving and loading
model = A2C.load("a2c_cartpole")
obs = env.reset()
while True:
action, _states = model.predict(obs)
obs, rewards, dones, info = env.step(action)
env.render()
Parameters¶
-
class
stable_baselines.a2c.A2C(policy, env, gamma=0.99, n_steps=5, vf_coef=0.25, ent_coef=0.01, max_grad_norm=0.5, learning_rate=0.0007, alpha=0.99, epsilon=1e-05, lr_schedule='constant', verbose=0, tensorboard_log=None, _init_setup_model=True, policy_kwargs=None, full_tensorboard_log=False, seed=None, n_cpu_tf_sess=None)[source]¶ The A2C (Advantage Actor Critic) model class, https://arxiv.org/abs/1602.01783
Parameters: - policy – (ActorCriticPolicy or str) The policy model to use (MlpPolicy, CnnPolicy, CnnLstmPolicy, …)
- env – (Gym environment or str) The environment to learn from (if registered in Gym, can be str)
- gamma – (float) Discount factor
- n_steps – (int) The number of steps to run for each environment per update (i.e. batch size is n_steps * n_env where n_env is number of environment copies running in parallel)
- vf_coef – (float) Value function coefficient for the loss calculation
- ent_coef – (float) Entropy coefficient for the loss calculation
- max_grad_norm – (float) The maximum value for the gradient clipping
- learning_rate – (float) The learning rate
- alpha – (float) RMSProp decay parameter (default: 0.99)
- epsilon – (float) RMSProp epsilon (stabilizes square root computation in denominator of RMSProp update) (default: 1e-5)
- lr_schedule – (str) The type of scheduler for the learning rate update (‘linear’, ‘constant’, ‘double_linear_con’, ‘middle_drop’ or ‘double_middle_drop’)
- verbose – (int) the verbosity level: 0 none, 1 training information, 2 tensorflow debug
- tensorboard_log – (str) the log location for tensorboard (if None, no logging)
- _init_setup_model – (bool) Whether or not to build the network at the creation of the instance (used only for loading)
- policy_kwargs – (dict) additional arguments to be passed to the policy on creation
- full_tensorboard_log – (bool) enable additional logging when using tensorboard WARNING: this logging can take a lot of space quickly
- seed – (int) Seed for the pseudo-random generators (python, numpy, tensorflow). If None (default), use random seed. Note that if you want completely deterministic results, you must set n_cpu_tf_sess to 1.
- n_cpu_tf_sess – (int) The number of threads for TensorFlow operations If None, the number of cpu of the current machine will be used.
-
action_probability(observation, state=None, mask=None, actions=None, logp=False)¶ If
actionsisNone, then get the model’s action probability distribution from a given observation.- Depending on the action space the output is:
- Discrete: probability for each possible action
- Box: mean and standard deviation of the action output
However if
actionsis notNone, this function will return the probability that the given actions are taken with the given parameters (observation, state, …) on this model. For discrete action spaces, it returns the probability mass; for continuous action spaces, the probability density. This is since the probability mass will always be zero in continuous spaces, see http://blog.christianperone.com/2019/01/ for a good explanationParameters: - observation – (np.ndarray) the input observation
- state – (np.ndarray) The last states (can be None, used in recurrent policies)
- mask – (np.ndarray) The last masks (can be None, used in recurrent policies)
- actions – (np.ndarray) (OPTIONAL) For calculating the likelihood that the given actions are chosen by the model for each of the given parameters. Must have the same number of actions and observations. (set to None to return the complete action probability distribution)
- logp – (bool) (OPTIONAL) When specified with actions, returns probability in log-space. This has no effect if actions is None.
Returns: (np.ndarray) the model’s (log) action probability
-
get_env()¶ returns the current environment (can be None if not defined)
Returns: (Gym Environment) The current environment
-
get_parameter_list()¶ Get tensorflow Variables of model’s parameters
This includes all variables necessary for continuing training (saving / loading).
Returns: (list) List of tensorflow Variables
-
get_parameters()¶ Get current model parameters as dictionary of variable name -> ndarray.
Returns: (OrderedDict) Dictionary of variable name -> ndarray of model’s parameters.
-
get_vec_normalize_env() → Optional[stable_baselines.common.vec_env.vec_normalize.VecNormalize]¶ Return the
VecNormalizewrapper of the training env if it exists.Returns: Optional[VecNormalize] The VecNormalizeenv.
-
learn(total_timesteps, callback=None, log_interval=100, tb_log_name='A2C', reset_num_timesteps=True)[source]¶ Return a trained model.
Parameters: - total_timesteps – (int) The total number of samples to train on
- callback – (Union[callable, [callable], BaseCallback]) function called at every steps with state of the algorithm. It takes the local and global variables. If it returns False, training is aborted. When the callback inherits from BaseCallback, you will have access to additional stages of the training (training start/end), please read the documentation for more details.
- log_interval – (int) The number of timesteps before logging.
- tb_log_name – (str) the name of the run for tensorboard log
- reset_num_timesteps – (bool) whether or not to reset the current timestep number (used in logging)
Returns: (BaseRLModel) the trained model
-
classmethod
load(load_path, env=None, custom_objects=None, **kwargs)¶ Load the model from file
Parameters: - load_path – (str or file-like) the saved parameter location
- env – (Gym Environment) the new environment to run the loaded model on (can be None if you only need prediction from a trained model)
- custom_objects – (dict) Dictionary of objects to replace upon loading. If a variable is present in this dictionary as a key, it will not be deserialized and the corresponding item will be used instead. Similar to custom_objects in keras.models.load_model. Useful when you have an object in file that can not be deserialized.
- kwargs – extra arguments to change the model when loading
-
load_parameters(load_path_or_dict, exact_match=True)¶ Load model parameters from a file or a dictionary
Dictionary keys should be tensorflow variable names, which can be obtained with
get_parametersfunction. Ifexact_matchis True, dictionary should contain keys for all model’s parameters, otherwise RunTimeError is raised. If False, only variables included in the dictionary will be updated.This does not load agent’s hyper-parameters.
Warning
This function does not update trainer/optimizer variables (e.g. momentum). As such training after using this function may lead to less-than-optimal results.
Parameters: - load_path_or_dict – (str or file-like or dict) Save parameter location or dict of parameters as variable.name -> ndarrays to be loaded.
- exact_match – (bool) If True, expects load dictionary to contain keys for all variables in the model. If False, loads parameters only for variables mentioned in the dictionary. Defaults to True.
-
predict(observation, state=None, mask=None, deterministic=False)¶ Get the model’s action from an observation
Parameters: - observation – (np.ndarray) the input observation
- state – (np.ndarray) The last states (can be None, used in recurrent policies)
- mask – (np.ndarray) The last masks (can be None, used in recurrent policies)
- deterministic – (bool) Whether or not to return deterministic actions.
Returns: (np.ndarray, np.ndarray) the model’s action and the next state (used in recurrent policies)
-
pretrain(dataset, n_epochs=10, learning_rate=0.0001, adam_epsilon=1e-08, val_interval=None)¶ Pretrain a model using behavior cloning: supervised learning given an expert dataset.
NOTE: only Box and Discrete spaces are supported for now.
Parameters: - dataset – (ExpertDataset) Dataset manager
- n_epochs – (int) Number of iterations on the training set
- learning_rate – (float) Learning rate
- adam_epsilon – (float) the epsilon value for the adam optimizer
- val_interval – (int) Report training and validation losses every n epochs. By default, every 10th of the maximum number of epochs.
Returns: (BaseRLModel) the pretrained model
-
save(save_path, cloudpickle=False)[source]¶ Save the current parameters to file
Parameters: - save_path – (str or file-like) The save location
- cloudpickle – (bool) Use older cloudpickle format instead of zip-archives.
-
set_env(env)¶ Checks the validity of the environment, and if it is coherent, set it as the current environment.
Parameters: env – (Gym Environment) The environment for learning a policy
-
set_random_seed(seed: Optional[int]) → None¶ Parameters: seed – (Optional[int]) Seed for the pseudo-random generators. If None, do not change the seeds.
ACER¶
Sample Efficient Actor-Critic with Experience Replay (ACER) combines several ideas of previous algorithms: it uses multiple workers (as A2C), implements a replay buffer (as in DQN), uses Retrace for Q-value estimation, importance sampling and a trust region.
Notes¶
- Original paper: https://arxiv.org/abs/1611.01224
python -m stable_baselines.acer.run_atariruns the algorithm for 40M frames = 10M timesteps on an Atari game. See help (-h) for more options.
Can I use?¶
- Recurrent policies: ✔️
- Multi processing: ✔️
- Gym spaces:
| Space | Action | Observation |
|---|---|---|
| Discrete | ✔️ | ✔️ |
| Box | ❌ | ✔️ |
| MultiDiscrete | ❌ | ✔️ |
| MultiBinary | ❌ | ✔️ |
Example¶
import gym
from stable_baselines.common.policies import MlpPolicy, MlpLstmPolicy, MlpLnLstmPolicy
from stable_baselines.common import make_vec_env
from stable_baselines import ACER
# multiprocess environment
env = make_vec_env('CartPole-v1', n_envs=4)
model = ACER(MlpPolicy, env, verbose=1)
model.learn(total_timesteps=25000)
model.save("acer_cartpole")
del model # remove to demonstrate saving and loading
model = ACER.load("acer_cartpole")
obs = env.reset()
while True:
action, _states = model.predict(obs)
obs, rewards, dones, info = env.step(action)
env.render()
Parameters¶
-
class
stable_baselines.acer.ACER(policy, env, gamma=0.99, n_steps=20, num_procs=None, q_coef=0.5, ent_coef=0.01, max_grad_norm=10, learning_rate=0.0007, lr_schedule='linear', rprop_alpha=0.99, rprop_epsilon=1e-05, buffer_size=5000, replay_ratio=4, replay_start=1000, correction_term=10.0, trust_region=True, alpha=0.99, delta=1, verbose=0, tensorboard_log=None, _init_setup_model=True, policy_kwargs=None, full_tensorboard_log=False, seed=None, n_cpu_tf_sess=1)[source]¶ The ACER (Actor-Critic with Experience Replay) model class, https://arxiv.org/abs/1611.01224
Parameters: - policy – (ActorCriticPolicy or str) The policy model to use (MlpPolicy, CnnPolicy, CnnLstmPolicy, …)
- env – (Gym environment or str) The environment to learn from (if registered in Gym, can be str)
- gamma – (float) The discount value
- n_steps – (int) The number of steps to run for each environment per update (i.e. batch size is n_steps * n_env where n_env is number of environment copies running in parallel)
- num_procs –
(int) The number of threads for TensorFlow operations
Deprecated since version 2.9.0: Use n_cpu_tf_sess instead.
- q_coef – (float) The weight for the loss on the Q value
- ent_coef – (float) The weight for the entropy loss
- max_grad_norm – (float) The clipping value for the maximum gradient
- learning_rate – (float) The initial learning rate for the RMS prop optimizer
- lr_schedule – (str) The type of scheduler for the learning rate update (‘linear’, ‘constant’, ‘double_linear_con’, ‘middle_drop’ or ‘double_middle_drop’)
- rprop_epsilon – (float) RMSProp epsilon (stabilizes square root computation in denominator of RMSProp update) (default: 1e-5)
- rprop_alpha – (float) RMSProp decay parameter (default: 0.99)
- buffer_size – (int) The buffer size in number of steps
- replay_ratio – (float) The number of replay learning per on policy learning on average, using a poisson distribution
- replay_start – (int) The minimum number of steps in the buffer, before learning replay
- correction_term – (float) Importance weight clipping factor (default: 10)
- trust_region – (bool) Whether or not algorithms estimates the gradient KL divergence between the old and updated policy and uses it to determine step size (default: True)
- alpha – (float) The decay rate for the Exponential moving average of the parameters
- delta – (float) max KL divergence between the old policy and updated policy (default: 1)
- verbose – (int) the verbosity level: 0 none, 1 training information, 2 tensorflow debug
- tensorboard_log – (str) the log location for tensorboard (if None, no logging)
- _init_setup_model – (bool) Whether or not to build the network at the creation of the instance
- policy_kwargs – (dict) additional arguments to be passed to the policy on creation
- full_tensorboard_log – (bool) enable additional logging when using tensorboard WARNING: this logging can take a lot of space quickly
- seed – (int) Seed for the pseudo-random generators (python, numpy, tensorflow). If None (default), use random seed. Note that if you want completely deterministic results, you must set n_cpu_tf_sess to 1.
- n_cpu_tf_sess – (int) The number of threads for TensorFlow operations If None, the number of cpu of the current machine will be used.
-
action_probability(observation, state=None, mask=None, actions=None, logp=False)¶ If
actionsisNone, then get the model’s action probability distribution from a given observation.- Depending on the action space the output is:
- Discrete: probability for each possible action
- Box: mean and standard deviation of the action output
However if
actionsis notNone, this function will return the probability that the given actions are taken with the given parameters (observation, state, …) on this model. For discrete action spaces, it returns the probability mass; for continuous action spaces, the probability density. This is since the probability mass will always be zero in continuous spaces, see http://blog.christianperone.com/2019/01/ for a good explanationParameters: - observation – (np.ndarray) the input observation
- state – (np.ndarray) The last states (can be None, used in recurrent policies)
- mask – (np.ndarray) The last masks (can be None, used in recurrent policies)
- actions – (np.ndarray) (OPTIONAL) For calculating the likelihood that the given actions are chosen by the model for each of the given parameters. Must have the same number of actions and observations. (set to None to return the complete action probability distribution)
- logp – (bool) (OPTIONAL) When specified with actions, returns probability in log-space. This has no effect if actions is None.
Returns: (np.ndarray) the model’s (log) action probability
-
get_env()¶ returns the current environment (can be None if not defined)
Returns: (Gym Environment) The current environment
-
get_parameter_list()¶ Get tensorflow Variables of model’s parameters
This includes all variables necessary for continuing training (saving / loading).
Returns: (list) List of tensorflow Variables
-
get_parameters()¶ Get current model parameters as dictionary of variable name -> ndarray.
Returns: (OrderedDict) Dictionary of variable name -> ndarray of model’s parameters.
-
get_vec_normalize_env() → Optional[stable_baselines.common.vec_env.vec_normalize.VecNormalize]¶ Return the
VecNormalizewrapper of the training env if it exists.Returns: Optional[VecNormalize] The VecNormalizeenv.
-
learn(total_timesteps, callback=None, log_interval=100, tb_log_name='ACER', reset_num_timesteps=True)[source]¶ Return a trained model.
Parameters: - total_timesteps – (int) The total number of samples to train on
- callback – (Union[callable, [callable], BaseCallback]) function called at every steps with state of the algorithm. It takes the local and global variables. If it returns False, training is aborted. When the callback inherits from BaseCallback, you will have access to additional stages of the training (training start/end), please read the documentation for more details.
- log_interval – (int) The number of timesteps before logging.
- tb_log_name – (str) the name of the run for tensorboard log
- reset_num_timesteps – (bool) whether or not to reset the current timestep number (used in logging)
Returns: (BaseRLModel) the trained model
-
classmethod
load(load_path, env=None, custom_objects=None, **kwargs)¶ Load the model from file
Parameters: - load_path – (str or file-like) the saved parameter location
- env – (Gym Environment) the new environment to run the loaded model on (can be None if you only need prediction from a trained model)
- custom_objects – (dict) Dictionary of objects to replace upon loading. If a variable is present in this dictionary as a key, it will not be deserialized and the corresponding item will be used instead. Similar to custom_objects in keras.models.load_model. Useful when you have an object in file that can not be deserialized.
- kwargs – extra arguments to change the model when loading
-
load_parameters(load_path_or_dict, exact_match=True)¶ Load model parameters from a file or a dictionary
Dictionary keys should be tensorflow variable names, which can be obtained with
get_parametersfunction. Ifexact_matchis True, dictionary should contain keys for all model’s parameters, otherwise RunTimeError is raised. If False, only variables included in the dictionary will be updated.This does not load agent’s hyper-parameters.
Warning
This function does not update trainer/optimizer variables (e.g. momentum). As such training after using this function may lead to less-than-optimal results.
Parameters: - load_path_or_dict – (str or file-like or dict) Save parameter location or dict of parameters as variable.name -> ndarrays to be loaded.
- exact_match – (bool) If True, expects load dictionary to contain keys for all variables in the model. If False, loads parameters only for variables mentioned in the dictionary. Defaults to True.
-
predict(observation, state=None, mask=None, deterministic=False)¶ Get the model’s action from an observation
Parameters: - observation – (np.ndarray) the input observation
- state – (np.ndarray) The last states (can be None, used in recurrent policies)
- mask – (np.ndarray) The last masks (can be None, used in recurrent policies)
- deterministic – (bool) Whether or not to return deterministic actions.
Returns: (np.ndarray, np.ndarray) the model’s action and the next state (used in recurrent policies)
-
pretrain(dataset, n_epochs=10, learning_rate=0.0001, adam_epsilon=1e-08, val_interval=None)¶ Pretrain a model using behavior cloning: supervised learning given an expert dataset.
NOTE: only Box and Discrete spaces are supported for now.
Parameters: - dataset – (ExpertDataset) Dataset manager
- n_epochs – (int) Number of iterations on the training set
- learning_rate – (float) Learning rate
- adam_epsilon – (float) the epsilon value for the adam optimizer
- val_interval – (int) Report training and validation losses every n epochs. By default, every 10th of the maximum number of epochs.
Returns: (BaseRLModel) the pretrained model
-
save(save_path, cloudpickle=False)[source]¶ Save the current parameters to file
Parameters: - save_path – (str or file-like) The save location
- cloudpickle – (bool) Use older cloudpickle format instead of zip-archives.
-
set_env(env)[source]¶ Checks the validity of the environment, and if it is coherent, set it as the current environment.
Parameters: env – (Gym Environment) The environment for learning a policy
-
set_random_seed(seed: Optional[int]) → None¶ Parameters: seed – (Optional[int]) Seed for the pseudo-random generators. If None, do not change the seeds.
ACKTR¶
Actor Critic using Kronecker-Factored Trust Region (ACKTR) uses Kronecker-factored approximate curvature (K-FAC) for trust region optimization.
Notes¶
- Original paper: https://arxiv.org/abs/1708.05144
- Baselines blog post: https://blog.openai.com/baselines-acktr-a2c/
python -m stable_baselines.acktr.run_atariruns the algorithm for 40M frames = 10M timesteps on an Atari game. See help (-h) for more options.
Can I use?¶
- Recurrent policies: ✔️
- Multi processing: ✔️
- Gym spaces:
| Space | Action | Observation |
|---|---|---|
| Discrete | ✔️ | ✔️ |
| Box | ✔️ | ✔️ |
| MultiDiscrete | ❌ | ✔️ |
| MultiBinary | ❌ | ✔️ |
Example¶
import gym
from stable_baselines.common.policies import MlpPolicy, MlpLstmPolicy, MlpLnLstmPolicy
from stable_baselines.common import make_vec_env
from stable_baselines import ACKTR
# multiprocess environment
env = make_vec_env('CartPole-v1', n_envs=4)
model = ACKTR(MlpPolicy, env, verbose=1)
model.learn(total_timesteps=25000)
model.save("acktr_cartpole")
del model # remove to demonstrate saving and loading
model = ACKTR.load("acktr_cartpole")
obs = env.reset()
while True:
action, _states = model.predict(obs)
obs, rewards, dones, info = env.step(action)
env.render()
Parameters¶
-
class
stable_baselines.acktr.ACKTR(policy, env, gamma=0.99, nprocs=None, n_steps=20, ent_coef=0.01, vf_coef=0.25, vf_fisher_coef=1.0, learning_rate=0.25, max_grad_norm=0.5, kfac_clip=0.001, lr_schedule='linear', verbose=0, tensorboard_log=None, _init_setup_model=True, async_eigen_decomp=False, kfac_update=1, gae_lambda=None, policy_kwargs=None, full_tensorboard_log=False, seed=None, n_cpu_tf_sess=1)[source]¶ The ACKTR (Actor Critic using Kronecker-Factored Trust Region) model class, https://arxiv.org/abs/1708.05144
Parameters: - policy – (ActorCriticPolicy or str) The policy model to use (MlpPolicy, CnnPolicy, CnnLstmPolicy, …)
- env – (Gym environment or str) The environment to learn from (if registered in Gym, can be str)
- gamma – (float) Discount factor
- nprocs –
(int) The number of threads for TensorFlow operations
Deprecated since version 2.9.0: Use n_cpu_tf_sess instead.
- n_steps – (int) The number of steps to run for each environment
- ent_coef – (float) The weight for the entropy loss
- vf_coef – (float) The weight for the loss on the value function
- vf_fisher_coef – (float) The weight for the fisher loss on the value function
- learning_rate – (float) The initial learning rate for the RMS prop optimizer
- max_grad_norm – (float) The clipping value for the maximum gradient
- kfac_clip – (float) gradient clipping for Kullback-Leibler
- lr_schedule – (str) The type of scheduler for the learning rate update (‘linear’, ‘constant’, ‘double_linear_con’, ‘middle_drop’ or ‘double_middle_drop’)
- verbose – (int) the verbosity level: 0 none, 1 training information, 2 tensorflow debug
- tensorboard_log – (str) the log location for tensorboard (if None, no logging)
- _init_setup_model – (bool) Whether or not to build the network at the creation of the instance
- async_eigen_decomp – (bool) Use async eigen decomposition
- kfac_update – (int) update kfac after kfac_update steps
- policy_kwargs – (dict) additional arguments to be passed to the policy on creation
- gae_lambda – (float) Factor for trade-off of bias vs variance for Generalized Advantage Estimator If None (default), then the classic advantage will be used instead of GAE
- full_tensorboard_log – (bool) enable additional logging when using tensorboard WARNING: this logging can take a lot of space quickly
- seed – (int) Seed for the pseudo-random generators (python, numpy, tensorflow). If None (default), use random seed. Note that if you want completely deterministic results, you must set n_cpu_tf_sess to 1.
- n_cpu_tf_sess – (int) The number of threads for TensorFlow operations If None, the number of cpu of the current machine will be used.
-
action_probability(observation, state=None, mask=None, actions=None, logp=False)¶ If
actionsisNone, then get the model’s action probability distribution from a given observation.- Depending on the action space the output is:
- Discrete: probability for each possible action
- Box: mean and standard deviation of the action output
However if
actionsis notNone, this function will return the probability that the given actions are taken with the given parameters (observation, state, …) on this model. For discrete action spaces, it returns the probability mass; for continuous action spaces, the probability density. This is since the probability mass will always be zero in continuous spaces, see http://blog.christianperone.com/2019/01/ for a good explanationParameters: - observation – (np.ndarray) the input observation
- state – (np.ndarray) The last states (can be None, used in recurrent policies)
- mask – (np.ndarray) The last masks (can be None, used in recurrent policies)
- actions – (np.ndarray) (OPTIONAL) For calculating the likelihood that the given actions are chosen by the model for each of the given parameters. Must have the same number of actions and observations. (set to None to return the complete action probability distribution)
- logp – (bool) (OPTIONAL) When specified with actions, returns probability in log-space. This has no effect if actions is None.
Returns: (np.ndarray) the model’s (log) action probability
-
get_env()¶ returns the current environment (can be None if not defined)
Returns: (Gym Environment) The current environment
-
get_parameter_list()¶ Get tensorflow Variables of model’s parameters
This includes all variables necessary for continuing training (saving / loading).
Returns: (list) List of tensorflow Variables
-
get_parameters()¶ Get current model parameters as dictionary of variable name -> ndarray.
Returns: (OrderedDict) Dictionary of variable name -> ndarray of model’s parameters.
-
get_vec_normalize_env() → Optional[stable_baselines.common.vec_env.vec_normalize.VecNormalize]¶ Return the
VecNormalizewrapper of the training env if it exists.Returns: Optional[VecNormalize] The VecNormalizeenv.
-
learn(total_timesteps, callback=None, log_interval=100, tb_log_name='ACKTR', reset_num_timesteps=True)[source]¶ Return a trained model.
Parameters: - total_timesteps – (int) The total number of samples to train on
- callback – (Union[callable, [callable], BaseCallback]) function called at every steps with state of the algorithm. It takes the local and global variables. If it returns False, training is aborted. When the callback inherits from BaseCallback, you will have access to additional stages of the training (training start/end), please read the documentation for more details.
- log_interval – (int) The number of timesteps before logging.
- tb_log_name – (str) the name of the run for tensorboard log
- reset_num_timesteps – (bool) whether or not to reset the current timestep number (used in logging)
Returns: (BaseRLModel) the trained model
-
classmethod
load(load_path, env=None, custom_objects=None, **kwargs)¶ Load the model from file
Parameters: - load_path – (str or file-like) the saved parameter location
- env – (Gym Environment) the new environment to run the loaded model on (can be None if you only need prediction from a trained model)
- custom_objects – (dict) Dictionary of objects to replace upon loading. If a variable is present in this dictionary as a key, it will not be deserialized and the corresponding item will be used instead. Similar to custom_objects in keras.models.load_model. Useful when you have an object in file that can not be deserialized.
- kwargs – extra arguments to change the model when loading
-
load_parameters(load_path_or_dict, exact_match=True)¶ Load model parameters from a file or a dictionary
Dictionary keys should be tensorflow variable names, which can be obtained with
get_parametersfunction. Ifexact_matchis True, dictionary should contain keys for all model’s parameters, otherwise RunTimeError is raised. If False, only variables included in the dictionary will be updated.This does not load agent’s hyper-parameters.
Warning
This function does not update trainer/optimizer variables (e.g. momentum). As such training after using this function may lead to less-than-optimal results.
Parameters: - load_path_or_dict – (str or file-like or dict) Save parameter location or dict of parameters as variable.name -> ndarrays to be loaded.
- exact_match – (bool) If True, expects load dictionary to contain keys for all variables in the model. If False, loads parameters only for variables mentioned in the dictionary. Defaults to True.
-
predict(observation, state=None, mask=None, deterministic=False)¶ Get the model’s action from an observation
Parameters: - observation – (np.ndarray) the input observation
- state – (np.ndarray) The last states (can be None, used in recurrent policies)
- mask – (np.ndarray) The last masks (can be None, used in recurrent policies)
- deterministic – (bool) Whether or not to return deterministic actions.
Returns: (np.ndarray, np.ndarray) the model’s action and the next state (used in recurrent policies)
-
pretrain(dataset, n_epochs=10, learning_rate=0.0001, adam_epsilon=1e-08, val_interval=None)¶ Pretrain a model using behavior cloning: supervised learning given an expert dataset.
NOTE: only Box and Discrete spaces are supported for now.
Parameters: - dataset – (ExpertDataset) Dataset manager
- n_epochs – (int) Number of iterations on the training set
- learning_rate – (float) Learning rate
- adam_epsilon – (float) the epsilon value for the adam optimizer
- val_interval – (int) Report training and validation losses every n epochs. By default, every 10th of the maximum number of epochs.
Returns: (BaseRLModel) the pretrained model
-
save(save_path, cloudpickle=False)[source]¶ Save the current parameters to file
Parameters: - save_path – (str or file-like) The save location
- cloudpickle – (bool) Use older cloudpickle format instead of zip-archives.
-
set_env(env)¶ Checks the validity of the environment, and if it is coherent, set it as the current environment.
Parameters: env – (Gym Environment) The environment for learning a policy
-
set_random_seed(seed: Optional[int]) → None¶ Parameters: seed – (Optional[int]) Seed for the pseudo-random generators. If None, do not change the seeds.
DDPG¶
Deep Deterministic Policy Gradient (DDPG)
Note
DDPG requires OpenMPI. If OpenMPI isn’t enabled, then DDPG isn’t
imported into the stable_baselines module.
Warning
The DDPG model does not support stable_baselines.common.policies because it uses q-value instead
of value estimation, as a result it must use its own policy models (see DDPG Policies).
Available Policies
MlpPolicy |
Policy object that implements actor critic, using a MLP (2 layers of 64) |
LnMlpPolicy |
Policy object that implements actor critic, using a MLP (2 layers of 64), with layer normalisation |
CnnPolicy |
Policy object that implements actor critic, using a CNN (the nature CNN) |
LnCnnPolicy |
Policy object that implements actor critic, using a CNN (the nature CNN), with layer normalisation |
Notes¶
- Original paper: https://arxiv.org/abs/1509.02971
- Baselines post: https://blog.openai.com/better-exploration-with-parameter-noise/
python -m stable_baselines.ddpg.mainruns the algorithm for 1M frames = 10M timesteps on a Mujoco environment. See help (-h) for more options.
Can I use?¶
- Recurrent policies: ❌
- Multi processing: ✔️ (using MPI)
- Gym spaces:
| Space | Action | Observation |
|---|---|---|
| Discrete | ❌ | ✔️ |
| Box | ✔️ | ✔️ |
| MultiDiscrete | ❌ | ✔️ |
| MultiBinary | ❌ | ✔️ |
Example¶
import gym
import numpy as np
from stable_baselines.ddpg.policies import MlpPolicy
from stable_baselines.common.noise import NormalActionNoise, OrnsteinUhlenbeckActionNoise, AdaptiveParamNoiseSpec
from stable_baselines import DDPG
env = gym.make('MountainCarContinuous-v0')
# the noise objects for DDPG
n_actions = env.action_space.shape[-1]
param_noise = None
action_noise = OrnsteinUhlenbeckActionNoise(mean=np.zeros(n_actions), sigma=float(0.5) * np.ones(n_actions))
model = DDPG(MlpPolicy, env, verbose=1, param_noise=param_noise, action_noise=action_noise)
model.learn(total_timesteps=400000)
model.save("ddpg_mountain")
del model # remove to demonstrate saving and loading
model = DDPG.load("ddpg_mountain")
obs = env.reset()
while True:
action, _states = model.predict(obs)
obs, rewards, dones, info = env.step(action)
env.render()
Parameters¶
-
class
stable_baselines.ddpg.DDPG(policy, env, gamma=0.99, memory_policy=None, eval_env=None, nb_train_steps=50, nb_rollout_steps=100, nb_eval_steps=100, param_noise=None, action_noise=None, normalize_observations=False, tau=0.001, batch_size=128, param_noise_adaption_interval=50, normalize_returns=False, enable_popart=False, observation_range=(-5.0, 5.0), critic_l2_reg=0.0, return_range=(-inf, inf), actor_lr=0.0001, critic_lr=0.001, clip_norm=None, reward_scale=1.0, render=False, render_eval=False, memory_limit=None, buffer_size=50000, random_exploration=0.0, verbose=0, tensorboard_log=None, _init_setup_model=True, policy_kwargs=None, full_tensorboard_log=False, seed=None, n_cpu_tf_sess=1)[source]¶ Deep Deterministic Policy Gradient (DDPG) model
DDPG: https://arxiv.org/pdf/1509.02971.pdf
Parameters: - policy – (DDPGPolicy or str) The policy model to use (MlpPolicy, CnnPolicy, LnMlpPolicy, …)
- env – (Gym environment or str) The environment to learn from (if registered in Gym, can be str)
- gamma – (float) the discount factor
- memory_policy –
(ReplayBuffer) the replay buffer (if None, default to baselines.deepq.replay_buffer.ReplayBuffer)
Deprecated since version 2.6.0: This parameter will be removed in a future version
- eval_env – (Gym Environment) the evaluation environment (can be None)
- nb_train_steps – (int) the number of training steps
- nb_rollout_steps – (int) the number of rollout steps
- nb_eval_steps – (int) the number of evaluation steps
- param_noise – (AdaptiveParamNoiseSpec) the parameter noise type (can be None)
- action_noise – (ActionNoise) the action noise type (can be None)
- param_noise_adaption_interval – (int) apply param noise every N steps
- tau – (float) the soft update coefficient (keep old values, between 0 and 1)
- normalize_returns – (bool) should the critic output be normalized
- enable_popart – (bool) enable pop-art normalization of the critic output (https://arxiv.org/pdf/1602.07714.pdf), normalize_returns must be set to True.
- normalize_observations – (bool) should the observation be normalized
- batch_size – (int) the size of the batch for learning the policy
- observation_range – (tuple) the bounding values for the observation
- return_range – (tuple) the bounding values for the critic output
- critic_l2_reg – (float) l2 regularizer coefficient
- actor_lr – (float) the actor learning rate
- critic_lr – (float) the critic learning rate
- clip_norm – (float) clip the gradients (disabled if None)
- reward_scale – (float) the value the reward should be scaled by
- render – (bool) enable rendering of the environment
- render_eval – (bool) enable rendering of the evaluation environment
- memory_limit –
(int) the max number of transitions to store, size of the replay buffer
Deprecated since version 2.6.0: Use buffer_size instead.
- buffer_size – (int) the max number of transitions to store, size of the replay buffer
- random_exploration – (float) Probability of taking a random action (as in an epsilon-greedy strategy) This is not needed for DDPG normally but can help exploring when using HER + DDPG. This hack was present in the original OpenAI Baselines repo (DDPG + HER)
- verbose – (int) the verbosity level: 0 none, 1 training information, 2 tensorflow debug
- tensorboard_log – (str) the log location for tensorboard (if None, no logging)
- _init_setup_model – (bool) Whether or not to build the network at the creation of the instance
- policy_kwargs – (dict) additional arguments to be passed to the policy on creation
- full_tensorboard_log – (bool) enable additional logging when using tensorboard WARNING: this logging can take a lot of space quickly
- seed – (int) Seed for the pseudo-random generators (python, numpy, tensorflow). If None (default), use random seed. Note that if you want completely deterministic results, you must set n_cpu_tf_sess to 1.
- n_cpu_tf_sess – (int) The number of threads for TensorFlow operations If None, the number of cpu of the current machine will be used.
-
action_probability(observation, state=None, mask=None, actions=None, logp=False)[source]¶ If
actionsisNone, then get the model’s action probability distribution from a given observation.- Depending on the action space the output is:
- Discrete: probability for each possible action
- Box: mean and standard deviation of the action output
However if
actionsis notNone, this function will return the probability that the given actions are taken with the given parameters (observation, state, …) on this model. For discrete action spaces, it returns the probability mass; for continuous action spaces, the probability density. This is since the probability mass will always be zero in continuous spaces, see http://blog.christianperone.com/2019/01/ for a good explanationParameters: - observation – (np.ndarray) the input observation
- state – (np.ndarray) The last states (can be None, used in recurrent policies)
- mask – (np.ndarray) The last masks (can be None, used in recurrent policies)
- actions – (np.ndarray) (OPTIONAL) For calculating the likelihood that the given actions are chosen by the model for each of the given parameters. Must have the same number of actions and observations. (set to None to return the complete action probability distribution)
- logp – (bool) (OPTIONAL) When specified with actions, returns probability in log-space. This has no effect if actions is None.
Returns: (np.ndarray) the model’s (log) action probability
-
get_env()¶ returns the current environment (can be None if not defined)
Returns: (Gym Environment) The current environment
-
get_parameter_list()[source]¶ Get tensorflow Variables of model’s parameters
This includes all variables necessary for continuing training (saving / loading).
Returns: (list) List of tensorflow Variables
-
get_parameters()¶ Get current model parameters as dictionary of variable name -> ndarray.
Returns: (OrderedDict) Dictionary of variable name -> ndarray of model’s parameters.
-
get_vec_normalize_env() → Optional[stable_baselines.common.vec_env.vec_normalize.VecNormalize]¶ Return the
VecNormalizewrapper of the training env if it exists.Returns: Optional[VecNormalize] The VecNormalizeenv.
-
learn(total_timesteps, callback=None, log_interval=100, tb_log_name='DDPG', reset_num_timesteps=True, replay_wrapper=None)[source]¶ Return a trained model.
Parameters: - total_timesteps – (int) The total number of samples to train on
- callback – (Union[callable, [callable], BaseCallback]) function called at every steps with state of the algorithm. It takes the local and global variables. If it returns False, training is aborted. When the callback inherits from BaseCallback, you will have access to additional stages of the training (training start/end), please read the documentation for more details.
- log_interval – (int) The number of timesteps before logging.
- tb_log_name – (str) the name of the run for tensorboard log
- reset_num_timesteps – (bool) whether or not to reset the current timestep number (used in logging)
Returns: (BaseRLModel) the trained model
-
classmethod
load(load_path, env=None, custom_objects=None, **kwargs)[source]¶ Load the model from file
Parameters: - load_path – (str or file-like) the saved parameter location
- env – (Gym Environment) the new environment to run the loaded model on (can be None if you only need prediction from a trained model)
- custom_objects – (dict) Dictionary of objects to replace upon loading. If a variable is present in this dictionary as a key, it will not be deserialized and the corresponding item will be used instead. Similar to custom_objects in keras.models.load_model. Useful when you have an object in file that can not be deserialized.
- kwargs – extra arguments to change the model when loading
-
load_parameters(load_path_or_dict, exact_match=True)¶ Load model parameters from a file or a dictionary
Dictionary keys should be tensorflow variable names, which can be obtained with
get_parametersfunction. Ifexact_matchis True, dictionary should contain keys for all model’s parameters, otherwise RunTimeError is raised. If False, only variables included in the dictionary will be updated.This does not load agent’s hyper-parameters.
Warning
This function does not update trainer/optimizer variables (e.g. momentum). As such training after using this function may lead to less-than-optimal results.
Parameters: - load_path_or_dict – (str or file-like or dict) Save parameter location or dict of parameters as variable.name -> ndarrays to be loaded.
- exact_match – (bool) If True, expects load dictionary to contain keys for all variables in the model. If False, loads parameters only for variables mentioned in the dictionary. Defaults to True.
-
predict(observation, state=None, mask=None, deterministic=True)[source]¶ Get the model’s action from an observation
Parameters: - observation – (np.ndarray) the input observation
- state – (np.ndarray) The last states (can be None, used in recurrent policies)
- mask – (np.ndarray) The last masks (can be None, used in recurrent policies)
- deterministic – (bool) Whether or not to return deterministic actions.
Returns: (np.ndarray, np.ndarray) the model’s action and the next state (used in recurrent policies)
-
pretrain(dataset, n_epochs=10, learning_rate=0.0001, adam_epsilon=1e-08, val_interval=None)¶ Pretrain a model using behavior cloning: supervised learning given an expert dataset.
NOTE: only Box and Discrete spaces are supported for now.
Parameters: - dataset – (ExpertDataset) Dataset manager
- n_epochs – (int) Number of iterations on the training set
- learning_rate – (float) Learning rate
- adam_epsilon – (float) the epsilon value for the adam optimizer
- val_interval – (int) Report training and validation losses every n epochs. By default, every 10th of the maximum number of epochs.
Returns: (BaseRLModel) the pretrained model
-
save(save_path, cloudpickle=False)[source]¶ Save the current parameters to file
Parameters: - save_path – (str or file-like) The save location
- cloudpickle – (bool) Use older cloudpickle format instead of zip-archives.
-
set_env(env)¶ Checks the validity of the environment, and if it is coherent, set it as the current environment.
Parameters: env – (Gym Environment) The environment for learning a policy
-
set_random_seed(seed: Optional[int]) → None¶ Parameters: seed – (Optional[int]) Seed for the pseudo-random generators. If None, do not change the seeds.
DDPG Policies¶
-
class
stable_baselines.ddpg.MlpPolicy(sess, ob_space, ac_space, n_env, n_steps, n_batch, reuse=False, **_kwargs)[source]¶ Policy object that implements actor critic, using a MLP (2 layers of 64)
Parameters: - sess – (TensorFlow session) The current TensorFlow session
- ob_space – (Gym Space) The observation space of the environment
- ac_space – (Gym Space) The action space of the environment
- n_env – (int) The number of environments to run
- n_steps – (int) The number of steps to run for each environment
- n_batch – (int) The number of batch to run (n_envs * n_steps)
- reuse – (bool) If the policy is reusable or not
- _kwargs – (dict) Extra keyword arguments for the nature CNN feature extraction
-
action_ph¶ tf.Tensor: placeholder for actions, shape (self.n_batch, ) + self.ac_space.shape.
-
initial_state¶ The initial state of the policy. For feedforward policies, None. For a recurrent policy, a NumPy array of shape (self.n_env, ) + state_shape.
-
is_discrete¶ bool: is action space discrete.
-
make_actor(obs=None, reuse=False, scope='pi')¶ creates an actor object
Parameters: - obs – (TensorFlow Tensor) The observation placeholder (can be None for default placeholder)
- reuse – (bool) whether or not to reuse parameters
- scope – (str) the scope name of the actor
Returns: (TensorFlow Tensor) the output tensor
-
make_critic(obs=None, action=None, reuse=False, scope='qf')¶ creates a critic object
Parameters: - obs – (TensorFlow Tensor) The observation placeholder (can be None for default placeholder)
- action – (TensorFlow Tensor) The action placeholder (can be None for default placeholder)
- reuse – (bool) whether or not to reuse parameters
- scope – (str) the scope name of the critic
Returns: (TensorFlow Tensor) the output tensor
-
obs_ph¶ tf.Tensor: placeholder for observations, shape (self.n_batch, ) + self.ob_space.shape.
-
proba_step(obs, state=None, mask=None)¶ Returns the action probability for a single step
Parameters: - obs – ([float] or [int]) The current observation of the environment
- state – ([float]) The last states (used in recurrent policies)
- mask – ([float]) The last masks (used in recurrent policies)
Returns: ([float]) the action probability
-
processed_obs¶ tf.Tensor: processed observations, shape (self.n_batch, ) + self.ob_space.shape.
The form of processing depends on the type of the observation space, and the parameters whether scale is passed to the constructor; see observation_input for more information.
-
step(obs, state=None, mask=None)¶ Returns the policy for a single step
Parameters: - obs – ([float] or [int]) The current observation of the environment
- state – ([float]) The last states (used in recurrent policies)
- mask – ([float]) The last masks (used in recurrent policies)
Returns: ([float]) actions
-
value(obs, action, state=None, mask=None)¶ Returns the value for a single step
Parameters: - obs – ([float] or [int]) The current observation of the environment
- action – ([float] or [int]) The taken action
- state – ([float]) The last states (used in recurrent policies)
- mask – ([float]) The last masks (used in recurrent policies)
Returns: ([float]) The associated value of the action
-
class
stable_baselines.ddpg.LnMlpPolicy(sess, ob_space, ac_space, n_env, n_steps, n_batch, reuse=False, **_kwargs)[source]¶ Policy object that implements actor critic, using a MLP (2 layers of 64), with layer normalisation
Parameters: - sess – (TensorFlow session) The current TensorFlow session
- ob_space – (Gym Space) The observation space of the environment
- ac_space – (Gym Space) The action space of the environment
- n_env – (int) The number of environments to run
- n_steps – (int) The number of steps to run for each environment
- n_batch – (int) The number of batch to run (n_envs * n_steps)
- reuse – (bool) If the policy is reusable or not
- _kwargs – (dict) Extra keyword arguments for the nature CNN feature extraction
-
action_ph¶ tf.Tensor: placeholder for actions, shape (self.n_batch, ) + self.ac_space.shape.
-
initial_state¶ The initial state of the policy. For feedforward policies, None. For a recurrent policy, a NumPy array of shape (self.n_env, ) + state_shape.
-
is_discrete¶ bool: is action space discrete.
-
make_actor(obs=None, reuse=False, scope='pi')¶ creates an actor object
Parameters: - obs – (TensorFlow Tensor) The observation placeholder (can be None for default placeholder)
- reuse – (bool) whether or not to reuse parameters
- scope – (str) the scope name of the actor
Returns: (TensorFlow Tensor) the output tensor
-
make_critic(obs=None, action=None, reuse=False, scope='qf')¶ creates a critic object
Parameters: - obs – (TensorFlow Tensor) The observation placeholder (can be None for default placeholder)
- action – (TensorFlow Tensor) The action placeholder (can be None for default placeholder)
- reuse – (bool) whether or not to reuse parameters
- scope – (str) the scope name of the critic
Returns: (TensorFlow Tensor) the output tensor
-
obs_ph¶ tf.Tensor: placeholder for observations, shape (self.n_batch, ) + self.ob_space.shape.
-
proba_step(obs, state=None, mask=None)¶ Returns the action probability for a single step
Parameters: - obs – ([float] or [int]) The current observation of the environment
- state – ([float]) The last states (used in recurrent policies)
- mask – ([float]) The last masks (used in recurrent policies)
Returns: ([float]) the action probability
-
processed_obs¶ tf.Tensor: processed observations, shape (self.n_batch, ) + self.ob_space.shape.
The form of processing depends on the type of the observation space, and the parameters whether scale is passed to the constructor; see observation_input for more information.
-
step(obs, state=None, mask=None)¶ Returns the policy for a single step
Parameters: - obs – ([float] or [int]) The current observation of the environment
- state – ([float]) The last states (used in recurrent policies)
- mask – ([float]) The last masks (used in recurrent policies)
Returns: ([float]) actions
-
value(obs, action, state=None, mask=None)¶ Returns the value for a single step
Parameters: - obs – ([float] or [int]) The current observation of the environment
- action – ([float] or [int]) The taken action
- state – ([float]) The last states (used in recurrent policies)
- mask – ([float]) The last masks (used in recurrent policies)
Returns: ([float]) The associated value of the action
-
class
stable_baselines.ddpg.CnnPolicy(sess, ob_space, ac_space, n_env, n_steps, n_batch, reuse=False, **_kwargs)[source]¶ Policy object that implements actor critic, using a CNN (the nature CNN)
Parameters: - sess – (TensorFlow session) The current TensorFlow session
- ob_space – (Gym Space) The observation space of the environment
- ac_space – (Gym Space) The action space of the environment
- n_env – (int) The number of environments to run
- n_steps – (int) The number of steps to run for each environment
- n_batch – (int) The number of batch to run (n_envs * n_steps)
- reuse – (bool) If the policy is reusable or not
- _kwargs – (dict) Extra keyword arguments for the nature CNN feature extraction
-
action_ph¶ tf.Tensor: placeholder for actions, shape (self.n_batch, ) + self.ac_space.shape.
-
initial_state¶ The initial state of the policy. For feedforward policies, None. For a recurrent policy, a NumPy array of shape (self.n_env, ) + state_shape.
-
is_discrete¶ bool: is action space discrete.
-
make_actor(obs=None, reuse=False, scope='pi')¶ creates an actor object
Parameters: - obs – (TensorFlow Tensor) The observation placeholder (can be None for default placeholder)
- reuse – (bool) whether or not to reuse parameters
- scope – (str) the scope name of the actor
Returns: (TensorFlow Tensor) the output tensor
-
make_critic(obs=None, action=None, reuse=False, scope='qf')¶ creates a critic object
Parameters: - obs – (TensorFlow Tensor) The observation placeholder (can be None for default placeholder)
- action – (TensorFlow Tensor) The action placeholder (can be None for default placeholder)
- reuse – (bool) whether or not to reuse parameters
- scope – (str) the scope name of the critic
Returns: (TensorFlow Tensor) the output tensor
-
obs_ph¶ tf.Tensor: placeholder for observations, shape (self.n_batch, ) + self.ob_space.shape.
-
proba_step(obs, state=None, mask=None)¶ Returns the action probability for a single step
Parameters: - obs – ([float] or [int]) The current observation of the environment
- state – ([float]) The last states (used in recurrent policies)
- mask – ([float]) The last masks (used in recurrent policies)
Returns: ([float]) the action probability
-
processed_obs¶ tf.Tensor: processed observations, shape (self.n_batch, ) + self.ob_space.shape.
The form of processing depends on the type of the observation space, and the parameters whether scale is passed to the constructor; see observation_input for more information.
-
step(obs, state=None, mask=None)¶ Returns the policy for a single step
Parameters: - obs – ([float] or [int]) The current observation of the environment
- state – ([float]) The last states (used in recurrent policies)
- mask – ([float]) The last masks (used in recurrent policies)
Returns: ([float]) actions
-
value(obs, action, state=None, mask=None)¶ Returns the value for a single step
Parameters: - obs – ([float] or [int]) The current observation of the environment
- action – ([float] or [int]) The taken action
- state – ([float]) The last states (used in recurrent policies)
- mask – ([float]) The last masks (used in recurrent policies)
Returns: ([float]) The associated value of the action
-
class
stable_baselines.ddpg.LnCnnPolicy(sess, ob_space, ac_space, n_env, n_steps, n_batch, reuse=False, **_kwargs)[source]¶ Policy object that implements actor critic, using a CNN (the nature CNN), with layer normalisation
Parameters: - sess – (TensorFlow session) The current TensorFlow session
- ob_space – (Gym Space) The observation space of the environment
- ac_space – (Gym Space) The action space of the environment
- n_env – (int) The number of environments to run
- n_steps – (int) The number of steps to run for each environment
- n_batch – (int) The number of batch to run (n_envs * n_steps)
- reuse – (bool) If the policy is reusable or not
- _kwargs – (dict) Extra keyword arguments for the nature CNN feature extraction
-
action_ph¶ tf.Tensor: placeholder for actions, shape (self.n_batch, ) + self.ac_space.shape.
-
initial_state¶ The initial state of the policy. For feedforward policies, None. For a recurrent policy, a NumPy array of shape (self.n_env, ) + state_shape.
-
is_discrete¶ bool: is action space discrete.
-
make_actor(obs=None, reuse=False, scope='pi')¶ creates an actor object
Parameters: - obs – (TensorFlow Tensor) The observation placeholder (can be None for default placeholder)
- reuse – (bool) whether or not to reuse parameters
- scope – (str) the scope name of the actor
Returns: (TensorFlow Tensor) the output tensor
-
make_critic(obs=None, action=None, reuse=False, scope='qf')¶ creates a critic object
Parameters: - obs – (TensorFlow Tensor) The observation placeholder (can be None for default placeholder)
- action – (TensorFlow Tensor) The action placeholder (can be None for default placeholder)
- reuse – (bool) whether or not to reuse parameters
- scope – (str) the scope name of the critic
Returns: (TensorFlow Tensor) the output tensor
-
obs_ph¶ tf.Tensor: placeholder for observations, shape (self.n_batch, ) + self.ob_space.shape.
-
proba_step(obs, state=None, mask=None)¶ Returns the action probability for a single step
Parameters: - obs – ([float] or [int]) The current observation of the environment
- state – ([float]) The last states (used in recurrent policies)
- mask – ([float]) The last masks (used in recurrent policies)
Returns: ([float]) the action probability
-
processed_obs¶ tf.Tensor: processed observations, shape (self.n_batch, ) + self.ob_space.shape.
The form of processing depends on the type of the observation space, and the parameters whether scale is passed to the constructor; see observation_input for more information.
-
step(obs, state=None, mask=None)¶ Returns the policy for a single step
Parameters: - obs – ([float] or [int]) The current observation of the environment
- state – ([float]) The last states (used in recurrent policies)
- mask – ([float]) The last masks (used in recurrent policies)
Returns: ([float]) actions
-
value(obs, action, state=None, mask=None)¶ Returns the value for a single step
Parameters: - obs – ([float] or [int]) The current observation of the environment
- action – ([float] or [int]) The taken action
- state – ([float]) The last states (used in recurrent policies)
- mask – ([float]) The last masks (used in recurrent policies)
Returns: ([float]) The associated value of the action
Action and Parameters Noise¶
-
class
stable_baselines.ddpg.AdaptiveParamNoiseSpec(initial_stddev=0.1, desired_action_stddev=0.1, adoption_coefficient=1.01)[source]¶ Implements adaptive parameter noise
Parameters: - initial_stddev – (float) the initial value for the standard deviation of the noise
- desired_action_stddev – (float) the desired value for the standard deviation of the noise
- adoption_coefficient – (float) the update coefficient for the standard deviation of the noise
-
class
stable_baselines.ddpg.NormalActionNoise(mean, sigma)[source]¶ A Gaussian action noise
Parameters: - mean – (float) the mean value of the noise
- sigma – (float) the scale of the noise (std here)
-
reset()¶ call end of episode reset for the noise
-
class
stable_baselines.ddpg.OrnsteinUhlenbeckActionNoise(mean, sigma, theta=0.15, dt=0.01, initial_noise=None)[source]¶ A Ornstein Uhlenbeck action noise, this is designed to approximate brownian motion with friction.
Based on http://math.stackexchange.com/questions/1287634/implementing-ornstein-uhlenbeck-in-matlab
Parameters: - mean – (float) the mean of the noise
- sigma – (float) the scale of the noise
- theta – (float) the rate of mean reversion
- dt – (float) the timestep for the noise
- initial_noise – ([float]) the initial value for the noise output, (if None: 0)
Custom Policy Network¶
Similarly to the example given in the examples page. You can easily define a custom architecture for the policy network:
import gym
from stable_baselines.ddpg.policies import FeedForwardPolicy
from stable_baselines import DDPG
# Custom MLP policy of two layers of size 16 each
class CustomDDPGPolicy(FeedForwardPolicy):
def __init__(self, *args, **kwargs):
super(CustomDDPGPolicy, self).__init__(*args, **kwargs,
layers=[16, 16],
layer_norm=False,
feature_extraction="mlp")
model = DDPG(CustomDDPGPolicy, 'Pendulum-v0', verbose=1)
# Train the agent
model.learn(total_timesteps=100000)
DQN¶
Deep Q Network (DQN) and its extensions (Double-DQN, Dueling-DQN, Prioritized Experience Replay).
Warning
The DQN model does not support stable_baselines.common.policies,
as a result it must use its own policy models (see DQN Policies).
Available Policies
MlpPolicy |
Policy object that implements DQN policy, using a MLP (2 layers of 64) |
LnMlpPolicy |
Policy object that implements DQN policy, using a MLP (2 layers of 64), with layer normalisation |
CnnPolicy |
Policy object that implements DQN policy, using a CNN (the nature CNN) |
LnCnnPolicy |
Policy object that implements DQN policy, using a CNN (the nature CNN), with layer normalisation |
Notes¶
- DQN paper: https://arxiv.org/abs/1312.5602
- Dueling DQN: https://arxiv.org/abs/1511.06581
- Double-Q Learning: https://arxiv.org/abs/1509.06461
- Prioritized Experience Replay: https://arxiv.org/abs/1511.05952
Note
By default, the DQN class has double q learning and dueling extensions enabled. See Issue #406 for disabling dueling. To disable double-q learning, you can change the default value in the constructor.
Can I use?¶
- Recurrent policies: ❌
- Multi processing: ❌
- Gym spaces:
| Space | Action | Observation |
|---|---|---|
| Discrete | ✔️ | ✔️ |
| Box | ❌ | ✔️ |
| MultiDiscrete | ❌ | ✔️ |
| MultiBinary | ❌ | ✔️ |
Example¶
import gym
from stable_baselines.common.vec_env import DummyVecEnv
from stable_baselines.deepq.policies import MlpPolicy
from stable_baselines import DQN
env = gym.make('CartPole-v1')
model = DQN(MlpPolicy, env, verbose=1)
model.learn(total_timesteps=25000)
model.save("deepq_cartpole")
del model # remove to demonstrate saving and loading
model = DQN.load("deepq_cartpole")
obs = env.reset()
while True:
action, _states = model.predict(obs)
obs, rewards, dones, info = env.step(action)
env.render()
With Atari:
from stable_baselines.common.atari_wrappers import make_atari
from stable_baselines.deepq.policies import MlpPolicy, CnnPolicy
from stable_baselines import DQN
env = make_atari('BreakoutNoFrameskip-v4')
model = DQN(CnnPolicy, env, verbose=1)
model.learn(total_timesteps=25000)
model.save("deepq_breakout")
del model # remove to demonstrate saving and loading
model = DQN.load("deepq_breakout")
obs = env.reset()
while True:
action, _states = model.predict(obs)
obs, rewards, dones, info = env.step(action)
env.render()
Parameters¶
-
class
stable_baselines.deepq.DQN(policy, env, gamma=0.99, learning_rate=0.0005, buffer_size=50000, exploration_fraction=0.1, exploration_final_eps=0.02, exploration_initial_eps=1.0, train_freq=1, batch_size=32, double_q=True, learning_starts=1000, target_network_update_freq=500, prioritized_replay=False, prioritized_replay_alpha=0.6, prioritized_replay_beta0=0.4, prioritized_replay_beta_iters=None, prioritized_replay_eps=1e-06, param_noise=False, n_cpu_tf_sess=None, verbose=0, tensorboard_log=None, _init_setup_model=True, policy_kwargs=None, full_tensorboard_log=False, seed=None)[source]¶ The DQN model class. DQN paper: https://arxiv.org/abs/1312.5602 Dueling DQN: https://arxiv.org/abs/1511.06581 Double-Q Learning: https://arxiv.org/abs/1509.06461 Prioritized Experience Replay: https://arxiv.org/abs/1511.05952
Parameters: - policy – (DQNPolicy or str) The policy model to use (MlpPolicy, CnnPolicy, LnMlpPolicy, …)
- env – (Gym environment or str) The environment to learn from (if registered in Gym, can be str)
- gamma – (float) discount factor
- learning_rate – (float) learning rate for adam optimizer
- buffer_size – (int) size of the replay buffer
- exploration_fraction – (float) fraction of entire training period over which the exploration rate is annealed
- exploration_final_eps – (float) final value of random action probability
- exploration_initial_eps – (float) initial value of random action probability
- train_freq – (int) update the model every train_freq steps. set to None to disable printing
- batch_size – (int) size of a batched sampled from replay buffer for training
- double_q – (bool) Whether to enable Double-Q learning or not.
- learning_starts – (int) how many steps of the model to collect transitions for before learning starts
- target_network_update_freq – (int) update the target network every target_network_update_freq steps.
- prioritized_replay – (bool) if True prioritized replay buffer will be used.
- prioritized_replay_alpha – (float)alpha parameter for prioritized replay buffer. It determines how much prioritization is used, with alpha=0 corresponding to the uniform case.
- prioritized_replay_beta0 – (float) initial value of beta for prioritized replay buffer
- prioritized_replay_beta_iters – (int) number of iterations over which beta will be annealed from initial value to 1.0. If set to None equals to max_timesteps.
- prioritized_replay_eps – (float) epsilon to add to the TD errors when updating priorities.
- param_noise – (bool) Whether or not to apply noise to the parameters of the policy.
- verbose – (int) the verbosity level: 0 none, 1 training information, 2 tensorflow debug
- tensorboard_log – (str) the log location for tensorboard (if None, no logging)
- _init_setup_model – (bool) Whether or not to build the network at the creation of the instance
- full_tensorboard_log – (bool) enable additional logging when using tensorboard WARNING: this logging can take a lot of space quickly
- seed – (int) Seed for the pseudo-random generators (python, numpy, tensorflow). If None (default), use random seed. Note that if you want completely deterministic results, you must set n_cpu_tf_sess to 1.
- n_cpu_tf_sess – (int) The number of threads for TensorFlow operations If None, the number of cpu of the current machine will be used.
-
action_probability(observation, state=None, mask=None, actions=None, logp=False)[source]¶ If
actionsisNone, then get the model’s action probability distribution from a given observation.- Depending on the action space the output is:
- Discrete: probability for each possible action
- Box: mean and standard deviation of the action output
However if
actionsis notNone, this function will return the probability that the given actions are taken with the given parameters (observation, state, …) on this model. For discrete action spaces, it returns the probability mass; for continuous action spaces, the probability density. This is since the probability mass will always be zero in continuous spaces, see http://blog.christianperone.com/2019/01/ for a good explanationParameters: - observation – (np.ndarray) the input observation
- state – (np.ndarray) The last states (can be None, used in recurrent policies)
- mask – (np.ndarray) The last masks (can be None, used in recurrent policies)
- actions – (np.ndarray) (OPTIONAL) For calculating the likelihood that the given actions are chosen by the model for each of the given parameters. Must have the same number of actions and observations. (set to None to return the complete action probability distribution)
- logp – (bool) (OPTIONAL) When specified with actions, returns probability in log-space. This has no effect if actions is None.
Returns: (np.ndarray) the model’s (log) action probability
-
get_env()¶ returns the current environment (can be None if not defined)
Returns: (Gym Environment) The current environment
-
get_parameter_list()[source]¶ Get tensorflow Variables of model’s parameters
This includes all variables necessary for continuing training (saving / loading).
Returns: (list) List of tensorflow Variables
-
get_parameters()¶ Get current model parameters as dictionary of variable name -> ndarray.
Returns: (OrderedDict) Dictionary of variable name -> ndarray of model’s parameters.
-
get_vec_normalize_env() → Optional[stable_baselines.common.vec_env.vec_normalize.VecNormalize]¶ Return the
VecNormalizewrapper of the training env if it exists.Returns: Optional[VecNormalize] The VecNormalizeenv.
-
learn(total_timesteps, callback=None, log_interval=100, tb_log_name='DQN', reset_num_timesteps=True, replay_wrapper=None)[source]¶ Return a trained model.
Parameters: - total_timesteps – (int) The total number of samples to train on
- callback – (Union[callable, [callable], BaseCallback]) function called at every steps with state of the algorithm. It takes the local and global variables. If it returns False, training is aborted. When the callback inherits from BaseCallback, you will have access to additional stages of the training (training start/end), please read the documentation for more details.
- log_interval – (int) The number of timesteps before logging.
- tb_log_name – (str) the name of the run for tensorboard log
- reset_num_timesteps – (bool) whether or not to reset the current timestep number (used in logging)
Returns: (BaseRLModel) the trained model
-
classmethod
load(load_path, env=None, custom_objects=None, **kwargs)¶ Load the model from file
Parameters: - load_path – (str or file-like) the saved parameter location
- env – (Gym Environment) the new environment to run the loaded model on (can be None if you only need prediction from a trained model)
- custom_objects – (dict) Dictionary of objects to replace upon loading. If a variable is present in this dictionary as a key, it will not be deserialized and the corresponding item will be used instead. Similar to custom_objects in keras.models.load_model. Useful when you have an object in file that can not be deserialized.
- kwargs – extra arguments to change the model when loading
-
load_parameters(load_path_or_dict, exact_match=True)¶ Load model parameters from a file or a dictionary
Dictionary keys should be tensorflow variable names, which can be obtained with
get_parametersfunction. Ifexact_matchis True, dictionary should contain keys for all model’s parameters, otherwise RunTimeError is raised. If False, only variables included in the dictionary will be updated.This does not load agent’s hyper-parameters.
Warning
This function does not update trainer/optimizer variables (e.g. momentum). As such training after using this function may lead to less-than-optimal results.
Parameters: - load_path_or_dict – (str or file-like or dict) Save parameter location or dict of parameters as variable.name -> ndarrays to be loaded.
- exact_match – (bool) If True, expects load dictionary to contain keys for all variables in the model. If False, loads parameters only for variables mentioned in the dictionary. Defaults to True.
-
predict(observation, state=None, mask=None, deterministic=True)[source]¶ Get the model’s action from an observation
Parameters: - observation – (np.ndarray) the input observation
- state – (np.ndarray) The last states (can be None, used in recurrent policies)
- mask – (np.ndarray) The last masks (can be None, used in recurrent policies)
- deterministic – (bool) Whether or not to return deterministic actions.
Returns: (np.ndarray, np.ndarray) the model’s action and the next state (used in recurrent policies)
-
pretrain(dataset, n_epochs=10, learning_rate=0.0001, adam_epsilon=1e-08, val_interval=None)¶ Pretrain a model using behavior cloning: supervised learning given an expert dataset.
NOTE: only Box and Discrete spaces are supported for now.
Parameters: - dataset – (ExpertDataset) Dataset manager
- n_epochs – (int) Number of iterations on the training set
- learning_rate – (float) Learning rate
- adam_epsilon – (float) the epsilon value for the adam optimizer
- val_interval – (int) Report training and validation losses every n epochs. By default, every 10th of the maximum number of epochs.
Returns: (BaseRLModel) the pretrained model
-
save(save_path, cloudpickle=False)[source]¶ Save the current parameters to file
Parameters: - save_path – (str or file-like) The save location
- cloudpickle – (bool) Use older cloudpickle format instead of zip-archives.
-
set_env(env)¶ Checks the validity of the environment, and if it is coherent, set it as the current environment.
Parameters: env – (Gym Environment) The environment for learning a policy
-
set_random_seed(seed: Optional[int]) → None¶ Parameters: seed – (Optional[int]) Seed for the pseudo-random generators. If None, do not change the seeds.
DQN Policies¶
-
class
stable_baselines.deepq.MlpPolicy(sess, ob_space, ac_space, n_env, n_steps, n_batch, reuse=False, obs_phs=None, dueling=True, **_kwargs)[source]¶ Policy object that implements DQN policy, using a MLP (2 layers of 64)
Parameters: - sess – (TensorFlow session) The current TensorFlow session
- ob_space – (Gym Space) The observation space of the environment
- ac_space – (Gym Space) The action space of the environment
- n_env – (int) The number of environments to run
- n_steps – (int) The number of steps to run for each environment
- n_batch – (int) The number of batch to run (n_envs * n_steps)
- reuse – (bool) If the policy is reusable or not
- obs_phs – (TensorFlow Tensor, TensorFlow Tensor) a tuple containing an override for observation placeholder and the processed observation placeholder respectively
- dueling – (bool) if true double the output MLP to compute a baseline for action scores
- _kwargs – (dict) Extra keyword arguments for the nature CNN feature extraction
-
action_ph¶ tf.Tensor: placeholder for actions, shape (self.n_batch, ) + self.ac_space.shape.
-
initial_state¶ The initial state of the policy. For feedforward policies, None. For a recurrent policy, a NumPy array of shape (self.n_env, ) + state_shape.
-
is_discrete¶ bool: is action space discrete.
-
obs_ph¶ tf.Tensor: placeholder for observations, shape (self.n_batch, ) + self.ob_space.shape.
-
proba_step(obs, state=None, mask=None)¶ Returns the action probability for a single step
Parameters: - obs – (np.ndarray float or int) The current observation of the environment
- state – (np.ndarray float) The last states (used in recurrent policies)
- mask – (np.ndarray float) The last masks (used in recurrent policies)
Returns: (np.ndarray float) the action probability
-
processed_obs¶ tf.Tensor: processed observations, shape (self.n_batch, ) + self.ob_space.shape.
The form of processing depends on the type of the observation space, and the parameters whether scale is passed to the constructor; see observation_input for more information.
-
step(obs, state=None, mask=None, deterministic=True)¶ Returns the q_values for a single step
Parameters: - obs – (np.ndarray float or int) The current observation of the environment
- state – (np.ndarray float) The last states (used in recurrent policies)
- mask – (np.ndarray float) The last masks (used in recurrent policies)
- deterministic – (bool) Whether or not to return deterministic actions.
Returns: (np.ndarray int, np.ndarray float, np.ndarray float) actions, q_values, states
-
class
stable_baselines.deepq.LnMlpPolicy(sess, ob_space, ac_space, n_env, n_steps, n_batch, reuse=False, obs_phs=None, dueling=True, **_kwargs)[source]¶ Policy object that implements DQN policy, using a MLP (2 layers of 64), with layer normalisation
Parameters: - sess – (TensorFlow session) The current TensorFlow session
- ob_space – (Gym Space) The observation space of the environment
- ac_space – (Gym Space) The action space of the environment
- n_env – (int) The number of environments to run
- n_steps – (int) The number of steps to run for each environment
- n_batch – (int) The number of batch to run (n_envs * n_steps)
- reuse – (bool) If the policy is reusable or not
- obs_phs – (TensorFlow Tensor, TensorFlow Tensor) a tuple containing an override for observation placeholder and the processed observation placeholder respectively
- dueling – (bool) if true double the output MLP to compute a baseline for action scores
- _kwargs – (dict) Extra keyword arguments for the nature CNN feature extraction
-
action_ph¶ tf.Tensor: placeholder for actions, shape (self.n_batch, ) + self.ac_space.shape.
-
initial_state¶ The initial state of the policy. For feedforward policies, None. For a recurrent policy, a NumPy array of shape (self.n_env, ) + state_shape.
-
is_discrete¶ bool: is action space discrete.
-
obs_ph¶ tf.Tensor: placeholder for observations, shape (self.n_batch, ) + self.ob_space.shape.
-
proba_step(obs, state=None, mask=None)¶ Returns the action probability for a single step
Parameters: - obs – (np.ndarray float or int) The current observation of the environment
- state – (np.ndarray float) The last states (used in recurrent policies)
- mask – (np.ndarray float) The last masks (used in recurrent policies)
Returns: (np.ndarray float) the action probability
-
processed_obs¶ tf.Tensor: processed observations, shape (self.n_batch, ) + self.ob_space.shape.
The form of processing depends on the type of the observation space, and the parameters whether scale is passed to the constructor; see observation_input for more information.
-
step(obs, state=None, mask=None, deterministic=True)¶ Returns the q_values for a single step
Parameters: - obs – (np.ndarray float or int) The current observation of the environment
- state – (np.ndarray float) The last states (used in recurrent policies)
- mask – (np.ndarray float) The last masks (used in recurrent policies)
- deterministic – (bool) Whether or not to return deterministic actions.
Returns: (np.ndarray int, np.ndarray float, np.ndarray float) actions, q_values, states
-
class
stable_baselines.deepq.CnnPolicy(sess, ob_space, ac_space, n_env, n_steps, n_batch, reuse=False, obs_phs=None, dueling=True, **_kwargs)[source]¶ Policy object that implements DQN policy, using a CNN (the nature CNN)
Parameters: - sess – (TensorFlow session) The current TensorFlow session
- ob_space – (Gym Space) The observation space of the environment
- ac_space – (Gym Space) The action space of the environment
- n_env – (int) The number of environments to run
- n_steps – (int) The number of steps to run for each environment
- n_batch – (int) The number of batch to run (n_envs * n_steps)
- reuse – (bool) If the policy is reusable or not
- obs_phs – (TensorFlow Tensor, TensorFlow Tensor) a tuple containing an override for observation placeholder and the processed observation placeholder respectively
- dueling – (bool) if true double the output MLP to compute a baseline for action scores
- _kwargs – (dict) Extra keyword arguments for the nature CNN feature extraction
-
action_ph¶ tf.Tensor: placeholder for actions, shape (self.n_batch, ) + self.ac_space.shape.
-
initial_state¶ The initial state of the policy. For feedforward policies, None. For a recurrent policy, a NumPy array of shape (self.n_env, ) + state_shape.
-
is_discrete¶ bool: is action space discrete.
-
obs_ph¶ tf.Tensor: placeholder for observations, shape (self.n_batch, ) + self.ob_space.shape.
-
proba_step(obs, state=None, mask=None)¶ Returns the action probability for a single step
Parameters: - obs – (np.ndarray float or int) The current observation of the environment
- state – (np.ndarray float) The last states (used in recurrent policies)
- mask – (np.ndarray float) The last masks (used in recurrent policies)
Returns: (np.ndarray float) the action probability
-
processed_obs¶ tf.Tensor: processed observations, shape (self.n_batch, ) + self.ob_space.shape.
The form of processing depends on the type of the observation space, and the parameters whether scale is passed to the constructor; see observation_input for more information.
-
step(obs, state=None, mask=None, deterministic=True)¶ Returns the q_values for a single step
Parameters: - obs – (np.ndarray float or int) The current observation of the environment
- state – (np.ndarray float) The last states (used in recurrent policies)
- mask – (np.ndarray float) The last masks (used in recurrent policies)
- deterministic – (bool) Whether or not to return deterministic actions.
Returns: (np.ndarray int, np.ndarray float, np.ndarray float) actions, q_values, states
-
class
stable_baselines.deepq.LnCnnPolicy(sess, ob_space, ac_space, n_env, n_steps, n_batch, reuse=False, obs_phs=None, dueling=True, **_kwargs)[source]¶ Policy object that implements DQN policy, using a CNN (the nature CNN), with layer normalisation
Parameters: - sess – (TensorFlow session) The current TensorFlow session
- ob_space – (Gym Space) The observation space of the environment
- ac_space – (Gym Space) The action space of the environment
- n_env – (int) The number of environments to run
- n_steps – (int) The number of steps to run for each environment
- n_batch – (int) The number of batch to run (n_envs * n_steps)
- reuse – (bool) If the policy is reusable or not
- obs_phs – (TensorFlow Tensor, TensorFlow Tensor) a tuple containing an override for observation placeholder and the processed observation placeholder respectively
- dueling – (bool) if true double the output MLP to compute a baseline for action scores
- _kwargs – (dict) Extra keyword arguments for the nature CNN feature extraction
-
action_ph¶ tf.Tensor: placeholder for actions, shape (self.n_batch, ) + self.ac_space.shape.
-
initial_state¶ The initial state of the policy. For feedforward policies, None. For a recurrent policy, a NumPy array of shape (self.n_env, ) + state_shape.
-
is_discrete¶ bool: is action space discrete.
-
obs_ph¶ tf.Tensor: placeholder for observations, shape (self.n_batch, ) + self.ob_space.shape.
-
proba_step(obs, state=None, mask=None)¶ Returns the action probability for a single step
Parameters: - obs – (np.ndarray float or int) The current observation of the environment
- state – (np.ndarray float) The last states (used in recurrent policies)
- mask – (np.ndarray float) The last masks (used in recurrent policies)
Returns: (np.ndarray float) the action probability
-
processed_obs¶ tf.Tensor: processed observations, shape (self.n_batch, ) + self.ob_space.shape.
The form of processing depends on the type of the observation space, and the parameters whether scale is passed to the constructor; see observation_input for more information.
-
step(obs, state=None, mask=None, deterministic=True)¶ Returns the q_values for a single step
Parameters: - obs – (np.ndarray float or int) The current observation of the environment
- state – (np.ndarray float) The last states (used in recurrent policies)
- mask – (np.ndarray float) The last masks (used in recurrent policies)
- deterministic – (bool) Whether or not to return deterministic actions.
Returns: (np.ndarray int, np.ndarray float, np.ndarray float) actions, q_values, states
Custom Policy Network¶
Similarly to the example given in the examples page. You can easily define a custom architecture for the policy network:
import gym
from stable_baselines.deepq.policies import FeedForwardPolicy
from stable_baselines.common.vec_env import DummyVecEnv
from stable_baselines import DQN
# Custom MLP policy of two layers of size 32 each
class CustomDQNPolicy(FeedForwardPolicy):
def __init__(self, *args, **kwargs):
super(CustomDQNPolicy, self).__init__(*args, **kwargs,
layers=[32, 32],
layer_norm=False,
feature_extraction="mlp")
# Create and wrap the environment
env = gym.make('LunarLander-v2')
env = DummyVecEnv([lambda: env])
model = DQN(CustomDQNPolicy, env, verbose=1)
# Train the agent
model.learn(total_timesteps=100000)
GAIL¶
The Generative Adversarial Imitation Learning (GAIL) uses expert trajectories to recover a cost function and then learn a policy.
Learning a cost function from expert demonstrations is called Inverse Reinforcement Learning (IRL). The connection between GAIL and Generative Adversarial Networks (GANs) is that it uses a discriminator that tries to seperate expert trajectory from trajectories of the learned policy, which has the role of the generator here.
Note
GAIL requires OpenMPI. If OpenMPI isn’t enabled, then GAIL isn’t
imported into the stable_baselines module.
Notes¶
- Original paper: https://arxiv.org/abs/1606.03476
Warning
Images are not yet handled properly by the current implementation
If you want to train an imitation learning agent¶
Step 1: Generate expert data¶
You can either train a RL algorithm in a classic setting, use another controller (e.g. a PID controller) or human demonstrations.
We recommend you to take a look at pre-training section
or directly look at stable_baselines/gail/dataset/ folder to learn more about the expected format for the dataset.
Here is an example of training a Soft Actor-Critic model to generate expert trajectories for GAIL:
from stable_baselines import SAC
from stable_baselines.gail import generate_expert_traj
# Generate expert trajectories (train expert)
model = SAC('MlpPolicy', 'Pendulum-v0', verbose=1)
# Train for 60000 timesteps and record 10 trajectories
# all the data will be saved in 'expert_pendulum.npz' file
generate_expert_traj(model, 'expert_pendulum', n_timesteps=60000, n_episodes=10)
Step 2: Run GAIL¶
In case you want to run Behavior Cloning (BC)
Use the .pretrain() method (cf guide).
Others
Thanks to the open source:
- @openai/imitation
- @carpedm20/deep-rl-tensorflow
Can I use?¶
- Recurrent policies: ❌
- Multi processing: ✔️ (using MPI)
- Gym spaces:
| Space | Action | Observation |
|---|---|---|
| Discrete | ✔️ | ✔️ |
| Box | ✔️ | ✔️ |
| MultiDiscrete | ❌ | ✔️ |
| MultiBinary | ❌ | ✔️ |
Example¶
import gym
from stable_baselines import GAIL, SAC
from stable_baselines.gail import ExpertDataset, generate_expert_traj
# Generate expert trajectories (train expert)
model = SAC('MlpPolicy', 'Pendulum-v0', verbose=1)
generate_expert_traj(model, 'expert_pendulum', n_timesteps=100, n_episodes=10)
# Load the expert dataset
dataset = ExpertDataset(expert_path='expert_pendulum.npz', traj_limitation=10, verbose=1)
model = GAIL('MlpPolicy', 'Pendulum-v0', dataset, verbose=1)
# Note: in practice, you need to train for 1M steps to have a working policy
model.learn(total_timesteps=1000)
model.save("gail_pendulum")
del model # remove to demonstrate saving and loading
model = GAIL.load("gail_pendulum")
env = gym.make('Pendulum-v0')
obs = env.reset()
while True:
action, _states = model.predict(obs)
obs, rewards, dones, info = env.step(action)
env.render()
Parameters¶
-
class
stable_baselines.gail.GAIL(policy, env, expert_dataset=None, hidden_size_adversary=100, adversary_entcoeff=0.001, g_step=3, d_step=1, d_stepsize=0.0003, verbose=0, _init_setup_model=True, **kwargs)[source]¶ Generative Adversarial Imitation Learning (GAIL)
Warning
Images are not yet handled properly by the current implementation
Parameters: - policy – (ActorCriticPolicy or str) The policy model to use (MlpPolicy, CnnPolicy, CnnLstmPolicy, …)
- env – (Gym environment or str) The environment to learn from (if registered in Gym, can be str)
- expert_dataset – (ExpertDataset) the dataset manager
- gamma – (float) the discount value
- timesteps_per_batch – (int) the number of timesteps to run per batch (horizon)
- max_kl – (float) the Kullback-Leibler loss threshold
- cg_iters – (int) the number of iterations for the conjugate gradient calculation
- lam – (float) GAE factor
- entcoeff – (float) the weight for the entropy loss
- cg_damping – (float) the compute gradient dampening factor
- vf_stepsize – (float) the value function stepsize
- vf_iters – (int) the value function’s number iterations for learning
- hidden_size – ([int]) the hidden dimension for the MLP
- g_step – (int) number of steps to train policy in each epoch
- d_step – (int) number of steps to train discriminator in each epoch
- d_stepsize – (float) the reward giver stepsize
- verbose – (int) the verbosity level: 0 none, 1 training information, 2 tensorflow debug
- _init_setup_model – (bool) Whether or not to build the network at the creation of the instance
- full_tensorboard_log – (bool) enable additional logging when using tensorboard WARNING: this logging can take a lot of space quickly
-
action_probability(observation, state=None, mask=None, actions=None, logp=False)¶ If
actionsisNone, then get the model’s action probability distribution from a given observation.- Depending on the action space the output is:
- Discrete: probability for each possible action
- Box: mean and standard deviation of the action output
However if
actionsis notNone, this function will return the probability that the given actions are taken with the given parameters (observation, state, …) on this model. For discrete action spaces, it returns the probability mass; for continuous action spaces, the probability density. This is since the probability mass will always be zero in continuous spaces, see http://blog.christianperone.com/2019/01/ for a good explanationParameters: - observation – (np.ndarray) the input observation
- state – (np.ndarray) The last states (can be None, used in recurrent policies)
- mask – (np.ndarray) The last masks (can be None, used in recurrent policies)
- actions – (np.ndarray) (OPTIONAL) For calculating the likelihood that the given actions are chosen by the model for each of the given parameters. Must have the same number of actions and observations. (set to None to return the complete action probability distribution)
- logp – (bool) (OPTIONAL) When specified with actions, returns probability in log-space. This has no effect if actions is None.
Returns: (np.ndarray) the model’s (log) action probability
-
get_env()¶ returns the current environment (can be None if not defined)
Returns: (Gym Environment) The current environment
-
get_parameter_list()¶ Get tensorflow Variables of model’s parameters
This includes all variables necessary for continuing training (saving / loading).
Returns: (list) List of tensorflow Variables
-
get_parameters()¶ Get current model parameters as dictionary of variable name -> ndarray.
Returns: (OrderedDict) Dictionary of variable name -> ndarray of model’s parameters.
-
get_vec_normalize_env() → Optional[stable_baselines.common.vec_env.vec_normalize.VecNormalize]¶ Return the
VecNormalizewrapper of the training env if it exists.Returns: Optional[VecNormalize] The VecNormalizeenv.
-
learn(total_timesteps, callback=None, log_interval=100, tb_log_name='GAIL', reset_num_timesteps=True)[source]¶ Return a trained model.
Parameters: - total_timesteps – (int) The total number of samples to train on
- callback – (Union[callable, [callable], BaseCallback]) function called at every steps with state of the algorithm. It takes the local and global variables. If it returns False, training is aborted. When the callback inherits from BaseCallback, you will have access to additional stages of the training (training start/end), please read the documentation for more details.
- log_interval – (int) The number of timesteps before logging.
- tb_log_name – (str) the name of the run for tensorboard log
- reset_num_timesteps – (bool) whether or not to reset the current timestep number (used in logging)
Returns: (BaseRLModel) the trained model
-
classmethod
load(load_path, env=None, custom_objects=None, **kwargs)¶ Load the model from file
Parameters: - load_path – (str or file-like) the saved parameter location
- env – (Gym Environment) the new environment to run the loaded model on (can be None if you only need prediction from a trained model)
- custom_objects – (dict) Dictionary of objects to replace upon loading. If a variable is present in this dictionary as a key, it will not be deserialized and the corresponding item will be used instead. Similar to custom_objects in keras.models.load_model. Useful when you have an object in file that can not be deserialized.
- kwargs – extra arguments to change the model when loading
-
load_parameters(load_path_or_dict, exact_match=True)¶ Load model parameters from a file or a dictionary
Dictionary keys should be tensorflow variable names, which can be obtained with
get_parametersfunction. Ifexact_matchis True, dictionary should contain keys for all model’s parameters, otherwise RunTimeError is raised. If False, only variables included in the dictionary will be updated.This does not load agent’s hyper-parameters.
Warning
This function does not update trainer/optimizer variables (e.g. momentum). As such training after using this function may lead to less-than-optimal results.
Parameters: - load_path_or_dict – (str or file-like or dict) Save parameter location or dict of parameters as variable.name -> ndarrays to be loaded.
- exact_match – (bool) If True, expects load dictionary to contain keys for all variables in the model. If False, loads parameters only for variables mentioned in the dictionary. Defaults to True.
-
predict(observation, state=None, mask=None, deterministic=False)¶ Get the model’s action from an observation
Parameters: - observation – (np.ndarray) the input observation
- state – (np.ndarray) The last states (can be None, used in recurrent policies)
- mask – (np.ndarray) The last masks (can be None, used in recurrent policies)
- deterministic – (bool) Whether or not to return deterministic actions.
Returns: (np.ndarray, np.ndarray) the model’s action and the next state (used in recurrent policies)
-
pretrain(dataset, n_epochs=10, learning_rate=0.0001, adam_epsilon=1e-08, val_interval=None)¶ Pretrain a model using behavior cloning: supervised learning given an expert dataset.
NOTE: only Box and Discrete spaces are supported for now.
Parameters: - dataset – (ExpertDataset) Dataset manager
- n_epochs – (int) Number of iterations on the training set
- learning_rate – (float) Learning rate
- adam_epsilon – (float) the epsilon value for the adam optimizer
- val_interval – (int) Report training and validation losses every n epochs. By default, every 10th of the maximum number of epochs.
Returns: (BaseRLModel) the pretrained model
-
save(save_path, cloudpickle=False)¶ Save the current parameters to file
Parameters: - save_path – (str or file-like) The save location
- cloudpickle – (bool) Use older cloudpickle format instead of zip-archives.
-
set_env(env)¶ Checks the validity of the environment, and if it is coherent, set it as the current environment.
Parameters: env – (Gym Environment) The environment for learning a policy
-
set_random_seed(seed: Optional[int]) → None¶ Parameters: seed – (Optional[int]) Seed for the pseudo-random generators. If None, do not change the seeds.
-
setup_model()¶ Create all the functions and tensorflow graphs necessary to train the model
HER¶
Hindsight Experience Replay (HER)
HER is a method wrapper that works with Off policy methods (DQN, SAC, TD3 and DDPG for example).
Note
HER was re-implemented from scratch in Stable-Baselines compared to the original OpenAI baselines. If you want to reproduce results from the paper, please use the rl baselines zoo in order to have the correct hyperparameters and at least 8 MPI workers with DDPG.
Warning
HER requires the environment to inherits from gym.GoalEnv
Warning
you must pass an environment or wrap it with HERGoalEnvWrapper in order to use the predict method
Notes¶
- Original paper: https://arxiv.org/abs/1707.01495
- OpenAI paper: Plappert et al. (2018)
- OpenAI blog post: https://openai.com/blog/ingredients-for-robotics-research/
Can I use?¶
Please refer to the wrapped model (DQN, SAC, TD3 or DDPG) for that section.
Example¶
from stable_baselines import HER, DQN, SAC, DDPG, TD3
from stable_baselines.her import GoalSelectionStrategy, HERGoalEnvWrapper
from stable_baselines.common.bit_flipping_env import BitFlippingEnv
model_class = DQN # works also with SAC, DDPG and TD3
env = BitFlippingEnv(N_BITS, continuous=model_class in [DDPG, SAC, TD3], max_steps=N_BITS)
# Available strategies (cf paper): future, final, episode, random
goal_selection_strategy = 'future' # equivalent to GoalSelectionStrategy.FUTURE
# Wrap the model
model = HER('MlpPolicy', env, model_class, n_sampled_goal=4, goal_selection_strategy=goal_selection_strategy,
verbose=1)
# Train the model
model.learn(1000)
model.save("./her_bit_env")
# WARNING: you must pass an env
# or wrap your environment with HERGoalEnvWrapper to use the predict method
model = HER.load('./her_bit_env', env=env)
obs = env.reset()
for _ in range(100):
action, _ = model.predict(obs)
obs, reward, done, _ = env.step(action)
if done:
obs = env.reset()
Parameters¶
-
class
stable_baselines.her.HER(policy, env, model_class, n_sampled_goal=4, goal_selection_strategy='future', *args, **kwargs)[source]¶ Hindsight Experience Replay (HER) https://arxiv.org/abs/1707.01495
Parameters: - policy – (BasePolicy or str) The policy model to use (MlpPolicy, CnnPolicy, CnnLstmPolicy, …)
- env – (Gym environment or str) The environment to learn from (if registered in Gym, can be str)
- model_class – (OffPolicyRLModel) The off policy RL model to apply Hindsight Experience Replay currently supported: DQN, DDPG, SAC
- n_sampled_goal – (int)
- goal_selection_strategy – (GoalSelectionStrategy or str)
-
action_probability(observation, state=None, mask=None, actions=None, logp=False)[source]¶ If
actionsisNone, then get the model’s action probability distribution from a given observation.- Depending on the action space the output is:
- Discrete: probability for each possible action
- Box: mean and standard deviation of the action output
However if
actionsis notNone, this function will return the probability that the given actions are taken with the given parameters (observation, state, …) on this model. For discrete action spaces, it returns the probability mass; for continuous action spaces, the probability density. This is since the probability mass will always be zero in continuous spaces, see http://blog.christianperone.com/2019/01/ for a good explanationParameters: - observation – (np.ndarray) the input observation
- state – (np.ndarray) The last states (can be None, used in recurrent policies)
- mask – (np.ndarray) The last masks (can be None, used in recurrent policies)
- actions – (np.ndarray) (OPTIONAL) For calculating the likelihood that the given actions are chosen by the model for each of the given parameters. Must have the same number of actions and observations. (set to None to return the complete action probability distribution)
- logp – (bool) (OPTIONAL) When specified with actions, returns probability in log-space. This has no effect if actions is None.
Returns: (np.ndarray) the model’s (log) action probability
-
get_env()[source]¶ returns the current environment (can be None if not defined)
Returns: (Gym Environment) The current environment
-
get_parameter_list()[source]¶ Get tensorflow Variables of model’s parameters
This includes all variables necessary for continuing training (saving / loading).
Returns: (list) List of tensorflow Variables
-
learn(total_timesteps, callback=None, log_interval=100, tb_log_name='HER', reset_num_timesteps=True)[source]¶ Return a trained model.
Parameters: - total_timesteps – (int) The total number of samples to train on
- callback – (Union[callable, [callable], BaseCallback]) function called at every steps with state of the algorithm. It takes the local and global variables. If it returns False, training is aborted. When the callback inherits from BaseCallback, you will have access to additional stages of the training (training start/end), please read the documentation for more details.
- log_interval – (int) The number of timesteps before logging.
- tb_log_name – (str) the name of the run for tensorboard log
- reset_num_timesteps – (bool) whether or not to reset the current timestep number (used in logging)
Returns: (BaseRLModel) the trained model
-
classmethod
load(load_path, env=None, custom_objects=None, **kwargs)[source]¶ Load the model from file
Parameters: - load_path – (str or file-like) the saved parameter location
- env – (Gym Environment) the new environment to run the loaded model on (can be None if you only need prediction from a trained model)
- custom_objects – (dict) Dictionary of objects to replace upon loading. If a variable is present in this dictionary as a key, it will not be deserialized and the corresponding item will be used instead. Similar to custom_objects in keras.models.load_model. Useful when you have an object in file that can not be deserialized.
- kwargs – extra arguments to change the model when loading
-
predict(observation, state=None, mask=None, deterministic=True)[source]¶ Get the model’s action from an observation
Parameters: - observation – (np.ndarray) the input observation
- state – (np.ndarray) The last states (can be None, used in recurrent policies)
- mask – (np.ndarray) The last masks (can be None, used in recurrent policies)
- deterministic – (bool) Whether or not to return deterministic actions.
Returns: (np.ndarray, np.ndarray) the model’s action and the next state (used in recurrent policies)
-
save(save_path, cloudpickle=False)[source]¶ Save the current parameters to file
Parameters: - save_path – (str or file-like) The save location
- cloudpickle – (bool) Use older cloudpickle format instead of zip-archives.
Goal Selection Strategies¶
Goal Env Wrapper¶
-
class
stable_baselines.her.HERGoalEnvWrapper(env)[source]¶ A wrapper that allow to use dict observation space (coming from GoalEnv) with the RL algorithms. It assumes that all the spaces of the dict space are of the same type.
Parameters: env – (gym.GoalEnv)
Replay Wrapper¶
-
class
stable_baselines.her.HindsightExperienceReplayWrapper(replay_buffer, n_sampled_goal, goal_selection_strategy, wrapped_env)[source]¶ Wrapper around a replay buffer in order to use HER. This implementation is inspired by to the one found in https://github.com/NervanaSystems/coach/.
Parameters: - replay_buffer – (ReplayBuffer)
- n_sampled_goal – (int) The number of artificial transitions to generate for each actual transition
- goal_selection_strategy – (GoalSelectionStrategy) The method that will be used to generate the goals for the artificial transitions.
- wrapped_env – (HERGoalEnvWrapper) the GoalEnv wrapped using HERGoalEnvWrapper, that enables to convert observation to dict, and vice versa
PPO1¶
The Proximal Policy Optimization algorithm combines ideas from A2C (having multiple workers) and TRPO (it uses a trust region to improve the actor).
The main idea is that after an update, the new policy should be not too far from the old policy.
For that, ppo uses clipping to avoid too large update.
Note
PPO1 requires OpenMPI. If OpenMPI isn’t enabled, then PPO1 isn’t
imported into the stable_baselines module.
Note
PPO1 uses MPI for multiprocessing unlike PPO2, which uses vectorized environments. PPO2 is the implementation OpenAI made for GPU.
Notes¶
- Original paper: https://arxiv.org/abs/1707.06347
- Clear explanation of PPO on Arxiv Insights channel: https://www.youtube.com/watch?v=5P7I-xPq8u8
- OpenAI blog post: https://blog.openai.com/openai-baselines-ppo/
mpirun -np 8 python -m stable_baselines.ppo1.run_atariruns the algorithm for 40M frames = 10M timesteps on an Atari game. See help (-h) for more options.python -m stable_baselines.ppo1.run_mujocoruns the algorithm for 1M frames on a Mujoco environment.- Train mujoco 3d humanoid (with optimal-ish hyperparameters):
mpirun -np 16 python -m stable_baselines.ppo1.run_humanoid --model-path=/path/to/model - Render the 3d humanoid:
python -m stable_baselines.ppo1.run_humanoid --play --model-path=/path/to/model
Can I use?¶
- Recurrent policies: ❌
- Multi processing: ✔️ (using MPI)
- Gym spaces:
| Space | Action | Observation |
|---|---|---|
| Discrete | ✔️ | ✔️ |
| Box | ✔️ | ✔️ |
| MultiDiscrete | ✔️ | ✔️ |
| MultiBinary | ✔️ | ✔️ |
Example¶
import gym
from stable_baselines.common.policies import MlpPolicy
from stable_baselines import PPO1
env = gym.make('CartPole-v1')
model = PPO1(MlpPolicy, env, verbose=1)
model.learn(total_timesteps=25000)
model.save("ppo1_cartpole")
del model # remove to demonstrate saving and loading
model = PPO1.load("ppo1_cartpole")
obs = env.reset()
while True:
action, _states = model.predict(obs)
obs, rewards, dones, info = env.step(action)
env.render()
Parameters¶
-
class
stable_baselines.ppo1.PPO1(policy, env, gamma=0.99, timesteps_per_actorbatch=256, clip_param=0.2, entcoeff=0.01, optim_epochs=4, optim_stepsize=0.001, optim_batchsize=64, lam=0.95, adam_epsilon=1e-05, schedule='linear', verbose=0, tensorboard_log=None, _init_setup_model=True, policy_kwargs=None, full_tensorboard_log=False, seed=None, n_cpu_tf_sess=1)[source]¶ Proximal Policy Optimization algorithm (MPI version). Paper: https://arxiv.org/abs/1707.06347
Parameters: - env – (Gym environment or str) The environment to learn from (if registered in Gym, can be str)
- policy – (ActorCriticPolicy or str) The policy model to use (MlpPolicy, CnnPolicy, CnnLstmPolicy, …)
- timesteps_per_actorbatch – (int) timesteps per actor per update
- clip_param – (float) clipping parameter epsilon
- entcoeff – (float) the entropy loss weight
- optim_epochs – (float) the optimizer’s number of epochs
- optim_stepsize – (float) the optimizer’s stepsize
- optim_batchsize – (int) the optimizer’s the batch size
- gamma – (float) discount factor
- lam – (float) advantage estimation
- adam_epsilon – (float) the epsilon value for the adam optimizer
- schedule – (str) The type of scheduler for the learning rate update (‘linear’, ‘constant’, ‘double_linear_con’, ‘middle_drop’ or ‘double_middle_drop’)
- verbose – (int) the verbosity level: 0 none, 1 training information, 2 tensorflow debug
- tensorboard_log – (str) the log location for tensorboard (if None, no logging)
- _init_setup_model – (bool) Whether or not to build the network at the creation of the instance
- policy_kwargs – (dict) additional arguments to be passed to the policy on creation
- full_tensorboard_log – (bool) enable additional logging when using tensorboard WARNING: this logging can take a lot of space quickly
- seed – (int) Seed for the pseudo-random generators (python, numpy, tensorflow). If None (default), use random seed. Note that if you want completely deterministic results, you must set n_cpu_tf_sess to 1.
- n_cpu_tf_sess – (int) The number of threads for TensorFlow operations If None, the number of cpu of the current machine will be used.
-
action_probability(observation, state=None, mask=None, actions=None, logp=False)¶ If
actionsisNone, then get the model’s action probability distribution from a given observation.- Depending on the action space the output is:
- Discrete: probability for each possible action
- Box: mean and standard deviation of the action output
However if
actionsis notNone, this function will return the probability that the given actions are taken with the given parameters (observation, state, …) on this model. For discrete action spaces, it returns the probability mass; for continuous action spaces, the probability density. This is since the probability mass will always be zero in continuous spaces, see http://blog.christianperone.com/2019/01/ for a good explanationParameters: - observation – (np.ndarray) the input observation
- state – (np.ndarray) The last states (can be None, used in recurrent policies)
- mask – (np.ndarray) The last masks (can be None, used in recurrent policies)
- actions – (np.ndarray) (OPTIONAL) For calculating the likelihood that the given actions are chosen by the model for each of the given parameters. Must have the same number of actions and observations. (set to None to return the complete action probability distribution)
- logp – (bool) (OPTIONAL) When specified with actions, returns probability in log-space. This has no effect if actions is None.
Returns: (np.ndarray) the model’s (log) action probability
-
get_env()¶ returns the current environment (can be None if not defined)
Returns: (Gym Environment) The current environment
-
get_parameter_list()¶ Get tensorflow Variables of model’s parameters
This includes all variables necessary for continuing training (saving / loading).
Returns: (list) List of tensorflow Variables
-
get_parameters()¶ Get current model parameters as dictionary of variable name -> ndarray.
Returns: (OrderedDict) Dictionary of variable name -> ndarray of model’s parameters.
-
get_vec_normalize_env() → Optional[stable_baselines.common.vec_env.vec_normalize.VecNormalize]¶ Return the
VecNormalizewrapper of the training env if it exists.Returns: Optional[VecNormalize] The VecNormalizeenv.
-
learn(total_timesteps, callback=None, log_interval=100, tb_log_name='PPO1', reset_num_timesteps=True)[source]¶ Return a trained model.
Parameters: - total_timesteps – (int) The total number of samples to train on
- callback – (Union[callable, [callable], BaseCallback]) function called at every steps with state of the algorithm. It takes the local and global variables. If it returns False, training is aborted. When the callback inherits from BaseCallback, you will have access to additional stages of the training (training start/end), please read the documentation for more details.
- log_interval – (int) The number of timesteps before logging.
- tb_log_name – (str) the name of the run for tensorboard log
- reset_num_timesteps – (bool) whether or not to reset the current timestep number (used in logging)
Returns: (BaseRLModel) the trained model
-
classmethod
load(load_path, env=None, custom_objects=None, **kwargs)¶ Load the model from file
Parameters: - load_path – (str or file-like) the saved parameter location
- env – (Gym Environment) the new environment to run the loaded model on (can be None if you only need prediction from a trained model)
- custom_objects – (dict) Dictionary of objects to replace upon loading. If a variable is present in this dictionary as a key, it will not be deserialized and the corresponding item will be used instead. Similar to custom_objects in keras.models.load_model. Useful when you have an object in file that can not be deserialized.
- kwargs – extra arguments to change the model when loading
-
load_parameters(load_path_or_dict, exact_match=True)¶ Load model parameters from a file or a dictionary
Dictionary keys should be tensorflow variable names, which can be obtained with
get_parametersfunction. Ifexact_matchis True, dictionary should contain keys for all model’s parameters, otherwise RunTimeError is raised. If False, only variables included in the dictionary will be updated.This does not load agent’s hyper-parameters.
Warning
This function does not update trainer/optimizer variables (e.g. momentum). As such training after using this function may lead to less-than-optimal results.
Parameters: - load_path_or_dict – (str or file-like or dict) Save parameter location or dict of parameters as variable.name -> ndarrays to be loaded.
- exact_match – (bool) If True, expects load dictionary to contain keys for all variables in the model. If False, loads parameters only for variables mentioned in the dictionary. Defaults to True.
-
predict(observation, state=None, mask=None, deterministic=False)¶ Get the model’s action from an observation
Parameters: - observation – (np.ndarray) the input observation
- state – (np.ndarray) The last states (can be None, used in recurrent policies)
- mask – (np.ndarray) The last masks (can be None, used in recurrent policies)
- deterministic – (bool) Whether or not to return deterministic actions.
Returns: (np.ndarray, np.ndarray) the model’s action and the next state (used in recurrent policies)
-
pretrain(dataset, n_epochs=10, learning_rate=0.0001, adam_epsilon=1e-08, val_interval=None)¶ Pretrain a model using behavior cloning: supervised learning given an expert dataset.
NOTE: only Box and Discrete spaces are supported for now.
Parameters: - dataset – (ExpertDataset) Dataset manager
- n_epochs – (int) Number of iterations on the training set
- learning_rate – (float) Learning rate
- adam_epsilon – (float) the epsilon value for the adam optimizer
- val_interval – (int) Report training and validation losses every n epochs. By default, every 10th of the maximum number of epochs.
Returns: (BaseRLModel) the pretrained model
-
save(save_path, cloudpickle=False)[source]¶ Save the current parameters to file
Parameters: - save_path – (str or file-like) The save location
- cloudpickle – (bool) Use older cloudpickle format instead of zip-archives.
-
set_env(env)¶ Checks the validity of the environment, and if it is coherent, set it as the current environment.
Parameters: env – (Gym Environment) The environment for learning a policy
-
set_random_seed(seed: Optional[int]) → None¶ Parameters: seed – (Optional[int]) Seed for the pseudo-random generators. If None, do not change the seeds.
PPO2¶
The Proximal Policy Optimization algorithm combines ideas from A2C (having multiple workers) and TRPO (it uses a trust region to improve the actor).
The main idea is that after an update, the new policy should be not too far form the old policy. For that, ppo uses clipping to avoid too large update.
Note
PPO2 is the implementation of OpenAI made for GPU. For multiprocessing, it uses vectorized environments compared to PPO1 which uses MPI.
Note
PPO2 contains several modifications from the original algorithm not documented by OpenAI: value function is also clipped and advantages are normalized.
Notes¶
- Original paper: https://arxiv.org/abs/1707.06347
- Clear explanation of PPO on Arxiv Insights channel: https://www.youtube.com/watch?v=5P7I-xPq8u8
- OpenAI blog post: https://blog.openai.com/openai-baselines-ppo/
python -m stable_baselines.ppo2.run_atariruns the algorithm for 40M- frames = 10M timesteps on an Atari game. See help (
-h) for more options.
python -m stable_baselines.ppo2.run_mujocoruns the algorithm for 1M- frames on a Mujoco environment.
Can I use?¶
- Recurrent policies: ✔️
- Multi processing: ✔️
- Gym spaces:
| Space | Action | Observation |
|---|---|---|
| Discrete | ✔️ | ✔️ |
| Box | ✔️ | ✔️ |
| MultiDiscrete | ✔️ | ✔️ |
| MultiBinary | ✔️ | ✔️ |
Example¶
Train a PPO agent on CartPole-v1 using 4 processes.
import gym
from stable_baselines.common.policies import MlpPolicy
from stable_baselines.common import make_vec_env
from stable_baselines import PPO2
# multiprocess environment
env = make_vec_env('CartPole-v1', n_envs=4)
model = PPO2(MlpPolicy, env, verbose=1)
model.learn(total_timesteps=25000)
model.save("ppo2_cartpole")
del model # remove to demonstrate saving and loading
model = PPO2.load("ppo2_cartpole")
# Enjoy trained agent
obs = env.reset()
while True:
action, _states = model.predict(obs)
obs, rewards, dones, info = env.step(action)
env.render()
Parameters¶
-
class
stable_baselines.ppo2.PPO2(policy, env, gamma=0.99, n_steps=128, ent_coef=0.01, learning_rate=0.00025, vf_coef=0.5, max_grad_norm=0.5, lam=0.95, nminibatches=4, noptepochs=4, cliprange=0.2, cliprange_vf=None, verbose=0, tensorboard_log=None, _init_setup_model=True, policy_kwargs=None, full_tensorboard_log=False, seed=None, n_cpu_tf_sess=None)[source]¶ Proximal Policy Optimization algorithm (GPU version). Paper: https://arxiv.org/abs/1707.06347
Parameters: - policy – (ActorCriticPolicy or str) The policy model to use (MlpPolicy, CnnPolicy, CnnLstmPolicy, …)
- env – (Gym environment or str) The environment to learn from (if registered in Gym, can be str)
- gamma – (float) Discount factor
- n_steps – (int) The number of steps to run for each environment per update (i.e. batch size is n_steps * n_env where n_env is number of environment copies running in parallel)
- ent_coef – (float) Entropy coefficient for the loss calculation
- learning_rate – (float or callable) The learning rate, it can be a function
- vf_coef – (float) Value function coefficient for the loss calculation
- max_grad_norm – (float) The maximum value for the gradient clipping
- lam – (float) Factor for trade-off of bias vs variance for Generalized Advantage Estimator
- nminibatches – (int) Number of training minibatches per update. For recurrent policies, the number of environments run in parallel should be a multiple of nminibatches.
- noptepochs – (int) Number of epoch when optimizing the surrogate
- cliprange – (float or callable) Clipping parameter, it can be a function
- cliprange_vf – (float or callable) Clipping parameter for the value function, it can be a function. This is a parameter specific to the OpenAI implementation. If None is passed (default), then cliprange (that is used for the policy) will be used. IMPORTANT: this clipping depends on the reward scaling. To deactivate value function clipping (and recover the original PPO implementation), you have to pass a negative value (e.g. -1).
- verbose – (int) the verbosity level: 0 none, 1 training information, 2 tensorflow debug
- tensorboard_log – (str) the log location for tensorboard (if None, no logging)
- _init_setup_model – (bool) Whether or not to build the network at the creation of the instance
- policy_kwargs – (dict) additional arguments to be passed to the policy on creation
- full_tensorboard_log – (bool) enable additional logging when using tensorboard WARNING: this logging can take a lot of space quickly
- seed – (int) Seed for the pseudo-random generators (python, numpy, tensorflow). If None (default), use random seed. Note that if you want completely deterministic results, you must set n_cpu_tf_sess to 1.
- n_cpu_tf_sess – (int) The number of threads for TensorFlow operations If None, the number of cpu of the current machine will be used.
-
action_probability(observation, state=None, mask=None, actions=None, logp=False)¶ If
actionsisNone, then get the model’s action probability distribution from a given observation.- Depending on the action space the output is:
- Discrete: probability for each possible action
- Box: mean and standard deviation of the action output
However if
actionsis notNone, this function will return the probability that the given actions are taken with the given parameters (observation, state, …) on this model. For discrete action spaces, it returns the probability mass; for continuous action spaces, the probability density. This is since the probability mass will always be zero in continuous spaces, see http://blog.christianperone.com/2019/01/ for a good explanationParameters: - observation – (np.ndarray) the input observation
- state – (np.ndarray) The last states (can be None, used in recurrent policies)
- mask – (np.ndarray) The last masks (can be None, used in recurrent policies)
- actions – (np.ndarray) (OPTIONAL) For calculating the likelihood that the given actions are chosen by the model for each of the given parameters. Must have the same number of actions and observations. (set to None to return the complete action probability distribution)
- logp – (bool) (OPTIONAL) When specified with actions, returns probability in log-space. This has no effect if actions is None.
Returns: (np.ndarray) the model’s (log) action probability
-
get_env()¶ returns the current environment (can be None if not defined)
Returns: (Gym Environment) The current environment
-
get_parameter_list()¶ Get tensorflow Variables of model’s parameters
This includes all variables necessary for continuing training (saving / loading).
Returns: (list) List of tensorflow Variables
-
get_parameters()¶ Get current model parameters as dictionary of variable name -> ndarray.
Returns: (OrderedDict) Dictionary of variable name -> ndarray of model’s parameters.
-
get_vec_normalize_env() → Optional[stable_baselines.common.vec_env.vec_normalize.VecNormalize]¶ Return the
VecNormalizewrapper of the training env if it exists.Returns: Optional[VecNormalize] The VecNormalizeenv.
-
learn(total_timesteps, callback=None, log_interval=1, tb_log_name='PPO2', reset_num_timesteps=True)[source]¶ Return a trained model.
Parameters: - total_timesteps – (int) The total number of samples to train on
- callback – (Union[callable, [callable], BaseCallback]) function called at every steps with state of the algorithm. It takes the local and global variables. If it returns False, training is aborted. When the callback inherits from BaseCallback, you will have access to additional stages of the training (training start/end), please read the documentation for more details.
- log_interval – (int) The number of timesteps before logging.
- tb_log_name – (str) the name of the run for tensorboard log
- reset_num_timesteps – (bool) whether or not to reset the current timestep number (used in logging)
Returns: (BaseRLModel) the trained model
-
classmethod
load(load_path, env=None, custom_objects=None, **kwargs)¶ Load the model from file
Parameters: - load_path – (str or file-like) the saved parameter location
- env – (Gym Environment) the new environment to run the loaded model on (can be None if you only need prediction from a trained model)
- custom_objects – (dict) Dictionary of objects to replace upon loading. If a variable is present in this dictionary as a key, it will not be deserialized and the corresponding item will be used instead. Similar to custom_objects in keras.models.load_model. Useful when you have an object in file that can not be deserialized.
- kwargs – extra arguments to change the model when loading
-
load_parameters(load_path_or_dict, exact_match=True)¶ Load model parameters from a file or a dictionary
Dictionary keys should be tensorflow variable names, which can be obtained with
get_parametersfunction. Ifexact_matchis True, dictionary should contain keys for all model’s parameters, otherwise RunTimeError is raised. If False, only variables included in the dictionary will be updated.This does not load agent’s hyper-parameters.
Warning
This function does not update trainer/optimizer variables (e.g. momentum). As such training after using this function may lead to less-than-optimal results.
Parameters: - load_path_or_dict – (str or file-like or dict) Save parameter location or dict of parameters as variable.name -> ndarrays to be loaded.
- exact_match – (bool) If True, expects load dictionary to contain keys for all variables in the model. If False, loads parameters only for variables mentioned in the dictionary. Defaults to True.
-
predict(observation, state=None, mask=None, deterministic=False)¶ Get the model’s action from an observation
Parameters: - observation – (np.ndarray) the input observation
- state – (np.ndarray) The last states (can be None, used in recurrent policies)
- mask – (np.ndarray) The last masks (can be None, used in recurrent policies)
- deterministic – (bool) Whether or not to return deterministic actions.
Returns: (np.ndarray, np.ndarray) the model’s action and the next state (used in recurrent policies)
-
pretrain(dataset, n_epochs=10, learning_rate=0.0001, adam_epsilon=1e-08, val_interval=None)¶ Pretrain a model using behavior cloning: supervised learning given an expert dataset.
NOTE: only Box and Discrete spaces are supported for now.
Parameters: - dataset – (ExpertDataset) Dataset manager
- n_epochs – (int) Number of iterations on the training set
- learning_rate – (float) Learning rate
- adam_epsilon – (float) the epsilon value for the adam optimizer
- val_interval – (int) Report training and validation losses every n epochs. By default, every 10th of the maximum number of epochs.
Returns: (BaseRLModel) the pretrained model
-
save(save_path, cloudpickle=False)[source]¶ Save the current parameters to file
Parameters: - save_path – (str or file-like) The save location
- cloudpickle – (bool) Use older cloudpickle format instead of zip-archives.
-
set_env(env)¶ Checks the validity of the environment, and if it is coherent, set it as the current environment.
Parameters: env – (Gym Environment) The environment for learning a policy
-
set_random_seed(seed: Optional[int]) → None¶ Parameters: seed – (Optional[int]) Seed for the pseudo-random generators. If None, do not change the seeds.
SAC¶
Soft Actor Critic (SAC) Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actor.
SAC is the successor of Soft Q-Learning SQL and incorporates the double Q-learning trick from TD3. A key feature of SAC, and a major difference with common RL algorithms, is that it is trained to maximize a trade-off between expected return and entropy, a measure of randomness in the policy.
Warning
The SAC model does not support stable_baselines.common.policies because it uses double q-values
and value estimation, as a result it must use its own policy models (see SAC Policies).
Available Policies
MlpPolicy |
Policy object that implements actor critic, using a MLP (2 layers of 64) |
LnMlpPolicy |
Policy object that implements actor critic, using a MLP (2 layers of 64), with layer normalisation |
CnnPolicy |
Policy object that implements actor critic, using a CNN (the nature CNN) |
LnCnnPolicy |
Policy object that implements actor critic, using a CNN (the nature CNN), with layer normalisation |
Notes¶
- Original paper: https://arxiv.org/abs/1801.01290
- OpenAI Spinning Guide for SAC: https://spinningup.openai.com/en/latest/algorithms/sac.html
- Original Implementation: https://github.com/haarnoja/sac
- Blog post on using SAC with real robots: https://bair.berkeley.edu/blog/2018/12/14/sac/
Note
In our implementation, we use an entropy coefficient (as in OpenAI Spinning or Facebook Horizon), which is the equivalent to the inverse of reward scale in the original SAC paper. The main reason is that it avoids having too high errors when updating the Q functions.
Note
The default policies for SAC differ a bit from others MlpPolicy: it uses ReLU instead of tanh activation, to match the original paper
Can I use?¶
- Recurrent policies: ❌
- Multi processing: ❌
- Gym spaces:
| Space | Action | Observation |
|---|---|---|
| Discrete | ❌ | ✔️ |
| Box | ✔️ | ✔️ |
| MultiDiscrete | ❌ | ✔️ |
| MultiBinary | ❌ | ✔️ |
Example¶
import gym
import numpy as np
from stable_baselines.sac.policies import MlpPolicy
from stable_baselines import SAC
env = gym.make('Pendulum-v0')
model = SAC(MlpPolicy, env, verbose=1)
model.learn(total_timesteps=50000, log_interval=10)
model.save("sac_pendulum")
del model # remove to demonstrate saving and loading
model = SAC.load("sac_pendulum")
obs = env.reset()
while True:
action, _states = model.predict(obs)
obs, rewards, dones, info = env.step(action)
env.render()
Parameters¶
-
class
stable_baselines.sac.SAC(policy, env, gamma=0.99, learning_rate=0.0003, buffer_size=50000, learning_starts=100, train_freq=1, batch_size=64, tau=0.005, ent_coef='auto', target_update_interval=1, gradient_steps=1, target_entropy='auto', action_noise=None, random_exploration=0.0, verbose=0, tensorboard_log=None, _init_setup_model=True, policy_kwargs=None, full_tensorboard_log=False, seed=None, n_cpu_tf_sess=None)[source]¶ Soft Actor-Critic (SAC) Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actor, This implementation borrows code from original implementation (https://github.com/haarnoja/sac) from OpenAI Spinning Up (https://github.com/openai/spinningup) and from the Softlearning repo (https://github.com/rail-berkeley/softlearning/) Paper: https://arxiv.org/abs/1801.01290 Introduction to SAC: https://spinningup.openai.com/en/latest/algorithms/sac.html
Parameters: - policy – (SACPolicy or str) The policy model to use (MlpPolicy, CnnPolicy, LnMlpPolicy, …)
- env – (Gym environment or str) The environment to learn from (if registered in Gym, can be str)
- gamma – (float) the discount factor
- learning_rate – (float or callable) learning rate for adam optimizer, the same learning rate will be used for all networks (Q-Values, Actor and Value function) it can be a function of the current progress (from 1 to 0)
- buffer_size – (int) size of the replay buffer
- batch_size – (int) Minibatch size for each gradient update
- tau – (float) the soft update coefficient (“polyak update”, between 0 and 1)
- ent_coef – (str or float) Entropy regularization coefficient. (Equivalent to inverse of reward scale in the original SAC paper.) Controlling exploration/exploitation trade-off. Set it to ‘auto’ to learn it automatically (and ‘auto_0.1’ for using 0.1 as initial value)
- train_freq – (int) Update the model every train_freq steps.
- learning_starts – (int) how many steps of the model to collect transitions for before learning starts
- target_update_interval – (int) update the target network every target_network_update_freq steps.
- gradient_steps – (int) How many gradient update after each step
- target_entropy – (str or float) target entropy when learning ent_coef (ent_coef = ‘auto’)
- action_noise – (ActionNoise) the action noise type (None by default), this can help for hard exploration problem. Cf DDPG for the different action noise type.
- random_exploration – (float) Probability of taking a random action (as in an epsilon-greedy strategy) This is not needed for SAC normally but can help exploring when using HER + SAC. This hack was present in the original OpenAI Baselines repo (DDPG + HER)
- verbose – (int) the verbosity level: 0 none, 1 training information, 2 tensorflow debug
- tensorboard_log – (str) the log location for tensorboard (if None, no logging)
- _init_setup_model – (bool) Whether or not to build the network at the creation of the instance
- policy_kwargs – (dict) additional arguments to be passed to the policy on creation
- full_tensorboard_log – (bool) enable additional logging when using tensorboard Note: this has no effect on SAC logging for now
- seed – (int) Seed for the pseudo-random generators (python, numpy, tensorflow). If None (default), use random seed. Note that if you want completely deterministic results, you must set n_cpu_tf_sess to 1.
- n_cpu_tf_sess – (int) The number of threads for TensorFlow operations If None, the number of cpu of the current machine will be used.
-
action_probability(observation, state=None, mask=None, actions=None, logp=False)[source]¶ If
actionsisNone, then get the model’s action probability distribution from a given observation.- Depending on the action space the output is:
- Discrete: probability for each possible action
- Box: mean and standard deviation of the action output
However if
actionsis notNone, this function will return the probability that the given actions are taken with the given parameters (observation, state, …) on this model. For discrete action spaces, it returns the probability mass; for continuous action spaces, the probability density. This is since the probability mass will always be zero in continuous spaces, see http://blog.christianperone.com/2019/01/ for a good explanationParameters: - observation – (np.ndarray) the input observation
- state – (np.ndarray) The last states (can be None, used in recurrent policies)
- mask – (np.ndarray) The last masks (can be None, used in recurrent policies)
- actions – (np.ndarray) (OPTIONAL) For calculating the likelihood that the given actions are chosen by the model for each of the given parameters. Must have the same number of actions and observations. (set to None to return the complete action probability distribution)
- logp – (bool) (OPTIONAL) When specified with actions, returns probability in log-space. This has no effect if actions is None.
Returns: (np.ndarray) the model’s (log) action probability
-
get_env()¶ returns the current environment (can be None if not defined)
Returns: (Gym Environment) The current environment
-
get_parameter_list()[source]¶ Get tensorflow Variables of model’s parameters
This includes all variables necessary for continuing training (saving / loading).
Returns: (list) List of tensorflow Variables
-
get_parameters()¶ Get current model parameters as dictionary of variable name -> ndarray.
Returns: (OrderedDict) Dictionary of variable name -> ndarray of model’s parameters.
-
get_vec_normalize_env() → Optional[stable_baselines.common.vec_env.vec_normalize.VecNormalize]¶ Return the
VecNormalizewrapper of the training env if it exists.Returns: Optional[VecNormalize] The VecNormalizeenv.
-
learn(total_timesteps, callback=None, log_interval=4, tb_log_name='SAC', reset_num_timesteps=True, replay_wrapper=None)[source]¶ Return a trained model.
Parameters: - total_timesteps – (int) The total number of samples to train on
- callback – (Union[callable, [callable], BaseCallback]) function called at every steps with state of the algorithm. It takes the local and global variables. If it returns False, training is aborted. When the callback inherits from BaseCallback, you will have access to additional stages of the training (training start/end), please read the documentation for more details.
- log_interval – (int) The number of timesteps before logging.
- tb_log_name – (str) the name of the run for tensorboard log
- reset_num_timesteps – (bool) whether or not to reset the current timestep number (used in logging)
Returns: (BaseRLModel) the trained model
-
classmethod
load(load_path, env=None, custom_objects=None, **kwargs)¶ Load the model from file
Parameters: - load_path – (str or file-like) the saved parameter location
- env – (Gym Environment) the new environment to run the loaded model on (can be None if you only need prediction from a trained model)
- custom_objects – (dict) Dictionary of objects to replace upon loading. If a variable is present in this dictionary as a key, it will not be deserialized and the corresponding item will be used instead. Similar to custom_objects in keras.models.load_model. Useful when you have an object in file that can not be deserialized.
- kwargs – extra arguments to change the model when loading
-
load_parameters(load_path_or_dict, exact_match=True)¶ Load model parameters from a file or a dictionary
Dictionary keys should be tensorflow variable names, which can be obtained with
get_parametersfunction. Ifexact_matchis True, dictionary should contain keys for all model’s parameters, otherwise RunTimeError is raised. If False, only variables included in the dictionary will be updated.This does not load agent’s hyper-parameters.
Warning
This function does not update trainer/optimizer variables (e.g. momentum). As such training after using this function may lead to less-than-optimal results.
Parameters: - load_path_or_dict – (str or file-like or dict) Save parameter location or dict of parameters as variable.name -> ndarrays to be loaded.
- exact_match – (bool) If True, expects load dictionary to contain keys for all variables in the model. If False, loads parameters only for variables mentioned in the dictionary. Defaults to True.
-
predict(observation, state=None, mask=None, deterministic=True)[source]¶ Get the model’s action from an observation
Parameters: - observation – (np.ndarray) the input observation
- state – (np.ndarray) The last states (can be None, used in recurrent policies)
- mask – (np.ndarray) The last masks (can be None, used in recurrent policies)
- deterministic – (bool) Whether or not to return deterministic actions.
Returns: (np.ndarray, np.ndarray) the model’s action and the next state (used in recurrent policies)
-
pretrain(dataset, n_epochs=10, learning_rate=0.0001, adam_epsilon=1e-08, val_interval=None)¶ Pretrain a model using behavior cloning: supervised learning given an expert dataset.
NOTE: only Box and Discrete spaces are supported for now.
Parameters: - dataset – (ExpertDataset) Dataset manager
- n_epochs – (int) Number of iterations on the training set
- learning_rate – (float) Learning rate
- adam_epsilon – (float) the epsilon value for the adam optimizer
- val_interval – (int) Report training and validation losses every n epochs. By default, every 10th of the maximum number of epochs.
Returns: (BaseRLModel) the pretrained model
-
save(save_path, cloudpickle=False)[source]¶ Save the current parameters to file
Parameters: - save_path – (str or file-like) The save location
- cloudpickle – (bool) Use older cloudpickle format instead of zip-archives.
-
set_env(env)¶ Checks the validity of the environment, and if it is coherent, set it as the current environment.
Parameters: env – (Gym Environment) The environment for learning a policy
-
set_random_seed(seed: Optional[int]) → None¶ Parameters: seed – (Optional[int]) Seed for the pseudo-random generators. If None, do not change the seeds.
SAC Policies¶
-
class
stable_baselines.sac.MlpPolicy(sess, ob_space, ac_space, n_env=1, n_steps=1, n_batch=None, reuse=False, **_kwargs)[source]¶ Policy object that implements actor critic, using a MLP (2 layers of 64)
Parameters: - sess – (TensorFlow session) The current TensorFlow session
- ob_space – (Gym Space) The observation space of the environment
- ac_space – (Gym Space) The action space of the environment
- n_env – (int) The number of environments to run
- n_steps – (int) The number of steps to run for each environment
- n_batch – (int) The number of batch to run (n_envs * n_steps)
- reuse – (bool) If the policy is reusable or not
- _kwargs – (dict) Extra keyword arguments for the nature CNN feature extraction
-
action_ph¶ tf.Tensor: placeholder for actions, shape (self.n_batch, ) + self.ac_space.shape.
-
initial_state¶ The initial state of the policy. For feedforward policies, None. For a recurrent policy, a NumPy array of shape (self.n_env, ) + state_shape.
-
is_discrete¶ bool: is action space discrete.
-
make_actor(obs=None, reuse=False, scope='pi')¶ Creates an actor object
Parameters: - obs – (TensorFlow Tensor) The observation placeholder (can be None for default placeholder)
- reuse – (bool) whether or not to reuse parameters
- scope – (str) the scope name of the actor
Returns: (TensorFlow Tensor) the output tensor
-
make_critics(obs=None, action=None, reuse=False, scope='values_fn', create_vf=True, create_qf=True)¶ Creates the two Q-Values approximator along with the Value function
Parameters: - obs – (TensorFlow Tensor) The observation placeholder (can be None for default placeholder)
- action – (TensorFlow Tensor) The action placeholder
- reuse – (bool) whether or not to reuse parameters
- scope – (str) the scope name
- create_vf – (bool) Whether to create Value fn or not
- create_qf – (bool) Whether to create Q-Values fn or not
Returns: ([tf.Tensor]) Mean, action and log probability
-
obs_ph¶ tf.Tensor: placeholder for observations, shape (self.n_batch, ) + self.ob_space.shape.
-
proba_step(obs, state=None, mask=None)¶ Returns the action probability params (mean, std) for a single step
Parameters: - obs – ([float] or [int]) The current observation of the environment
- state – ([float]) The last states (used in recurrent policies)
- mask – ([float]) The last masks (used in recurrent policies)
Returns: ([float], [float])
-
processed_obs¶ tf.Tensor: processed observations, shape (self.n_batch, ) + self.ob_space.shape.
The form of processing depends on the type of the observation space, and the parameters whether scale is passed to the constructor; see observation_input for more information.
-
step(obs, state=None, mask=None, deterministic=False)¶ Returns the policy for a single step
Parameters: - obs – ([float] or [int]) The current observation of the environment
- state – ([float]) The last states (used in recurrent policies)
- mask – ([float]) The last masks (used in recurrent policies)
- deterministic – (bool) Whether or not to return deterministic actions.
Returns: ([float]) actions
-
class
stable_baselines.sac.LnMlpPolicy(sess, ob_space, ac_space, n_env=1, n_steps=1, n_batch=None, reuse=False, **_kwargs)[source]¶ Policy object that implements actor critic, using a MLP (2 layers of 64), with layer normalisation
Parameters: - sess – (TensorFlow session) The current TensorFlow session
- ob_space – (Gym Space) The observation space of the environment
- ac_space – (Gym Space) The action space of the environment
- n_env – (int) The number of environments to run
- n_steps – (int) The number of steps to run for each environment
- n_batch – (int) The number of batch to run (n_envs * n_steps)
- reuse – (bool) If the policy is reusable or not
- _kwargs – (dict) Extra keyword arguments for the nature CNN feature extraction
-
action_ph¶ tf.Tensor: placeholder for actions, shape (self.n_batch, ) + self.ac_space.shape.
-
initial_state¶ The initial state of the policy. For feedforward policies, None. For a recurrent policy, a NumPy array of shape (self.n_env, ) + state_shape.
-
is_discrete¶ bool: is action space discrete.
-
make_actor(obs=None, reuse=False, scope='pi')¶ Creates an actor object
Parameters: - obs – (TensorFlow Tensor) The observation placeholder (can be None for default placeholder)
- reuse – (bool) whether or not to reuse parameters
- scope – (str) the scope name of the actor
Returns: (TensorFlow Tensor) the output tensor
-
make_critics(obs=None, action=None, reuse=False, scope='values_fn', create_vf=True, create_qf=True)¶ Creates the two Q-Values approximator along with the Value function
Parameters: - obs – (TensorFlow Tensor) The observation placeholder (can be None for default placeholder)
- action – (TensorFlow Tensor) The action placeholder
- reuse – (bool) whether or not to reuse parameters
- scope – (str) the scope name
- create_vf – (bool) Whether to create Value fn or not
- create_qf – (bool) Whether to create Q-Values fn or not
Returns: ([tf.Tensor]) Mean, action and log probability
-
obs_ph¶ tf.Tensor: placeholder for observations, shape (self.n_batch, ) + self.ob_space.shape.
-
proba_step(obs, state=None, mask=None)¶ Returns the action probability params (mean, std) for a single step
Parameters: - obs – ([float] or [int]) The current observation of the environment
- state – ([float]) The last states (used in recurrent policies)
- mask – ([float]) The last masks (used in recurrent policies)
Returns: ([float], [float])
-
processed_obs¶ tf.Tensor: processed observations, shape (self.n_batch, ) + self.ob_space.shape.
The form of processing depends on the type of the observation space, and the parameters whether scale is passed to the constructor; see observation_input for more information.
-
step(obs, state=None, mask=None, deterministic=False)¶ Returns the policy for a single step
Parameters: - obs – ([float] or [int]) The current observation of the environment
- state – ([float]) The last states (used in recurrent policies)
- mask – ([float]) The last masks (used in recurrent policies)
- deterministic – (bool) Whether or not to return deterministic actions.
Returns: ([float]) actions
-
class
stable_baselines.sac.CnnPolicy(sess, ob_space, ac_space, n_env=1, n_steps=1, n_batch=None, reuse=False, **_kwargs)[source]¶ Policy object that implements actor critic, using a CNN (the nature CNN)
Parameters: - sess – (TensorFlow session) The current TensorFlow session
- ob_space – (Gym Space) The observation space of the environment
- ac_space – (Gym Space) The action space of the environment
- n_env – (int) The number of environments to run
- n_steps – (int) The number of steps to run for each environment
- n_batch – (int) The number of batch to run (n_envs * n_steps)
- reuse – (bool) If the policy is reusable or not
- _kwargs – (dict) Extra keyword arguments for the nature CNN feature extraction
-
action_ph¶ tf.Tensor: placeholder for actions, shape (self.n_batch, ) + self.ac_space.shape.
-
initial_state¶ The initial state of the policy. For feedforward policies, None. For a recurrent policy, a NumPy array of shape (self.n_env, ) + state_shape.
-
is_discrete¶ bool: is action space discrete.
-
make_actor(obs=None, reuse=False, scope='pi')¶ Creates an actor object
Parameters: - obs – (TensorFlow Tensor) The observation placeholder (can be None for default placeholder)
- reuse – (bool) whether or not to reuse parameters
- scope – (str) the scope name of the actor
Returns: (TensorFlow Tensor) the output tensor
-
make_critics(obs=None, action=None, reuse=False, scope='values_fn', create_vf=True, create_qf=True)¶ Creates the two Q-Values approximator along with the Value function
Parameters: - obs – (TensorFlow Tensor) The observation placeholder (can be None for default placeholder)
- action – (TensorFlow Tensor) The action placeholder
- reuse – (bool) whether or not to reuse parameters
- scope – (str) the scope name
- create_vf – (bool) Whether to create Value fn or not
- create_qf – (bool) Whether to create Q-Values fn or not
Returns: ([tf.Tensor]) Mean, action and log probability
-
obs_ph¶ tf.Tensor: placeholder for observations, shape (self.n_batch, ) + self.ob_space.shape.
-
proba_step(obs, state=None, mask=None)¶ Returns the action probability params (mean, std) for a single step
Parameters: - obs – ([float] or [int]) The current observation of the environment
- state – ([float]) The last states (used in recurrent policies)
- mask – ([float]) The last masks (used in recurrent policies)
Returns: ([float], [float])
-
processed_obs¶ tf.Tensor: processed observations, shape (self.n_batch, ) + self.ob_space.shape.
The form of processing depends on the type of the observation space, and the parameters whether scale is passed to the constructor; see observation_input for more information.
-
step(obs, state=None, mask=None, deterministic=False)¶ Returns the policy for a single step
Parameters: - obs – ([float] or [int]) The current observation of the environment
- state – ([float]) The last states (used in recurrent policies)
- mask – ([float]) The last masks (used in recurrent policies)
- deterministic – (bool) Whether or not to return deterministic actions.
Returns: ([float]) actions
-
class
stable_baselines.sac.LnCnnPolicy(sess, ob_space, ac_space, n_env=1, n_steps=1, n_batch=None, reuse=False, **_kwargs)[source]¶ Policy object that implements actor critic, using a CNN (the nature CNN), with layer normalisation
Parameters: - sess – (TensorFlow session) The current TensorFlow session
- ob_space – (Gym Space) The observation space of the environment
- ac_space – (Gym Space) The action space of the environment
- n_env – (int) The number of environments to run
- n_steps – (int) The number of steps to run for each environment
- n_batch – (int) The number of batch to run (n_envs * n_steps)
- reuse – (bool) If the policy is reusable or not
- _kwargs – (dict) Extra keyword arguments for the nature CNN feature extraction
-
action_ph¶ tf.Tensor: placeholder for actions, shape (self.n_batch, ) + self.ac_space.shape.
-
initial_state¶ The initial state of the policy. For feedforward policies, None. For a recurrent policy, a NumPy array of shape (self.n_env, ) + state_shape.
-
is_discrete¶ bool: is action space discrete.
-
make_actor(obs=None, reuse=False, scope='pi')¶ Creates an actor object
Parameters: - obs – (TensorFlow Tensor) The observation placeholder (can be None for default placeholder)
- reuse – (bool) whether or not to reuse parameters
- scope – (str) the scope name of the actor
Returns: (TensorFlow Tensor) the output tensor
-
make_critics(obs=None, action=None, reuse=False, scope='values_fn', create_vf=True, create_qf=True)¶ Creates the two Q-Values approximator along with the Value function
Parameters: - obs – (TensorFlow Tensor) The observation placeholder (can be None for default placeholder)
- action – (TensorFlow Tensor) The action placeholder
- reuse – (bool) whether or not to reuse parameters
- scope – (str) the scope name
- create_vf – (bool) Whether to create Value fn or not
- create_qf – (bool) Whether to create Q-Values fn or not
Returns: ([tf.Tensor]) Mean, action and log probability
-
obs_ph¶ tf.Tensor: placeholder for observations, shape (self.n_batch, ) + self.ob_space.shape.
-
proba_step(obs, state=None, mask=None)¶ Returns the action probability params (mean, std) for a single step
Parameters: - obs – ([float] or [int]) The current observation of the environment
- state – ([float]) The last states (used in recurrent policies)
- mask – ([float]) The last masks (used in recurrent policies)
Returns: ([float], [float])
-
processed_obs¶ tf.Tensor: processed observations, shape (self.n_batch, ) + self.ob_space.shape.
The form of processing depends on the type of the observation space, and the parameters whether scale is passed to the constructor; see observation_input for more information.
-
step(obs, state=None, mask=None, deterministic=False)¶ Returns the policy for a single step
Parameters: - obs – ([float] or [int]) The current observation of the environment
- state – ([float]) The last states (used in recurrent policies)
- mask – ([float]) The last masks (used in recurrent policies)
- deterministic – (bool) Whether or not to return deterministic actions.
Returns: ([float]) actions
Custom Policy Network¶
Similarly to the example given in the examples page. You can easily define a custom architecture for the policy network:
import gym
from stable_baselines.sac.policies import FeedForwardPolicy
from stable_baselines.common.vec_env import DummyVecEnv
from stable_baselines import SAC
# Custom MLP policy of three layers of size 128 each
class CustomSACPolicy(FeedForwardPolicy):
def __init__(self, *args, **kwargs):
super(CustomSACPolicy, self).__init__(*args, **kwargs,
layers=[128, 128, 128],
layer_norm=False,
feature_extraction="mlp")
# Create and wrap the environment
env = gym.make('Pendulum-v0')
env = DummyVecEnv([lambda: env])
model = SAC(CustomSACPolicy, env, verbose=1)
# Train the agent
model.learn(total_timesteps=100000)
TD3¶
Twin Delayed DDPG (TD3) Addressing Function Approximation Error in Actor-Critic Methods.
TD3 is a direct successor of DDPG and improves it using three major tricks: clipped double Q-Learning, delayed policy update and target policy smoothing. We recommend reading OpenAI Spinning guide on TD3 to learn more about those.
Warning
The TD3 model does not support stable_baselines.common.policies because it uses double q-values
estimation, as a result it must use its own policy models (see TD3 Policies).
Available Policies
MlpPolicy |
Policy object that implements actor critic, using a MLP (2 layers of 64) |
LnMlpPolicy |
Policy object that implements actor critic, using a MLP (2 layers of 64), with layer normalisation |
CnnPolicy |
Policy object that implements actor critic, using a CNN (the nature CNN) |
LnCnnPolicy |
Policy object that implements actor critic, using a CNN (the nature CNN), with layer normalisation |
Notes¶
- Original paper: https://arxiv.org/pdf/1802.09477.pdf
- OpenAI Spinning Guide for TD3: https://spinningup.openai.com/en/latest/algorithms/td3.html
- Original Implementation: https://github.com/sfujim/TD3
Note
The default policies for TD3 differ a bit from others MlpPolicy: it uses ReLU instead of tanh activation, to match the original paper
Can I use?¶
- Recurrent policies: ❌
- Multi processing: ❌
- Gym spaces:
| Space | Action | Observation |
|---|---|---|
| Discrete | ❌ | ✔️ |
| Box | ✔️ | ✔️ |
| MultiDiscrete | ❌ | ✔️ |
| MultiBinary | ❌ | ✔️ |
Example¶
import gym
import numpy as np
from stable_baselines import TD3
from stable_baselines.td3.policies import MlpPolicy
from stable_baselines.common.vec_env import DummyVecEnv
from stable_baselines.ddpg.noise import NormalActionNoise, OrnsteinUhlenbeckActionNoise
env = gym.make('Pendulum-v0')
# The noise objects for TD3
n_actions = env.action_space.shape[-1]
action_noise = NormalActionNoise(mean=np.zeros(n_actions), sigma=0.1 * np.ones(n_actions))
model = TD3(MlpPolicy, env, action_noise=action_noise, verbose=1)
model.learn(total_timesteps=50000, log_interval=10)
model.save("td3_pendulum")
del model # remove to demonstrate saving and loading
model = TD3.load("td3_pendulum")
obs = env.reset()
while True:
action, _states = model.predict(obs)
obs, rewards, dones, info = env.step(action)
env.render()
Parameters¶
-
class
stable_baselines.td3.TD3(policy, env, gamma=0.99, learning_rate=0.0003, buffer_size=50000, learning_starts=100, train_freq=100, gradient_steps=100, batch_size=128, tau=0.005, policy_delay=2, action_noise=None, target_policy_noise=0.2, target_noise_clip=0.5, random_exploration=0.0, verbose=0, tensorboard_log=None, _init_setup_model=True, policy_kwargs=None, full_tensorboard_log=False, seed=None, n_cpu_tf_sess=None)[source]¶ Twin Delayed DDPG (TD3) Addressing Function Approximation Error in Actor-Critic Methods.
Original implementation: https://github.com/sfujim/TD3 Paper: https://arxiv.org/pdf/1802.09477.pdf Introduction to TD3: https://spinningup.openai.com/en/latest/algorithms/td3.html
Parameters: - policy – (TD3Policy or str) The policy model to use (MlpPolicy, CnnPolicy, LnMlpPolicy, …)
- env – (Gym environment or str) The environment to learn from (if registered in Gym, can be str)
- gamma – (float) the discount factor
- learning_rate – (float or callable) learning rate for adam optimizer, the same learning rate will be used for all networks (Q-Values and Actor networks) it can be a function of the current progress (from 1 to 0)
- buffer_size – (int) size of the replay buffer
- batch_size – (int) Minibatch size for each gradient update
- tau – (float) the soft update coefficient (“polyak update” of the target networks, between 0 and 1)
- policy_delay – (int) Policy and target networks will only be updated once every policy_delay steps per training steps. The Q values will be updated policy_delay more often (update every training step).
- action_noise – (ActionNoise) the action noise type. Cf DDPG for the different action noise type.
- target_policy_noise – (float) Standard deviation of Gaussian noise added to target policy (smoothing noise)
- target_noise_clip – (float) Limit for absolute value of target policy smoothing noise.
- train_freq – (int) Update the model every train_freq steps.
- learning_starts – (int) how many steps of the model to collect transitions for before learning starts
- gradient_steps – (int) How many gradient update after each step
- random_exploration – (float) Probability of taking a random action (as in an epsilon-greedy strategy) This is not needed for TD3 normally but can help exploring when using HER + TD3. This hack was present in the original OpenAI Baselines repo (DDPG + HER)
- verbose – (int) the verbosity level: 0 none, 1 training information, 2 tensorflow debug
- tensorboard_log – (str) the log location for tensorboard (if None, no logging)
- _init_setup_model – (bool) Whether or not to build the network at the creation of the instance
- policy_kwargs – (dict) additional arguments to be passed to the policy on creation
- full_tensorboard_log – (bool) enable additional logging when using tensorboard Note: this has no effect on TD3 logging for now
- seed – (int) Seed for the pseudo-random generators (python, numpy, tensorflow). If None (default), use random seed. Note that if you want completely deterministic results, you must set n_cpu_tf_sess to 1.
- n_cpu_tf_sess – (int) The number of threads for TensorFlow operations If None, the number of cpu of the current machine will be used.
-
action_probability(observation, state=None, mask=None, actions=None, logp=False)[source]¶ If
actionsisNone, then get the model’s action probability distribution from a given observation.- Depending on the action space the output is:
- Discrete: probability for each possible action
- Box: mean and standard deviation of the action output
However if
actionsis notNone, this function will return the probability that the given actions are taken with the given parameters (observation, state, …) on this model. For discrete action spaces, it returns the probability mass; for continuous action spaces, the probability density. This is since the probability mass will always be zero in continuous spaces, see http://blog.christianperone.com/2019/01/ for a good explanationParameters: - observation – (np.ndarray) the input observation
- state – (np.ndarray) The last states (can be None, used in recurrent policies)
- mask – (np.ndarray) The last masks (can be None, used in recurrent policies)
- actions – (np.ndarray) (OPTIONAL) For calculating the likelihood that the given actions are chosen by the model for each of the given parameters. Must have the same number of actions and observations. (set to None to return the complete action probability distribution)
- logp – (bool) (OPTIONAL) When specified with actions, returns probability in log-space. This has no effect if actions is None.
Returns: (np.ndarray) the model’s (log) action probability
-
get_env()¶ returns the current environment (can be None if not defined)
Returns: (Gym Environment) The current environment
-
get_parameter_list()[source]¶ Get tensorflow Variables of model’s parameters
This includes all variables necessary for continuing training (saving / loading).
Returns: (list) List of tensorflow Variables
-
get_parameters()¶ Get current model parameters as dictionary of variable name -> ndarray.
Returns: (OrderedDict) Dictionary of variable name -> ndarray of model’s parameters.
-
get_vec_normalize_env() → Optional[stable_baselines.common.vec_env.vec_normalize.VecNormalize]¶ Return the
VecNormalizewrapper of the training env if it exists.Returns: Optional[VecNormalize] The VecNormalizeenv.
-
learn(total_timesteps, callback=None, log_interval=4, tb_log_name='TD3', reset_num_timesteps=True, replay_wrapper=None)[source]¶ Return a trained model.
Parameters: - total_timesteps – (int) The total number of samples to train on
- callback – (Union[callable, [callable], BaseCallback]) function called at every steps with state of the algorithm. It takes the local and global variables. If it returns False, training is aborted. When the callback inherits from BaseCallback, you will have access to additional stages of the training (training start/end), please read the documentation for more details.
- log_interval – (int) The number of timesteps before logging.
- tb_log_name – (str) the name of the run for tensorboard log
- reset_num_timesteps – (bool) whether or not to reset the current timestep number (used in logging)
Returns: (BaseRLModel) the trained model
-
classmethod
load(load_path, env=None, custom_objects=None, **kwargs)¶ Load the model from file
Parameters: - load_path – (str or file-like) the saved parameter location
- env – (Gym Environment) the new environment to run the loaded model on (can be None if you only need prediction from a trained model)
- custom_objects – (dict) Dictionary of objects to replace upon loading. If a variable is present in this dictionary as a key, it will not be deserialized and the corresponding item will be used instead. Similar to custom_objects in keras.models.load_model. Useful when you have an object in file that can not be deserialized.
- kwargs – extra arguments to change the model when loading
-
load_parameters(load_path_or_dict, exact_match=True)¶ Load model parameters from a file or a dictionary
Dictionary keys should be tensorflow variable names, which can be obtained with
get_parametersfunction. Ifexact_matchis True, dictionary should contain keys for all model’s parameters, otherwise RunTimeError is raised. If False, only variables included in the dictionary will be updated.This does not load agent’s hyper-parameters.
Warning
This function does not update trainer/optimizer variables (e.g. momentum). As such training after using this function may lead to less-than-optimal results.
Parameters: - load_path_or_dict – (str or file-like or dict) Save parameter location or dict of parameters as variable.name -> ndarrays to be loaded.
- exact_match – (bool) If True, expects load dictionary to contain keys for all variables in the model. If False, loads parameters only for variables mentioned in the dictionary. Defaults to True.
-
predict(observation, state=None, mask=None, deterministic=True)[source]¶ Get the model’s action from an observation
Parameters: - observation – (np.ndarray) the input observation
- state – (np.ndarray) The last states (can be None, used in recurrent policies)
- mask – (np.ndarray) The last masks (can be None, used in recurrent policies)
- deterministic – (bool) Whether or not to return deterministic actions.
Returns: (np.ndarray, np.ndarray) the model’s action and the next state (used in recurrent policies)
-
pretrain(dataset, n_epochs=10, learning_rate=0.0001, adam_epsilon=1e-08, val_interval=None)¶ Pretrain a model using behavior cloning: supervised learning given an expert dataset.
NOTE: only Box and Discrete spaces are supported for now.
Parameters: - dataset – (ExpertDataset) Dataset manager
- n_epochs – (int) Number of iterations on the training set
- learning_rate – (float) Learning rate
- adam_epsilon – (float) the epsilon value for the adam optimizer
- val_interval – (int) Report training and validation losses every n epochs. By default, every 10th of the maximum number of epochs.
Returns: (BaseRLModel) the pretrained model
-
save(save_path, cloudpickle=False)[source]¶ Save the current parameters to file
Parameters: - save_path – (str or file-like) The save location
- cloudpickle – (bool) Use older cloudpickle format instead of zip-archives.
-
set_env(env)¶ Checks the validity of the environment, and if it is coherent, set it as the current environment.
Parameters: env – (Gym Environment) The environment for learning a policy
-
set_random_seed(seed: Optional[int]) → None¶ Parameters: seed – (Optional[int]) Seed for the pseudo-random generators. If None, do not change the seeds.
TD3 Policies¶
-
class
stable_baselines.td3.MlpPolicy(sess, ob_space, ac_space, n_env=1, n_steps=1, n_batch=None, reuse=False, **_kwargs)[source]¶ Policy object that implements actor critic, using a MLP (2 layers of 64)
Parameters: - sess – (TensorFlow session) The current TensorFlow session
- ob_space – (Gym Space) The observation space of the environment
- ac_space – (Gym Space) The action space of the environment
- n_env – (int) The number of environments to run
- n_steps – (int) The number of steps to run for each environment
- n_batch – (int) The number of batch to run (n_envs * n_steps)
- reuse – (bool) If the policy is reusable or not
- _kwargs – (dict) Extra keyword arguments for the nature CNN feature extraction
-
action_ph¶ tf.Tensor: placeholder for actions, shape (self.n_batch, ) + self.ac_space.shape.
-
initial_state¶ The initial state of the policy. For feedforward policies, None. For a recurrent policy, a NumPy array of shape (self.n_env, ) + state_shape.
-
is_discrete¶ bool: is action space discrete.
-
make_actor(obs=None, reuse=False, scope='pi')¶ Creates an actor object
Parameters: - obs – (TensorFlow Tensor) The observation placeholder (can be None for default placeholder)
- reuse – (bool) whether or not to reuse parameters
- scope – (str) the scope name of the actor
Returns: (TensorFlow Tensor) the output tensor
-
make_critics(obs=None, action=None, reuse=False, scope='values_fn')¶ Creates the two Q-Values approximator
Parameters: - obs – (TensorFlow Tensor) The observation placeholder (can be None for default placeholder)
- action – (TensorFlow Tensor) The action placeholder
- reuse – (bool) whether or not to reuse parameters
- scope – (str) the scope name
Returns: ([tf.Tensor]) Mean, action and log probability
-
obs_ph¶ tf.Tensor: placeholder for observations, shape (self.n_batch, ) + self.ob_space.shape.
-
proba_step(obs, state=None, mask=None)¶ Returns the policy for a single step
Parameters: - obs – ([float] or [int]) The current observation of the environment
- state – ([float]) The last states (used in recurrent policies)
- mask – ([float]) The last masks (used in recurrent policies)
Returns: ([float]) actions
-
processed_obs¶ tf.Tensor: processed observations, shape (self.n_batch, ) + self.ob_space.shape.
The form of processing depends on the type of the observation space, and the parameters whether scale is passed to the constructor; see observation_input for more information.
-
step(obs, state=None, mask=None)¶ Returns the policy for a single step
Parameters: - obs – ([float] or [int]) The current observation of the environment
- state – ([float]) The last states (used in recurrent policies)
- mask – ([float]) The last masks (used in recurrent policies)
Returns: ([float]) actions
-
class
stable_baselines.td3.LnMlpPolicy(sess, ob_space, ac_space, n_env=1, n_steps=1, n_batch=None, reuse=False, **_kwargs)[source]¶ Policy object that implements actor critic, using a MLP (2 layers of 64), with layer normalisation
Parameters: - sess – (TensorFlow session) The current TensorFlow session
- ob_space – (Gym Space) The observation space of the environment
- ac_space – (Gym Space) The action space of the environment
- n_env – (int) The number of environments to run
- n_steps – (int) The number of steps to run for each environment
- n_batch – (int) The number of batch to run (n_envs * n_steps)
- reuse – (bool) If the policy is reusable or not
- _kwargs – (dict) Extra keyword arguments for the nature CNN feature extraction
-
action_ph¶ tf.Tensor: placeholder for actions, shape (self.n_batch, ) + self.ac_space.shape.
-
initial_state¶ The initial state of the policy. For feedforward policies, None. For a recurrent policy, a NumPy array of shape (self.n_env, ) + state_shape.
-
is_discrete¶ bool: is action space discrete.
-
make_actor(obs=None, reuse=False, scope='pi')¶ Creates an actor object
Parameters: - obs – (TensorFlow Tensor) The observation placeholder (can be None for default placeholder)
- reuse – (bool) whether or not to reuse parameters
- scope – (str) the scope name of the actor
Returns: (TensorFlow Tensor) the output tensor
-
make_critics(obs=None, action=None, reuse=False, scope='values_fn')¶ Creates the two Q-Values approximator
Parameters: - obs – (TensorFlow Tensor) The observation placeholder (can be None for default placeholder)
- action – (TensorFlow Tensor) The action placeholder
- reuse – (bool) whether or not to reuse parameters
- scope – (str) the scope name
Returns: ([tf.Tensor]) Mean, action and log probability
-
obs_ph¶ tf.Tensor: placeholder for observations, shape (self.n_batch, ) + self.ob_space.shape.
-
proba_step(obs, state=None, mask=None)¶ Returns the policy for a single step
Parameters: - obs – ([float] or [int]) The current observation of the environment
- state – ([float]) The last states (used in recurrent policies)
- mask – ([float]) The last masks (used in recurrent policies)
Returns: ([float]) actions
-
processed_obs¶ tf.Tensor: processed observations, shape (self.n_batch, ) + self.ob_space.shape.
The form of processing depends on the type of the observation space, and the parameters whether scale is passed to the constructor; see observation_input for more information.
-
step(obs, state=None, mask=None)¶ Returns the policy for a single step
Parameters: - obs – ([float] or [int]) The current observation of the environment
- state – ([float]) The last states (used in recurrent policies)
- mask – ([float]) The last masks (used in recurrent policies)
Returns: ([float]) actions
-
class
stable_baselines.td3.CnnPolicy(sess, ob_space, ac_space, n_env=1, n_steps=1, n_batch=None, reuse=False, **_kwargs)[source]¶ Policy object that implements actor critic, using a CNN (the nature CNN)
Parameters: - sess – (TensorFlow session) The current TensorFlow session
- ob_space – (Gym Space) The observation space of the environment
- ac_space – (Gym Space) The action space of the environment
- n_env – (int) The number of environments to run
- n_steps – (int) The number of steps to run for each environment
- n_batch – (int) The number of batch to run (n_envs * n_steps)
- reuse – (bool) If the policy is reusable or not
- _kwargs – (dict) Extra keyword arguments for the nature CNN feature extraction
-
action_ph¶ tf.Tensor: placeholder for actions, shape (self.n_batch, ) + self.ac_space.shape.
-
initial_state¶ The initial state of the policy. For feedforward policies, None. For a recurrent policy, a NumPy array of shape (self.n_env, ) + state_shape.
-
is_discrete¶ bool: is action space discrete.
-
make_actor(obs=None, reuse=False, scope='pi')¶ Creates an actor object
Parameters: - obs – (TensorFlow Tensor) The observation placeholder (can be None for default placeholder)
- reuse – (bool) whether or not to reuse parameters
- scope – (str) the scope name of the actor
Returns: (TensorFlow Tensor) the output tensor
-
make_critics(obs=None, action=None, reuse=False, scope='values_fn')¶ Creates the two Q-Values approximator
Parameters: - obs – (TensorFlow Tensor) The observation placeholder (can be None for default placeholder)
- action – (TensorFlow Tensor) The action placeholder
- reuse – (bool) whether or not to reuse parameters
- scope – (str) the scope name
Returns: ([tf.Tensor]) Mean, action and log probability
-
obs_ph¶ tf.Tensor: placeholder for observations, shape (self.n_batch, ) + self.ob_space.shape.
-
proba_step(obs, state=None, mask=None)¶ Returns the policy for a single step
Parameters: - obs – ([float] or [int]) The current observation of the environment
- state – ([float]) The last states (used in recurrent policies)
- mask – ([float]) The last masks (used in recurrent policies)
Returns: ([float]) actions
-
processed_obs¶ tf.Tensor: processed observations, shape (self.n_batch, ) + self.ob_space.shape.
The form of processing depends on the type of the observation space, and the parameters whether scale is passed to the constructor; see observation_input for more information.
-
step(obs, state=None, mask=None)¶ Returns the policy for a single step
Parameters: - obs – ([float] or [int]) The current observation of the environment
- state – ([float]) The last states (used in recurrent policies)
- mask – ([float]) The last masks (used in recurrent policies)
Returns: ([float]) actions
-
class
stable_baselines.td3.LnCnnPolicy(sess, ob_space, ac_space, n_env=1, n_steps=1, n_batch=None, reuse=False, **_kwargs)[source]¶ Policy object that implements actor critic, using a CNN (the nature CNN), with layer normalisation
Parameters: - sess – (TensorFlow session) The current TensorFlow session
- ob_space – (Gym Space) The observation space of the environment
- ac_space – (Gym Space) The action space of the environment
- n_env – (int) The number of environments to run
- n_steps – (int) The number of steps to run for each environment
- n_batch – (int) The number of batch to run (n_envs * n_steps)
- reuse – (bool) If the policy is reusable or not
- _kwargs – (dict) Extra keyword arguments for the nature CNN feature extraction
-
action_ph¶ tf.Tensor: placeholder for actions, shape (self.n_batch, ) + self.ac_space.shape.
-
initial_state¶ The initial state of the policy. For feedforward policies, None. For a recurrent policy, a NumPy array of shape (self.n_env, ) + state_shape.
-
is_discrete¶ bool: is action space discrete.
-
make_actor(obs=None, reuse=False, scope='pi')¶ Creates an actor object
Parameters: - obs – (TensorFlow Tensor) The observation placeholder (can be None for default placeholder)
- reuse – (bool) whether or not to reuse parameters
- scope – (str) the scope name of the actor
Returns: (TensorFlow Tensor) the output tensor
-
make_critics(obs=None, action=None, reuse=False, scope='values_fn')¶ Creates the two Q-Values approximator
Parameters: - obs – (TensorFlow Tensor) The observation placeholder (can be None for default placeholder)
- action – (TensorFlow Tensor) The action placeholder
- reuse – (bool) whether or not to reuse parameters
- scope – (str) the scope name
Returns: ([tf.Tensor]) Mean, action and log probability
-
obs_ph¶ tf.Tensor: placeholder for observations, shape (self.n_batch, ) + self.ob_space.shape.
-
proba_step(obs, state=None, mask=None)¶ Returns the policy for a single step
Parameters: - obs – ([float] or [int]) The current observation of the environment
- state – ([float]) The last states (used in recurrent policies)
- mask – ([float]) The last masks (used in recurrent policies)
Returns: ([float]) actions
-
processed_obs¶ tf.Tensor: processed observations, shape (self.n_batch, ) + self.ob_space.shape.
The form of processing depends on the type of the observation space, and the parameters whether scale is passed to the constructor; see observation_input for more information.
-
step(obs, state=None, mask=None)¶ Returns the policy for a single step
Parameters: - obs – ([float] or [int]) The current observation of the environment
- state – ([float]) The last states (used in recurrent policies)
- mask – ([float]) The last masks (used in recurrent policies)
Returns: ([float]) actions
Custom Policy Network¶
Similarly to the example given in the examples page. You can easily define a custom architecture for the policy network:
import gym
import numpy as np
from stable_baselines import TD3
from stable_baselines.td3.policies import FeedForwardPolicy
from stable_baselines.common.vec_env import DummyVecEnv
from stable_baselines.ddpg.noise import NormalActionNoise, OrnsteinUhlenbeckActionNoise
# Custom MLP policy with two layers
class CustomTD3Policy(FeedForwardPolicy):
def __init__(self, *args, **kwargs):
super(CustomTD3Policy, self).__init__(*args, **kwargs,
layers=[400, 300],
layer_norm=False,
feature_extraction="mlp")
# Create and wrap the environment
env = gym.make('Pendulum-v0')
env = DummyVecEnv([lambda: env])
# The noise objects for TD3
n_actions = env.action_space.shape[-1]
action_noise = NormalActionNoise(mean=np.zeros(n_actions), sigma=0.1 * np.ones(n_actions))
model = TD3(CustomTD3Policy, env, action_noise=action_noise, verbose=1)
# Train the agent
model.learn(total_timesteps=80000)
TRPO¶
Trust Region Policy Optimization (TRPO) is an iterative approach for optimizing policies with guaranteed monotonic improvement.
Note
TRPO requires OpenMPI. If OpenMPI isn’t enabled, then TRPO isn’t
imported into the stable_baselines module.
Notes¶
- Original paper: https://arxiv.org/abs/1502.05477
- OpenAI blog post: https://blog.openai.com/openai-baselines-ppo/
mpirun -np 16 python -m stable_baselines.trpo_mpi.run_atariruns the algorithm for 40M frames = 10M timesteps on an Atari game. See help (-h) for more options.python -m stable_baselines.trpo_mpi.run_mujocoruns the algorithm for 1M timesteps on a Mujoco environment.
Can I use?¶
- Recurrent policies: ❌
- Multi processing: ✔️ (using MPI)
- Gym spaces:
| Space | Action | Observation |
|---|---|---|
| Discrete | ✔️ | ✔️ |
| Box | ✔️ | ✔️ |
| MultiDiscrete | ✔️ | ✔️ |
| MultiBinary | ✔️ | ✔️ |
Example¶
import gym
from stable_baselines.common.policies import MlpPolicy
from stable_baselines import TRPO
env = gym.make('CartPole-v1')
model = TRPO(MlpPolicy, env, verbose=1)
model.learn(total_timesteps=25000)
model.save("trpo_cartpole")
del model # remove to demonstrate saving and loading
model = TRPO.load("trpo_cartpole")
obs = env.reset()
while True:
action, _states = model.predict(obs)
obs, rewards, dones, info = env.step(action)
env.render()
Parameters¶
-
class
stable_baselines.trpo_mpi.TRPO(policy, env, gamma=0.99, timesteps_per_batch=1024, max_kl=0.01, cg_iters=10, lam=0.98, entcoeff=0.0, cg_damping=0.01, vf_stepsize=0.0003, vf_iters=3, verbose=0, tensorboard_log=None, _init_setup_model=True, policy_kwargs=None, full_tensorboard_log=False, seed=None, n_cpu_tf_sess=1)[source]¶ Trust Region Policy Optimization (https://arxiv.org/abs/1502.05477)
Parameters: - policy – (ActorCriticPolicy or str) The policy model to use (MlpPolicy, CnnPolicy, CnnLstmPolicy, …)
- env – (Gym environment or str) The environment to learn from (if registered in Gym, can be str)
- gamma – (float) the discount value
- timesteps_per_batch – (int) the number of timesteps to run per batch (horizon)
- max_kl – (float) the Kullback-Leibler loss threshold
- cg_iters – (int) the number of iterations for the conjugate gradient calculation
- lam – (float) GAE factor
- entcoeff – (float) the weight for the entropy loss
- cg_damping – (float) the compute gradient dampening factor
- vf_stepsize – (float) the value function stepsize
- vf_iters – (int) the value function’s number iterations for learning
- verbose – (int) the verbosity level: 0 none, 1 training information, 2 tensorflow debug
- tensorboard_log – (str) the log location for tensorboard (if None, no logging)
- _init_setup_model – (bool) Whether or not to build the network at the creation of the instance
- policy_kwargs – (dict) additional arguments to be passed to the policy on creation
- full_tensorboard_log – (bool) enable additional logging when using tensorboard WARNING: this logging can take a lot of space quickly
- seed – (int) Seed for the pseudo-random generators (python, numpy, tensorflow). If None (default), use random seed. Note that if you want completely deterministic results, you must set n_cpu_tf_sess to 1.
- n_cpu_tf_sess – (int) The number of threads for TensorFlow operations If None, the number of cpu of the current machine will be used.
-
action_probability(observation, state=None, mask=None, actions=None, logp=False)¶ If
actionsisNone, then get the model’s action probability distribution from a given observation.- Depending on the action space the output is:
- Discrete: probability for each possible action
- Box: mean and standard deviation of the action output
However if
actionsis notNone, this function will return the probability that the given actions are taken with the given parameters (observation, state, …) on this model. For discrete action spaces, it returns the probability mass; for continuous action spaces, the probability density. This is since the probability mass will always be zero in continuous spaces, see http://blog.christianperone.com/2019/01/ for a good explanationParameters: - observation – (np.ndarray) the input observation
- state – (np.ndarray) The last states (can be None, used in recurrent policies)
- mask – (np.ndarray) The last masks (can be None, used in recurrent policies)
- actions – (np.ndarray) (OPTIONAL) For calculating the likelihood that the given actions are chosen by the model for each of the given parameters. Must have the same number of actions and observations. (set to None to return the complete action probability distribution)
- logp – (bool) (OPTIONAL) When specified with actions, returns probability in log-space. This has no effect if actions is None.
Returns: (np.ndarray) the model’s (log) action probability
-
get_env()¶ returns the current environment (can be None if not defined)
Returns: (Gym Environment) The current environment
-
get_parameter_list()¶ Get tensorflow Variables of model’s parameters
This includes all variables necessary for continuing training (saving / loading).
Returns: (list) List of tensorflow Variables
-
get_parameters()¶ Get current model parameters as dictionary of variable name -> ndarray.
Returns: (OrderedDict) Dictionary of variable name -> ndarray of model’s parameters.
-
get_vec_normalize_env() → Optional[stable_baselines.common.vec_env.vec_normalize.VecNormalize]¶ Return the
VecNormalizewrapper of the training env if it exists.Returns: Optional[VecNormalize] The VecNormalizeenv.
-
learn(total_timesteps, callback=None, log_interval=100, tb_log_name='TRPO', reset_num_timesteps=True)[source]¶ Return a trained model.
Parameters: - total_timesteps – (int) The total number of samples to train on
- callback – (Union[callable, [callable], BaseCallback]) function called at every steps with state of the algorithm. It takes the local and global variables. If it returns False, training is aborted. When the callback inherits from BaseCallback, you will have access to additional stages of the training (training start/end), please read the documentation for more details.
- log_interval – (int) The number of timesteps before logging.
- tb_log_name – (str) the name of the run for tensorboard log
- reset_num_timesteps – (bool) whether or not to reset the current timestep number (used in logging)
Returns: (BaseRLModel) the trained model
-
classmethod
load(load_path, env=None, custom_objects=None, **kwargs)¶ Load the model from file
Parameters: - load_path – (str or file-like) the saved parameter location
- env – (Gym Environment) the new environment to run the loaded model on (can be None if you only need prediction from a trained model)
- custom_objects – (dict) Dictionary of objects to replace upon loading. If a variable is present in this dictionary as a key, it will not be deserialized and the corresponding item will be used instead. Similar to custom_objects in keras.models.load_model. Useful when you have an object in file that can not be deserialized.
- kwargs – extra arguments to change the model when loading
-
load_parameters(load_path_or_dict, exact_match=True)¶ Load model parameters from a file or a dictionary
Dictionary keys should be tensorflow variable names, which can be obtained with
get_parametersfunction. Ifexact_matchis True, dictionary should contain keys for all model’s parameters, otherwise RunTimeError is raised. If False, only variables included in the dictionary will be updated.This does not load agent’s hyper-parameters.
Warning
This function does not update trainer/optimizer variables (e.g. momentum). As such training after using this function may lead to less-than-optimal results.
Parameters: - load_path_or_dict – (str or file-like or dict) Save parameter location or dict of parameters as variable.name -> ndarrays to be loaded.
- exact_match – (bool) If True, expects load dictionary to contain keys for all variables in the model. If False, loads parameters only for variables mentioned in the dictionary. Defaults to True.
-
predict(observation, state=None, mask=None, deterministic=False)¶ Get the model’s action from an observation
Parameters: - observation – (np.ndarray) the input observation
- state – (np.ndarray) The last states (can be None, used in recurrent policies)
- mask – (np.ndarray) The last masks (can be None, used in recurrent policies)
- deterministic – (bool) Whether or not to return deterministic actions.
Returns: (np.ndarray, np.ndarray) the model’s action and the next state (used in recurrent policies)
-
pretrain(dataset, n_epochs=10, learning_rate=0.0001, adam_epsilon=1e-08, val_interval=None)¶ Pretrain a model using behavior cloning: supervised learning given an expert dataset.
NOTE: only Box and Discrete spaces are supported for now.
Parameters: - dataset – (ExpertDataset) Dataset manager
- n_epochs – (int) Number of iterations on the training set
- learning_rate – (float) Learning rate
- adam_epsilon – (float) the epsilon value for the adam optimizer
- val_interval – (int) Report training and validation losses every n epochs. By default, every 10th of the maximum number of epochs.
Returns: (BaseRLModel) the pretrained model
-
save(save_path, cloudpickle=False)[source]¶ Save the current parameters to file
Parameters: - save_path – (str or file-like) The save location
- cloudpickle – (bool) Use older cloudpickle format instead of zip-archives.
-
set_env(env)¶ Checks the validity of the environment, and if it is coherent, set it as the current environment.
Parameters: env – (Gym Environment) The environment for learning a policy
-
set_random_seed(seed: Optional[int]) → None¶ Parameters: seed – (Optional[int]) Seed for the pseudo-random generators. If None, do not change the seeds.
Probability Distributions¶
Probability distributions used for the different action spaces:
CategoricalProbabilityDistribution-> DiscreteDiagGaussianProbabilityDistribution-> Box (continuous actions)MultiCategoricalProbabilityDistribution-> MultiDiscreteBernoulliProbabilityDistribution-> MultiBinary
The policy networks output parameters for the distributions (named flat in the methods).
Actions are then sampled from those distributions.
For instance, in the case of discrete actions. The policy network outputs probability
of taking each action. The CategoricalProbabilityDistribution allows to sample from it,
computes the entropy, the negative log probability (neglogp) and backpropagate the gradient.
In the case of continuous actions, a Gaussian distribution is used. The policy network outputs
mean and (log) std of the distribution (assumed to be a DiagGaussianProbabilityDistribution).
-
class
stable_baselines.common.distributions.BernoulliProbabilityDistribution(logits)[source]¶ -
-
classmethod
fromflat(flat)[source]¶ Create an instance of this from new Bernoulli input
Parameters: flat – ([float]) the Bernoulli input data Returns: (ProbabilityDistribution) the instance from the given Bernoulli input data
-
kl(other)[source]¶ Calculates the Kullback-Leibler divergence from the given probability distribution
Parameters: other – ([float]) the distribution to compare with Returns: (float) the KL divergence of the two distributions
-
classmethod
-
class
stable_baselines.common.distributions.BernoulliProbabilityDistributionType(size)[source]¶ -
-
proba_distribution_from_latent(pi_latent_vector, vf_latent_vector, init_scale=1.0, init_bias=0.0)[source]¶ returns the probability distribution from latent values
Parameters: - pi_latent_vector – ([float]) the latent pi values
- vf_latent_vector – ([float]) the latent vf values
- init_scale – (float) the initial scale of the distribution
- init_bias – (float) the initial bias of the distribution
Returns: (ProbabilityDistribution) the instance of the ProbabilityDistribution associated
-
-
class
stable_baselines.common.distributions.CategoricalProbabilityDistribution(logits)[source]¶ -
-
classmethod
fromflat(flat)[source]¶ Create an instance of this from new logits values
Parameters: flat – ([float]) the categorical logits input Returns: (ProbabilityDistribution) the instance from the given categorical input
-
kl(other)[source]¶ Calculates the Kullback-Leibler divergence from the given probability distribution
Parameters: other – ([float]) the distribution to compare with Returns: (float) the KL divergence of the two distributions
-
classmethod
-
class
stable_baselines.common.distributions.CategoricalProbabilityDistributionType(n_cat)[source]¶ -
-
proba_distribution_from_latent(pi_latent_vector, vf_latent_vector, init_scale=1.0, init_bias=0.0)[source]¶ returns the probability distribution from latent values
Parameters: - pi_latent_vector – ([float]) the latent pi values
- vf_latent_vector – ([float]) the latent vf values
- init_scale – (float) the initial scale of the distribution
- init_bias – (float) the initial bias of the distribution
Returns: (ProbabilityDistribution) the instance of the ProbabilityDistribution associated
-
-
class
stable_baselines.common.distributions.DiagGaussianProbabilityDistribution(flat)[source]¶ -
-
classmethod
fromflat(flat)[source]¶ Create an instance of this from new multivariate Gaussian input
Parameters: flat – ([float]) the multivariate Gaussian input data Returns: (ProbabilityDistribution) the instance from the given multivariate Gaussian input data
-
kl(other)[source]¶ Calculates the Kullback-Leibler divergence from the given probability distribution
Parameters: other – ([float]) the distribution to compare with Returns: (float) the KL divergence of the two distributions
-
classmethod
-
class
stable_baselines.common.distributions.DiagGaussianProbabilityDistributionType(size)[source]¶ -
-
proba_distribution_from_flat(flat)[source]¶ returns the probability distribution from flat probabilities
Parameters: flat – ([float]) the flat probabilities Returns: (ProbabilityDistribution) the instance of the ProbabilityDistribution associated
-
proba_distribution_from_latent(pi_latent_vector, vf_latent_vector, init_scale=1.0, init_bias=0.0)[source]¶ returns the probability distribution from latent values
Parameters: - pi_latent_vector – ([float]) the latent pi values
- vf_latent_vector – ([float]) the latent vf values
- init_scale – (float) the initial scale of the distribution
- init_bias – (float) the initial bias of the distribution
Returns: (ProbabilityDistribution) the instance of the ProbabilityDistribution associated
-
-
class
stable_baselines.common.distributions.MultiCategoricalProbabilityDistribution(nvec, flat)[source]¶ -
-
classmethod
fromflat(flat)[source]¶ Create an instance of this from new logits values
Parameters: flat – ([float]) the multi categorical logits input Returns: (ProbabilityDistribution) the instance from the given multi categorical input
-
kl(other)[source]¶ Calculates the Kullback-Leibler divergence from the given probability distribution
Parameters: other – ([float]) the distribution to compare with Returns: (float) the KL divergence of the two distributions
-
classmethod
-
class
stable_baselines.common.distributions.MultiCategoricalProbabilityDistributionType(n_vec)[source]¶ -
-
proba_distribution_from_flat(flat)[source]¶ Returns the probability distribution from flat probabilities flat: flattened vector of parameters of probability distribution
Parameters: flat – ([float]) the flat probabilities Returns: (ProbabilityDistribution) the instance of the ProbabilityDistribution associated
-
proba_distribution_from_latent(pi_latent_vector, vf_latent_vector, init_scale=1.0, init_bias=0.0)[source]¶ returns the probability distribution from latent values
Parameters: - pi_latent_vector – ([float]) the latent pi values
- vf_latent_vector – ([float]) the latent vf values
- init_scale – (float) the initial scale of the distribution
- init_bias – (float) the initial bias of the distribution
Returns: (ProbabilityDistribution) the instance of the ProbabilityDistribution associated
-
-
class
stable_baselines.common.distributions.ProbabilityDistribution[source]¶ Base class for describing a probability distribution.
-
kl(other)[source]¶ Calculates the Kullback-Leibler divergence from the given probability distribution
Parameters: other – ([float]) the distribution to compare with Returns: (float) the KL divergence of the two distributions
-
logp(x)[source]¶ returns the of the log likelihood
Parameters: x – (str) the labels of each index Returns: ([float]) The log likelihood of the distribution
-
-
class
stable_baselines.common.distributions.ProbabilityDistributionType[source]¶ Parametrized family of probability distributions
-
param_placeholder(prepend_shape, name=None)[source]¶ returns the TensorFlow placeholder for the input parameters
Parameters: - prepend_shape – ([int]) the prepend shape
- name – (str) the placeholder name
Returns: (TensorFlow Tensor) the placeholder
-
proba_distribution_from_flat(flat)[source]¶ Returns the probability distribution from flat probabilities flat: flattened vector of parameters of probability distribution
Parameters: flat – ([float]) the flat probabilities Returns: (ProbabilityDistribution) the instance of the ProbabilityDistribution associated
-
proba_distribution_from_latent(pi_latent_vector, vf_latent_vector, init_scale=1.0, init_bias=0.0)[source]¶ returns the probability distribution from latent values
Parameters: - pi_latent_vector – ([float]) the latent pi values
- vf_latent_vector – ([float]) the latent vf values
- init_scale – (float) the initial scale of the distribution
- init_bias – (float) the initial bias of the distribution
Returns: (ProbabilityDistribution) the instance of the ProbabilityDistribution associated
-
probability_distribution_class()[source]¶ returns the ProbabilityDistribution class of this type
Returns: (Type ProbabilityDistribution) the probability distribution class associated
-
-
stable_baselines.common.distributions.make_proba_dist_type(ac_space)[source]¶ return an instance of ProbabilityDistributionType for the correct type of action space
Parameters: ac_space – (Gym Space) the input action space Returns: (ProbabilityDistributionType) the appropriate instance of a ProbabilityDistributionType
Tensorflow Utils¶
-
stable_baselines.common.tf_util.avg_norm(tensor)[source]¶ Return an average of the L2 normalization of the batch
Parameters: tensor – (TensorFlow Tensor) The input tensor Returns: (TensorFlow Tensor) Average L2 normalization of the batch
-
stable_baselines.common.tf_util.batch_to_seq(tensor_batch, n_batch, n_steps, flat=False)[source]¶ Transform a batch of Tensors, into a sequence of Tensors for recurrent policies
Parameters: - tensor_batch – (TensorFlow Tensor) The input tensor to unroll
- n_batch – (int) The number of batch to run (n_envs * n_steps)
- n_steps – (int) The number of steps to run for each environment
- flat – (bool) If the input Tensor is flat
Returns: (TensorFlow Tensor) sequence of Tensors for recurrent policies
-
stable_baselines.common.tf_util.calc_entropy(logits)[source]¶ Calculates the entropy of the output values of the network
Parameters: logits – (TensorFlow Tensor) The input probability for each action Returns: (TensorFlow Tensor) The Entropy of the output values of the network
-
stable_baselines.common.tf_util.check_shape(tensors, shapes)[source]¶ Verifies the tensors match the given shape, will raise an error if the shapes do not match
Parameters: - tensors – ([TensorFlow Tensor]) The tensors that should be checked
- shapes – ([list]) The list of shapes for each tensor
-
stable_baselines.common.tf_util.flatgrad(loss, var_list, clip_norm=None)[source]¶ calculates the gradient and flattens it
Parameters: - loss – (float) the loss value
- var_list – ([TensorFlow Tensor]) the variables
- clip_norm – (float) clip the gradients (disabled if None)
Returns: ([TensorFlow Tensor]) flattened gradient
-
stable_baselines.common.tf_util.function(inputs, outputs, updates=None, givens=None)[source]¶ Take a bunch of tensorflow placeholders and expressions computed based on those placeholders and produces f(inputs) -> outputs. Function f takes values to be fed to the input’s placeholders and produces the values of the expressions in outputs. Just like a Theano function.
Input values can be passed in the same order as inputs or can be provided as kwargs based on placeholder name (passed to constructor or accessible via placeholder.op.name).
- Example:
>>> x = tf.placeholder(tf.int32, (), name="x") >>> y = tf.placeholder(tf.int32, (), name="y") >>> z = 3 * x + 2 * y >>> lin = function([x, y], z, givens={y: 0}) >>> with single_threaded_session(): >>> initialize() >>> assert lin(2) == 6 >>> assert lin(x=3) == 9 >>> assert lin(2, 2) == 10
Parameters: - inputs – (TensorFlow Tensor or Object with make_feed_dict) list of input arguments
- outputs – (TensorFlow Tensor) list of outputs or a single output to be returned from function. Returned value will also have the same shape.
- updates – ([tf.Operation] or tf.Operation) list of update functions or single update function that will be run whenever the function is called. The return is ignored.
- givens – (dict) the values known for the output
-
stable_baselines.common.tf_util.get_globals_vars(name)[source]¶ returns the trainable variables
Parameters: name – (str) the scope Returns: ([TensorFlow Variable])
-
stable_baselines.common.tf_util.get_trainable_vars(name)[source]¶ returns the trainable variables
Parameters: name – (str) the scope Returns: ([TensorFlow Variable])
-
stable_baselines.common.tf_util.gradient_add(grad_1, grad_2, param, verbose=0)[source]¶ Sum two gradients
Parameters: - grad_1 – (TensorFlow Tensor) The first gradient
- grad_2 – (TensorFlow Tensor) The second gradient
- param – (TensorFlow parameters) The trainable parameters
- verbose – (int) verbosity level
Returns: (TensorFlow Tensor) the sum of the gradients
-
stable_baselines.common.tf_util.huber_loss(tensor, delta=1.0)[source]¶ Reference: https://en.wikipedia.org/wiki/Huber_loss
Parameters: - tensor – (TensorFlow Tensor) the input value
- delta – (float) Huber loss delta value
Returns: (TensorFlow Tensor) Huber loss output
-
stable_baselines.common.tf_util.in_session(func)[source]¶ Wraps a function so that it is in a TensorFlow Session
Parameters: func – (function) the function to wrap Returns: (function)
-
stable_baselines.common.tf_util.initialize(sess=None)[source]¶ Initialize all the uninitialized variables in the global scope.
Parameters: sess – (TensorFlow Session)
-
stable_baselines.common.tf_util.intprod(tensor)[source]¶ calculates the product of all the elements in a list
Parameters: tensor – ([Number]) the list of elements Returns: (int) the product truncated
-
stable_baselines.common.tf_util.is_image(tensor)[source]¶ Check if a tensor has the shape of a valid image for tensorboard logging. Valid image: RGB, RGBD, GrayScale
Parameters: tensor – (np.ndarray or tf.placeholder) Returns: (bool)
-
stable_baselines.common.tf_util.make_session(num_cpu=None, make_default=False, graph=None)[source]¶ Returns a session that will use <num_cpu> CPU’s only
Parameters: - num_cpu – (int) number of CPUs to use for TensorFlow
- make_default – (bool) if this should return an InteractiveSession or a normal Session
- graph – (TensorFlow Graph) the graph of the session
Returns: (TensorFlow session)
-
stable_baselines.common.tf_util.mse(pred, target)[source]¶ Returns the Mean squared error between prediction and target
Parameters: - pred – (TensorFlow Tensor) The predicted value
- target – (TensorFlow Tensor) The target value
Returns: (TensorFlow Tensor) The Mean squared error between prediction and target
-
stable_baselines.common.tf_util.numel(tensor)[source]¶ get TensorFlow Tensor’s number of elements
Parameters: tensor – (TensorFlow Tensor) the input tensor Returns: (int) the number of elements
-
stable_baselines.common.tf_util.outer_scope_getter(scope, new_scope='')[source]¶ remove a scope layer for the getter
Parameters: - scope – (str) the layer to remove
- new_scope – (str) optional replacement name
Returns: (function (function, str,
*args,**kwargs): Tensorflow Tensor)
-
stable_baselines.common.tf_util.q_explained_variance(q_pred, q_true)[source]¶ Calculates the explained variance of the Q value
Parameters: - q_pred – (TensorFlow Tensor) The predicted Q value
- q_true – (TensorFlow Tensor) The expected Q value
Returns: (TensorFlow Tensor) the explained variance of the Q value
-
stable_baselines.common.tf_util.sample(logits)[source]¶ Creates a sampling Tensor for non deterministic policies when using categorical distribution. It uses the Gumbel-max trick: http://amid.fish/humble-gumbel
Parameters: logits – (TensorFlow Tensor) The input probability for each action Returns: (TensorFlow Tensor) The sampled action
-
stable_baselines.common.tf_util.seq_to_batch(tensor_sequence, flat=False)[source]¶ Transform a sequence of Tensors, into a batch of Tensors for recurrent policies
Parameters: - tensor_sequence – (TensorFlow Tensor) The input tensor to batch
- flat – (bool) If the input Tensor is flat
Returns: (TensorFlow Tensor) batch of Tensors for recurrent policies
-
stable_baselines.common.tf_util.single_threaded_session(make_default=False, graph=None)[source]¶ Returns a session which will only use a single CPU
Parameters: - make_default – (bool) if this should return an InteractiveSession or a normal Session
- graph – (TensorFlow Graph) the graph of the session
Returns: (TensorFlow session)
-
stable_baselines.common.tf_util.total_episode_reward_logger(rew_acc, rewards, masks, writer, steps)[source]¶ calculates the cumulated episode reward, and prints to tensorflow log the output
Parameters: - rew_acc – (np.array float) the total running reward
- rewards – (np.array float) the rewards
- masks – (np.array bool) the end of episodes
- writer – (TensorFlow Session.writer) the writer to log to
- steps – (int) the current timestep
Returns: (np.array float) the updated total running reward
Returns: (np.array float) the updated total running reward
Command Utils¶
Helpers for scripts like run_atari.py.
-
stable_baselines.common.cmd_util.arg_parser()[source]¶ Create an empty argparse.ArgumentParser.
Returns: (ArgumentParser)
-
stable_baselines.common.cmd_util.atari_arg_parser()[source]¶ Create an argparse.ArgumentParser for run_atari.py.
Returns: (ArgumentParser) parser {‘–env’: ‘BreakoutNoFrameskip-v4’, ‘–seed’: 0, ‘–num-timesteps’: int(1e7)}
-
stable_baselines.common.cmd_util.make_atari_env(env_id, num_env, seed, wrapper_kwargs=None, start_index=0, allow_early_resets=True, start_method=None, use_subprocess=False)[source]¶ Create a wrapped, monitored VecEnv for Atari.
Parameters: - env_id – (str) the environment ID
- num_env – (int) the number of environment you wish to have in subprocesses
- seed – (int) the initial seed for RNG
- wrapper_kwargs – (dict) the parameters for wrap_deepmind function
- start_index – (int) start rank index
- allow_early_resets – (bool) allows early reset of the environment
- start_method – (str) method used to start the subprocesses. See SubprocVecEnv doc for more information
- use_subprocess – (bool) Whether to use SubprocVecEnv or DummyVecEnv when num_env > 1, DummyVecEnv is usually faster. Default: False
Returns: (VecEnv) The atari environment
-
stable_baselines.common.cmd_util.make_mujoco_env(env_id, seed, allow_early_resets=True)[source]¶ Create a wrapped, monitored gym.Env for MuJoCo.
Parameters: - env_id – (str) the environment ID
- seed – (int) the initial seed for RNG
- allow_early_resets – (bool) allows early reset of the environment
Returns: (Gym Environment) The mujoco environment
-
stable_baselines.common.cmd_util.make_robotics_env(env_id, seed, rank=0, allow_early_resets=True)[source]¶ Create a wrapped, monitored gym.Env for MuJoCo.
Parameters: - env_id – (str) the environment ID
- seed – (int) the initial seed for RNG
- rank – (int) the rank of the environment (for logging)
- allow_early_resets – (bool) allows early reset of the environment
Returns: (Gym Environment) The robotic environment
-
stable_baselines.common.cmd_util.make_vec_env(env_id, n_envs=1, seed=None, start_index=0, monitor_dir=None, wrapper_class=None, env_kwargs=None, vec_env_cls=None, vec_env_kwargs=None)[source]¶ Create a wrapped, monitored VecEnv. By default it uses a DummyVecEnv which is usually faster than a SubprocVecEnv.
Parameters: - env_id – (str or Type[gym.Env]) the environment ID or the environment class
- n_envs – (int) the number of environments you wish to have in parallel
- seed – (int) the initial seed for the random number generator
- start_index – (int) start rank index
- monitor_dir – (str) Path to a folder where the monitor files will be saved. If None, no file will be written, however, the env will still be wrapped in a Monitor wrapper to provide additional information about training.
- wrapper_class – (gym.Wrapper or callable) Additional wrapper to use on the environment. This can also be a function with single argument that wraps the environment in many things.
- env_kwargs – (dict) Optional keyword argument to pass to the env constructor
- vec_env_cls – (Type[VecEnv]) A custom VecEnv class constructor. Default: None.
- vec_env_kwargs – (dict) Keyword arguments to pass to the VecEnv class constructor.
Returns: (VecEnv) The wrapped environment
Schedules¶
Schedules are used as hyperparameter for most of the algorithms, in order to change value of a parameter over time (usually the learning rate).
This file is used for specifying various schedules that evolve over time throughout the execution of the algorithm, such as:
- learning rate for the optimizer
- exploration epsilon for the epsilon greedy exploration strategy
- beta parameter for beta parameter in prioritized replay
Each schedule has a function value(t) which returns the current value of the parameter given the timestep t of the optimization procedure.
-
class
stable_baselines.common.schedules.ConstantSchedule(value)[source]¶ Value remains constant over time.
Parameters: value – (float) Constant value of the schedule
-
class
stable_baselines.common.schedules.LinearSchedule(schedule_timesteps, final_p, initial_p=1.0)[source]¶ Linear interpolation between initial_p and final_p over schedule_timesteps. After this many timesteps pass final_p is returned.
Parameters: - schedule_timesteps – (int) Number of timesteps for which to linearly anneal initial_p to final_p
- initial_p – (float) initial output value
- final_p – (float) final output value
-
class
stable_baselines.common.schedules.PiecewiseSchedule(endpoints, interpolation=<function linear_interpolation>, outside_value=None)[source]¶ Piecewise schedule.
Parameters: - endpoints – ([(int, int)]) list of pairs (time, value) meaning that schedule should output value when t==time. All the values for time must be sorted in an increasing order. When t is between two times, e.g. (time_a, value_a) and (time_b, value_b), such that time_a <= t < time_b then value outputs interpolation(value_a, value_b, alpha) where alpha is a fraction of time passed between time_a and time_b for time t.
- interpolation – (lambda (float, float, float): float) a function that takes value to the left and to the right of t according to the endpoints. Alpha is the fraction of distance from left endpoint to right endpoint that t has covered. See linear_interpolation for example.
- outside_value – (float) if the value is requested outside of all the intervals specified in endpoints this value is returned. If None then AssertionError is raised when outside value is requested.
-
stable_baselines.common.schedules.constant(_)[source]¶ Returns a constant value for the Scheduler
Parameters: _ – ignored Returns: (float) 1
-
stable_baselines.common.schedules.constfn(val)[source]¶ Create a function that returns a constant It is useful for learning rate schedule (to avoid code duplication)
Parameters: val – (float) Returns: (function)
-
stable_baselines.common.schedules.double_linear_con(progress)[source]¶ Returns a linear value (x2) with a flattened tail for the Scheduler
Parameters: progress – (float) Current progress status (in [0, 1]) Returns: (float) 1 - progress*2 if (1 - progress*2) >= 0.125 else 0.125
-
stable_baselines.common.schedules.double_middle_drop(progress)[source]¶ Returns a linear value with two drops near the middle to a constant value for the Scheduler
Parameters: progress – (float) Current progress status (in [0, 1]) Returns: (float) if 0.75 <= 1 - p: 1 - p, if 0.25 <= 1 - p < 0.75: 0.75, if 1 - p < 0.25: 0.125
-
stable_baselines.common.schedules.get_schedule_fn(value_schedule)[source]¶ Transform (if needed) learning rate and clip range to callable.
Parameters: value_schedule – (callable or float) Returns: (function)
-
stable_baselines.common.schedules.linear_interpolation(left, right, alpha)[source]¶ Linear interpolation between left and right.
Parameters: - left – (float) left boundary
- right – (float) right boundary
- alpha – (float) coeff in [0, 1]
Returns: (float)
Evaluation Helper¶
-
stable_baselines.common.evaluation.evaluate_policy(model, env, n_eval_episodes=10, deterministic=True, render=False, callback=None, reward_threshold=None, return_episode_rewards=False)[source]¶ Runs policy for n_eval_episodes episodes and returns average reward. This is made to work only with one env.
Parameters: - model – (BaseRLModel) The RL agent you want to evaluate.
- env – (gym.Env or VecEnv) The gym environment. In the case of a VecEnv this must contain only one environment.
- n_eval_episodes – (int) Number of episode to evaluate the agent
- deterministic – (bool) Whether to use deterministic or stochastic actions
- render – (bool) Whether to render the environment or not
- callback – (callable) callback function to do additional checks, called after each step.
- reward_threshold – (float) Minimum expected reward per episode, this will raise an error if the performance is not met
- return_episode_rewards – (bool) If True, a list of reward per episode will be returned instead of the mean.
Returns: (float, float) Mean reward per episode, std of reward per episode returns ([float], [int]) when return_episode_rewards is True
Gym Environment Checker¶
-
stable_baselines.common.env_checker.check_env(env: <MagicMock id='140553655973984'>, warn: bool = True, skip_render_check: bool = True) → None[source]¶ Check that an environment follows Gym API. This is particularly useful when using a custom environment. Please take a look at https://github.com/openai/gym/blob/master/gym/core.py for more information about the API.
It also optionally check that the environment is compatible with Stable-Baselines.
Parameters: - env – (gym.Env) The Gym environment that will be checked
- warn – (bool) Whether to output additional warnings mainly related to the interaction with Stable Baselines
- skip_render_check – (bool) Whether to skip the checks for the render method. True by default (useful for the CI)
Changelog¶
For download links, please look at Github release page.
Release 2.10.0 (2020-03-11)¶
Callback collection, cleanup and bug fixes
Breaking Changes:¶
evaluate_policynow returns the standard deviation of the reward per episode as second return value (instead ofn_steps)evaluate_policynow returns as second return value a list of the episode lengths whenreturn_episode_rewardsis set toTrue(instead ofn_steps)Callback are now called after each
env.step()for consistency (it was called everyn_stepsbefore in algorithm likeA2CorPPO2)Removed unused code in
common/a2c/utils.py(calc_entropy_softmax,make_path)Refactoring, including removed files and moving functions.
Algorithms no longer import from each other, and
commondoes not import from algorithms.a2c/utils.pyremoved and split into other files:- common/tf_util.py:
sample,calc_entropy,mse,avg_norm,total_episode_reward_logger,q_explained_variance,gradient_add,avg_norm,check_shape,seq_to_batch,batch_to_seq. - common/tf_layers.py:
conv,linear,lstm,_ln,lnlstm,conv_to_fc,ortho_init. - a2c/a2c.py:
discount_with_dones. - acer/acer_simple.py:
get_by_index,EpisodeStats. - common/schedules.py:
constant,linear_schedule,middle_drop,double_linear_con,double_middle_drop,SCHEDULES,Scheduler.
- common/tf_util.py:
trpo_mpi/utils.pyfunctions moved (traj_segment_generatormoved tocommon/runners.py,flatten_liststocommon/misc_util.py).ppo2/ppo2.pyfunctions moved (safe_meantocommon/math_util.py,constfnandget_schedule_fntocommon/schedules.py).sac/policies.pyfunctionmlpmoved tocommon/tf_layers.py.sac/sac.pyfunctionget_varsremoved (replaced withtf.util.get_trainable_vars).deepq/replay_buffer.pyrenamed tocommon/buffers.py.
New Features:¶
- Parallelized updating and sampling from the replay buffer in DQN. (@flodorner)
- Docker build script, scripts/build_docker.sh, can push images automatically.
- Added callback collection
- Added
unwrap_vec_normalizeandsync_envs_normalizationin thevec_envmodule to synchronize two VecNormalize environment - Added a seeding method for vectorized environments. (@NeoExtended)
- Added extend method to store batches of experience in ReplayBuffer. (@solliet)
Bug Fixes:¶
- Fixed Docker images via
scripts/build_docker.shandDockerfile: GPU image now containstensorflow-gpu, and both images havestable_baselinesinstalled in developer mode at correct directory for mounting. - Fixed Docker GPU run script,
scripts/run_docker_gpu.sh, to work with new NVidia Container Toolkit. - Repeated calls to
RLModel.learn()now preserve internal counters for some episode logging statistics that used to be zeroed at the start of every call. - Fix DummyVecEnv.render for
num_envs > 1. This used to print a warning and then not render at all. (@shwang) - Fixed a bug in PPO2, ACER, A2C, and ACKTR where repeated calls to
learn(total_timesteps)reset the environment on every call, potentially biasing samples toward early episode timesteps. (@shwang) - Fixed by adding lazy property
ActorCriticRLModel.runner. Subclasses now use lazily-generated self.runnerinstead of reinitializing a new Runner every timelearn()is called.
- Fixed by adding lazy property
- Fixed a bug in
check_envwhere it would fail on high dimensional action spaces - Fixed
Monitor.close()that was not calling the parent method - Fixed a bug in
BaseRLModelwhen seeding vectorized environments. (@NeoExtended) - Fixed
num_timestepscomputation to be consistent between algorithms (updated afterenv.step()) OnlyTRPOandPPO1update it differently (after synchronization) because they rely on MPI - Fixed bug in
TRPOwith NaN standardized advantages (@richardwu) - Fixed partial minibatch computation in ExpertDataset (@richardwu)
- Fixed normalization (with
VecNormalize) for off-policy algorithms - Fixed
sync_envs_normalizationto sync the reward normalization too - Bump minimum Gym version (>=0.11)
Deprecations:¶
Others:¶
- Removed redundant return value from
a2c.utils::total_episode_reward_logger. (@shwang) - Cleanup and refactoring in
common/identity_env.py(@shwang) - Added a Makefile to simplify common development tasks (build the doc, type check, run the tests)
Documentation:¶
- Add dedicated page for callbacks
- Fixed example for creating a GIF (@KuKuXia)
- Change Colab links in the README to point to the notebooks repo
- Fix typo in Reinforcement Learning Tips and Tricks page. (@mmcenta)
Release 2.9.0 (2019-12-20)¶
Reproducible results, automatic ``VecEnv`` wrapping, env checker and more usability improvements
Breaking Changes:¶
- The
seedargument has been moved from learn() method to model constructor in order to have reproducible results allow_early_resetsof theMonitorwrapper now default toTruemake_atari_envnow returns aDummyVecEnvby default (instead of aSubprocVecEnv) this usually improves performance.- Fix inconsistency of sample type, so that mode/sample function returns tensor of tf.int64 in CategoricalProbabilityDistribution/MultiCategoricalProbabilityDistribution (@seheevic)
New Features:¶
Add
n_cpu_tf_sessto model constructor to choose the number of threads used by TensorflowEnvironments are automatically wrapped in a
DummyVecEnvif needed when passing them to the model constructorAdded
stable_baselines.common.make_vec_envhelper to simplify VecEnv creationAdded
stable_baselines.common.evaluation.evaluate_policyhelper to simplify model evaluationVecNormalizechanges:- Now supports being pickled and unpickled (@AdamGleave).
- New methods
.normalize_obs(obs)and normalize_reward(rews) apply normalization to arbitrary observation or rewards without updating statistics (@shwang) .get_original_reward()returns the unnormalized rewards from the most recent timestep.reset()now collects observation statistics (used to only apply normalization)
Add parameter
exploration_initial_epsto DQN. (@jdossgollin)Add type checking and PEP 561 compliance. Note: most functions are still not annotated, this will be a gradual process.
DDPG, TD3 and SAC accept non-symmetric action spaces. (@Antymon)
Add
check_envutil to check if a custom environment follows the gym interface (@araffin and @justinkterry)
Bug Fixes:¶
- Fix seeding, so it is now possible to have deterministic results on cpu
- Fix a bug in DDPG where
predictmethod with deterministic=False would fail - Fix a bug in TRPO: mean_losses was not initialized causing the logger to crash when there was no gradients (@MarvineGothic)
- Fix a bug in
cmd_utilfrom API change in recent Gym versions - Fix a bug in DDPG, TD3 and SAC where warmup and random exploration actions would end up scaled in the replay buffer (@Antymon)
Deprecations:¶
nprocs(ACKTR) andnum_procs(ACER) are deprecated in favor ofn_cpu_tf_sesswhich is now common to all algorithmsVecNormalize:load_running_averageandsave_running_averageare deprecated in favour of using pickle.
Others:¶
- Add upper bound for Tensorflow version (<2.0.0).
- Refactored test to remove duplicated code
- Add pull request template
- Replaced redundant code in load_results (@jbulow)
- Minor PEP8 fixes in dqn.py (@justinkterry)
- Add a message to the assert in
PPO2 - Update replay buffer doctring
- Fix
VecEnvdocstrings
Documentation:¶
- Add plotting to the Monitor example (@rusu24edward)
- Add Snake Game AI project (@pedrohbtp)
- Add note on the support Tensorflow versions.
- Remove unnecessary steps required for Windows installation.
- Remove
DummyVecEnvcreation when not needed - Added
make_vec_envto the examples to simplify VecEnv creation - Add QuaRL project (@srivatsankrishnan)
- Add Pwnagotchi project (@evilsocket)
- Fix multiprocessing example (@rusu24edward)
- Fix
result_plotterexample - Add JNRR19 tutorial (by @edbeeching, @hill-a and @araffin)
- Updated notebooks link
- Fix typo in algos.rst, “containes” to “contains” (@SyllogismRXS)
- Fix outdated source documentation for load_results
- Add PPO_CPP project (@Antymon)
- Add section on C++ portability of Tensorflow models (@Antymon)
- Update custom env documentation to reflect new gym API for the
close()method (@justinkterry) - Update custom env documentation to clarify what step and reset return (@justinkterry)
- Add RL tips and tricks for doing RL experiments
- Corrected lots of typos
- Add spell check to documentation if available
Release 2.8.0 (2019-09-29)¶
MPI dependency optional, new save format, ACKTR with continuous actions
Breaking Changes:¶
- OpenMPI-dependent algorithms (PPO1, TRPO, GAIL, DDPG) are disabled in the
default installation of stable_baselines.
mpi4pyis now installed as an extra. Whenmpi4pyis not available, stable-baselines skips imports of OpenMPI-dependent algorithms. See installation notes and Issue #430. - SubprocVecEnv now defaults to a thread-safe start method,
forkserverwhen available and otherwisespawn. This may require application code be wrapped inif __name__ == '__main__'. You can restore previous behavior by explicitly settingstart_method = 'fork'. See PR #428. - Updated dependencies: tensorflow v1.8.0 is now required
- Removed
checkpoint_pathandcheckpoint_freqargument fromDQNthat were not used - Removed
bench/benchmark.pythat was not used - Removed several functions from
common/tf_util.pythat were not used - Removed
ppo1/run_humanoid.py
New Features:¶
- important change Switch to using zip-archived JSON and Numpy
savezfor storing models for better support across library/Python versions. (@Miffyli) - ACKTR now supports continuous actions
- Add
double_qargument toDQNconstructor
Bug Fixes:¶
- Skip automatic imports of OpenMPI-dependent algorithms to avoid an issue where OpenMPI would cause stable-baselines to hang on Ubuntu installs. See installation notes and Issue #430.
- Fix a bug when calling
logger.configure()with MPI enabled (@keshaviyengar) - set
allow_pickle=Truefor numpy>=1.17.0 when loading expert dataset - Fix a bug when using VecCheckNan with numpy ndarray as state. Issue #489. (@ruifeng96150)
Deprecations:¶
- Models saved with cloudpickle format (stable-baselines<=2.7.0) are now deprecated in favor of zip-archive format for better support across Python/Tensorflow versions. (@Miffyli)
Others:¶
- Implementations of noise classes (
AdaptiveParamNoiseSpec,NormalActionNoise,OrnsteinUhlenbeckActionNoise) were moved from stable_baselines.ddpg.noise tostable_baselines.common.noise. The API remains backward-compatible; for examplefrom stable_baselines.ddpg.noise import NormalActionNoiseis still okay. (@shwang) - Docker images were updated
- Cleaned up files in
common/folder and in acktr/ folder that were only used by old ACKTR version (e.g. filter.py) - Renamed acktr_disc.py to acktr.py
Documentation:¶
- Add WaveRL project (@jaberkow)
- Add Fenics-DRL project (@DonsetPG)
- Fix and rename custom policy names (@eavelardev)
- Add documentation on exporting models.
- Update maintainers list (Welcome to @Miffyli)
Release 2.7.0 (2019-07-31)¶
Twin Delayed DDPG (TD3) and GAE bug fix (TRPO, PPO1, GAIL)
Breaking Changes:¶
New Features:¶
- added Twin Delayed DDPG (TD3) algorithm, with HER support
- added support for continuous action spaces to
action_probability, computing the PDF of a Gaussian policy in addition to the existing support for categorical stochastic policies. - added flag to
action_probabilityto return log-probabilities. - added support for python lists and numpy arrays in
logger.writekvs. (@dwiel) - the info dict returned by VecEnvs now include a
terminal_observationkey providing access to the last observation in a trajectory. (@qxcv)
Bug Fixes:¶
- fixed a bug in
traj_segment_generatorwhere theepisode_startswas wrongly recorded, resulting in wrong calculation of Generalized Advantage Estimation (GAE), this affects TRPO, PPO1 and GAIL (thanks to @miguelrass for spotting the bug) - added missing property
n_batchinBasePolicy.
Deprecations:¶
Others:¶
- renamed some keys in
traj_segment_generatorto be more meaningful - retrieve unnormalized reward when using Monitor wrapper with TRPO, PPO1 and GAIL to display them in the logs (mean episode reward)
- clean up DDPG code (renamed variables)
Documentation:¶
- doc fix for the hyperparameter tuning command in the rl zoo
- added an example on how to log additional variable with tensorboard and a callback
Release 2.6.0 (2019-06-12)¶
Hindsight Experience Replay (HER) - Reloaded | get/load parameters
Breaking Changes:¶
- breaking change removed
stable_baselines.ddpg.memoryin favor ofstable_baselines.deepq.replay_buffer(see fix below)
Breaking Change: DDPG replay buffer was unified with DQN/SAC replay buffer. As a result, when loading a DDPG model trained with stable_baselines<2.6.0, it throws an import error. You can fix that using:
import sys
import pkg_resources
import stable_baselines
# Fix for breaking change for DDPG buffer in v2.6.0
if pkg_resources.get_distribution("stable_baselines").version >= "2.6.0":
sys.modules['stable_baselines.ddpg.memory'] = stable_baselines.deepq.replay_buffer
stable_baselines.deepq.replay_buffer.Memory = stable_baselines.deepq.replay_buffer.ReplayBuffer
We recommend you to save again the model afterward, so the fix won’t be needed the next time the trained agent is loaded.
New Features:¶
- revamped HER implementation: clean re-implementation from scratch, now supports DQN, SAC and DDPG
- add
action_noiseparam for SAC, it helps exploration for problem with deceptive reward - The parameter
filter_sizeof the functionconvin A2C utils now supports passing a list/tuple of two integers (height and width), in order to have non-squared kernel matrix. (@yutingsz) - add
random_explorationparameter for DDPG and SAC, it may be useful when using HER + DDPG/SAC. This hack was present in the original OpenAI Baselines DDPG + HER implementation. - added
load_parametersandget_parametersto base RL class. With these methods, users are able to load and get parameters to/from existing model, without touching tensorflow. (@Miffyli) - added specific hyperparameter for PPO2 to clip the value function (
cliprange_vf) - added
VecCheckNanwrapper
Bug Fixes:¶
- bugfix for
VecEnvWrapper.__getattr__which enables access to class attributes inherited from parent classes. - fixed path splitting in
TensorboardWriter._get_latest_run_id()on Windows machines (@PatrickWalter214) - fixed a bug where initial learning rate is logged instead of its placeholder in
A2C.setup_model(@sc420) - fixed a bug where number of timesteps is incorrectly updated and logged in
A2C.learnandA2C._train_step(@sc420) - fixed
num_timesteps(total_timesteps) variable in PPO2 that was wrongly computed. - fixed a bug in DDPG/DQN/SAC, when there were the number of samples in the replay buffer was lesser than the batch size (thanks to @dwiel for spotting the bug)
- removed
a2c.utils.find_trainable_paramsplease usecommon.tf_util.get_trainable_varsinstead.find_trainable_paramswas returning all trainable variables, discarding the scope argument. This bug was causing the model to save duplicated parameters (for DDPG and SAC) but did not affect the performance.
Deprecations:¶
- deprecated
memory_limitandmemory_policyin DDPG, please usebuffer_sizeinstead. (will be removed in v3.x.x)
Others:¶
- important change switched to using dictionaries rather than lists when storing parameters, with tensorflow Variable names being the keys. (@Miffyli)
- removed unused dependencies (tdqm, dill, progressbar2, seaborn, glob2, click)
- removed
get_available_gpusfunction which hadn’t been used anywhere (@Pastafarianist)
Documentation:¶
- added guide for managing
NaNandinf - updated ven_env doc
- misc doc updates
Release 2.5.1 (2019-05-04)¶
Bug fixes + improvements in the VecEnv
Warning: breaking changes when using custom policies
- doc update (fix example of result plotter + improve doc)
- fixed logger issues when stdout lacks
readfunction - fixed a bug in
common.dataset.Datasetwhere shuffling was not disabled properly (it affects only PPO1 with recurrent policies) - fixed output layer name for DDPG q function, used in pop-art normalization and l2 regularization of the critic
- added support for multi env recording to
generate_expert_traj(@XMaster96) - added support for LSTM model recording to
generate_expert_traj(@XMaster96) GAIL: remove mandatory matplotlib dependency and refactor as subclass ofTRPO(@kantneel and @AdamGleave)- added
get_attr(),env_method()andset_attr()methods for all VecEnv. Those methods now all acceptindiceskeyword to select a subset of envs.set_attrnow returnsNonerather than a list ofNone. (@kantneel) GAIL:gail.dataset.ExpertDatasetsupports loading from memory rather than file, andgail.dataset.record_expertsupports returning in-memory rather than saving to file.- added support in
VecEnvWrapperfor accessing attributes of arbitrarily deeply nested instances ofVecEnvWrapperandVecEnv. This is allowed as long as the attribute belongs to exactly one of the nested instances i.e. it must be unambiguous. (@kantneel) - fixed bug where result plotter would crash on very short runs (@Pastafarianist)
- added option to not trim output of result plotter by number of timesteps (@Pastafarianist)
- clarified the public interface of
BasePolicyandActorCriticPolicy. Breaking change when using custom policies:masks_phis now calleddones_ph, and most placeholders were made private: e.g.self.value_fnis nowself._value_fn - support for custom stateful policies.
- fixed episode length recording in
trpo_mpi.utils.traj_segment_generator(@GerardMaggiolino)
Release 2.5.0 (2019-03-28)¶
Working GAIL, pretrain RL models and hotfix for A2C with continuous actions
- fixed various bugs in GAIL
- added scripts to generate dataset for gail
- added tests for GAIL + data for Pendulum-v0
- removed unused
utilsfile in DQN folder - fixed a bug in A2C where actions were cast to
int32even in the continuous case - added addional logging to A2C when Monitor wrapper is used
- changed logging for PPO2: do not display NaN when reward info is not present
- change default value of A2C lr schedule
- removed behavior cloning script
- added
pretrainmethod to base class, in order to use behavior cloning on all models - fixed
close()method for DummyVecEnv. - added support for Dict spaces in DummyVecEnv and SubprocVecEnv. (@AdamGleave)
- added support for arbitrary multiprocessing start methods and added a warning about SubprocVecEnv that are not thread-safe by default. (@AdamGleave)
- added support for Discrete actions for GAIL
- fixed deprecation warning for tf: replaces
tf.to_float()bytf.cast() - fixed bug in saving and loading ddpg model when using normalization of obs or returns (@tperol)
- changed DDPG default buffer size from 100 to 50000.
- fixed a bug in
ddpg.pyincombined_statsfor eval. Computed mean oneval_episode_rewardsandeval_qs(@keshaviyengar) - fixed a bug in
setup.pythat would error on non-GPU systems without TensorFlow installed
Release 2.4.1 (2019-02-11)¶
Bug fixes and improvements
- fixed computation of training metrics in TRPO and PPO1
- added
reset_num_timestepskeyword when calling train() to continue tensorboard learning curves - reduced the size taken by tensorboard logs (added a
full_tensorboard_logto enable full logging, which was the previous behavior) - fixed image detection for tensorboard logging
- fixed ACKTR for recurrent policies
- fixed gym breaking changes
- fixed custom policy examples in the doc for DQN and DDPG
- remove gym spaces patch for equality functions
- fixed tensorflow dependency: cpu version was installed overwritting tensorflow-gpu when present.
- fixed a bug in
traj_segment_generator(used in ppo1 and trpo) wherenewwas not updated. (spotted by @junhyeokahn)
Release 2.4.0 (2019-01-17)¶
Soft Actor-Critic (SAC) and policy kwargs
- added Soft Actor-Critic (SAC) model
- fixed a bug in DQN where prioritized_replay_beta_iters param was not used
- fixed DDPG that did not save target network parameters
- fixed bug related to shape of true_reward (@abhiskk)
- fixed example code in documentation of tf_util:Function (@JohannesAck)
- added learning rate schedule for SAC
- fixed action probability for continuous actions with actor-critic models
- added optional parameter to action_probability for likelihood calculation of given action being taken.
- added more flexible custom LSTM policies
- added auto entropy coefficient optimization for SAC
- clip continuous actions at test time too for all algorithms (except SAC/DDPG where it is not needed)
- added a mean to pass kwargs to policy when creating a model (+ save those kwargs)
- fixed DQN examples in DQN folder
- added possibility to pass activation function for DDPG, DQN and SAC
Release 2.3.0 (2018-12-05)¶
- added support for storing model in file like object. (thanks to @erniejunior)
- fixed wrong image detection when using tensorboard logging with DQN
- fixed bug in ppo2 when passing non callable lr after loading
- fixed tensorboard logging in ppo2 when nminibatches=1
- added early stoppping via callback return value (@erniejunior)
- added more flexible custom mlp policies (@erniejunior)
Release 2.2.1 (2018-11-18)¶
- added VecVideoRecorder to record mp4 videos from environment.
Release 2.2.0 (2018-11-07)¶
- Hotfix for ppo2, the wrong placeholder was used for the value function
Release 2.1.2 (2018-11-06)¶
- added
async_eigen_decompparameter for ACKTR and set it toFalseby default (remove deprecation warnings) - added methods for calling env methods/setting attributes inside a VecEnv (thanks to @bjmuld)
- updated gym minimum version
Release 2.1.1 (2018-10-20)¶
- fixed MpiAdam synchronization issue in PPO1 (thanks to @brendenpetersen) issue #50
- fixed dependency issues (new mujoco-py requires a mujoco license + gym broke MultiDiscrete space shape)
Release 2.1.0 (2018-10-2)¶
Warning
This version contains breaking changes for DQN policies, please read the full details
Bug fixes + doc update
- added patch fix for equal function using gym.spaces.MultiDiscrete and gym.spaces.MultiBinary
- fixes for DQN action_probability
- re-added double DQN + refactored DQN policies breaking changes
- replaced
asyncwithasync_eigen_decompin ACKTR/KFAC for python 3.7 compatibility - removed action clipping for prediction of continuous actions (see issue #36)
- fixed NaN issue due to clipping the continuous action in the wrong place (issue #36)
- documentation was updated (policy + DDPG example hyperparameters)
Release 2.0.0 (2018-09-18)¶
Warning
This version contains breaking changes, please read the full details
Tensorboard, refactoring and bug fixes
- Renamed DeepQ to DQN breaking changes
- Renamed DeepQPolicy to DQNPolicy breaking changes
- fixed DDPG behavior breaking changes
- changed default policies for DDPG, so that DDPG now works correctly breaking changes
- added more documentation (some modules from common).
- added doc about using custom env
- added Tensorboard support for A2C, ACER, ACKTR, DDPG, DeepQ, PPO1, PPO2 and TRPO
- added episode reward to Tensorboard
- added documentation for Tensorboard usage
- added Identity for Box action space
- fixed render function ignoring parameters when using wrapped environments
- fixed PPO1 and TRPO done values for recurrent policies
- fixed image normalization not occurring when using images
- updated VecEnv objects for the new Gym version
- added test for DDPG
- refactored DQN policies
- added registry for policies, can be passed as string to the agent
- added documentation for custom policies + policy registration
- fixed numpy warning when using DDPG Memory
- fixed DummyVecEnv not copying the observation array when stepping and resetting
- added pre-built docker images + installation instructions
- added
deterministicargument in the predict function - added assert in PPO2 for recurrent policies
- fixed predict function to handle both vectorized and unwrapped environment
- added input check to the predict function
- refactored ActorCritic models to reduce code duplication
- refactored Off Policy models (to begin HER and replay_buffer refactoring)
- added tests for auto vectorization detection
- fixed render function, to handle positional arguments
Release 1.0.7 (2018-08-29)¶
Bug fixes and documentation
- added html documentation using sphinx + integration with read the docs
- cleaned up README + typos
- fixed normalization for DQN with images
- fixed DQN identity test
Release 1.0.1 (2018-08-20)¶
Refactored Stable Baselines
- refactored A2C, ACER, ACTKR, DDPG, DeepQ, GAIL, TRPO, PPO1 and PPO2 under a single constant class
- added callback to refactored algorithm training
- added saving and loading to refactored algorithms
- refactored ACER, DDPG, GAIL, PPO1 and TRPO to fit with A2C, PPO2 and ACKTR policies
- added new policies for most algorithms (Mlp, MlpLstm, MlpLnLstm, Cnn, CnnLstm and CnnLnLstm)
- added dynamic environment switching (so continual RL learning is now feasible)
- added prediction from observation and action probability from observation for all the algorithms
- fixed graphs issues, so models wont collide in names
- fixed behavior_clone weight loading for GAIL
- fixed Tensorflow using all the GPU VRAM
- fixed models so that they are all compatible with vectorized environments
- fixed
set_global_seedto updategym.spaces’s random seed - fixed PPO1 and TRPO performance issues when learning identity function
- added new tests for loading, saving, continuous actions and learning the identity function
- fixed DQN wrapping for atari
- added saving and loading for Vecnormalize wrapper
- added automatic detection of action space (for the policy network)
- fixed ACER buffer with constant values assuming n_stack=4
- fixed some RL algorithms not clipping the action to be in the action_space, when using
gym.spaces.Box - refactored algorithms can take either a
gym.Environmentor astr([if the environment name is registered](https://github.com/openai/gym/wiki/Environments)) - Hoftix in ACER (compared to v1.0.0)
Future Work :
- Finish refactoring HER
- Refactor ACKTR and ACER for continuous implementation
Release 0.1.6 (2018-07-27)¶
Deobfuscation of the code base + pep8 and fixes
- Fixed
tf.session().__enter__()being used, rather thansess = tf.session()and passing the session to the objects - Fixed uneven scoping of TensorFlow Sessions throughout the code
- Fixed rolling vecwrapper to handle observations that are not only grayscale images
- Fixed deepq saving the environment when trying to save itself
- Fixed
ValueError: Cannot take the length of Shape with unknown rank.inacktr, when runningrun_atari.pyscript. - Fixed calling baselines sequentially no longer creates graph conflicts
- Fixed mean on empty array warning with deepq
- Fixed kfac eigen decomposition not cast to float64, when the parameter use_float64 is set to True
- Fixed Dataset data loader, not correctly resetting id position if shuffling is disabled
- Fixed
EOFErrorwhen reading from connection in theworkerinsubproc_vec_env.py - Fixed
behavior_cloneweight loading and saving for GAIL - Avoid taking root square of negative number in
trpo_mpi.py - Removed some duplicated code (a2cpolicy, trpo_mpi)
- Removed unused, undocumented and crashing function
reset_taskinsubproc_vec_env.py - Reformated code to PEP8 style
- Documented all the codebase
- Added atari tests
- Added logger tests
Missing: tests for acktr continuous (+ HER, rely on mujoco…)
Maintainers¶
Stable-Baselines is currently maintained by Ashley Hill (aka @hill-a), Antonin Raffin (aka @araffin), Maximilian Ernestus (aka @erniejunior), Adam Gleave (@AdamGleave) and Anssi Kanervisto (aka @Miffyli).
Contributors (since v2.0.0):¶
In random order…
Thanks to @bjmuld @iambenzo @iandanforth @r7vme @brendenpetersen @huvar @abhiskk @JohannesAck @EliasHasle @mrakgr @Bleyddyn @antoine-galataud @junhyeokahn @AdamGleave @keshaviyengar @tperol @XMaster96 @kantneel @Pastafarianist @GerardMaggiolino @PatrickWalter214 @yutingsz @sc420 @Aaahh @billtubbs @Miffyli @dwiel @miguelrass @qxcv @jaberkow @eavelardev @ruifeng96150 @pedrohbtp @srivatsankrishnan @evilsocket @MarvineGothic @jdossgollin @SyllogismRXS @rusu24edward @jbulow @Antymon @seheevic @justinkterry @edbeeching @flodorner @KuKuXia @NeoExtended @solliet @mmcenta @richardwu
Projects¶
This is a list of projects using stable-baselines. Please tell us, if you want your project to appear on this page ;)
Learning to drive in a day¶
Implementation of reinforcement learning approach to make a donkey car learn to drive. Uses DDPG on VAE features (reproducing paper from wayve.ai)
Donkey Gym¶
OpenAI gym environment for donkeycar simulator.
Self-driving FZERO Artificial Intelligence¶
Series of videos on how to make a self-driving FZERO artificial intelligence using reinforcement learning algorithms PPO2 and A2C.
S-RL Toolbox¶
S-RL Toolbox: Reinforcement Learning (RL) and State Representation Learning (SRL) for Robotics. Stable-Baselines was originally developped for this project.
Roboschool simulations training on Amazon SageMaker¶
“In this notebook example, we will make HalfCheetah learn to walk using the stable-baselines […]”
MarathonEnvs + OpenAi.Baselines¶
Experimental - using OpenAI baselines with MarathonEnvs (ML-Agents)
Learning to drive smoothly in minutes¶
Implementation of reinforcement learning approach to make a car learn to drive smoothly in minutes. Uses SAC on VAE features.
Making Roboy move with elegance¶
Project around Roboy, a tendon-driven robot, that enabled it to move its shoulder in simulation to reach a pre-defined point in 3D space. The agent used Proximal Policy Optimization (PPO) or Soft Actor-Critic (SAC) and was tested on the real hardware.
Train a ROS-integrated mobile robot (differential drive) to avoid dynamic objects¶
The RL-agent serves as local planner and is trained in a simulator, fusion of the Flatland Simulator and the crowd simulator Pedsim. This was tested on a real mobile robot. The Proximal Policy Optimization (PPO) algorithm is applied.
Adversarial Policies: Attacking Deep Reinforcement Learning¶
Uses Stable Baselines to train adversarial policies that attack pre-trained victim policies in a zero-sum multi-agent environments. May be useful as an example of how to integrate Stable Baselines with Ray to perform distributed experiments and Sacred for experiment configuration and monitoring.
WaveRL: Training RL agents to perform active damping¶
Reinforcement learning is used to train agents to control pistons attached to a bridge to cancel out vibrations. The bridge is modeled as a one dimensional oscillating system and dynamics are simulated using a finite difference solver. Agents were trained using Proximal Policy Optimization. See presentation for environment detalis.
Fenics-DRL: Fluid mechanics and Deep Reinforcement Learning¶
Deep Reinforcement Learning is used to control the position or the shape of obstacles in different fluids in order to optimize drag or lift. Fenics is used for the Fluid Mechanics part, and Stable Baselines is used for the DRL.
Air Learning: An AI Research Platform Algorithm Hardware Benchmarking of Autonomous Aerial Robots¶
Aerial robotics is a cross-layer, interdisciplinary field. Air Learning is an effort to bridge seemingly disparate fields.
Designing an autonomous robot to perform a task involves interactions between various boundaries spanning from modeling the environment down to the choice of onboard computer platform available in the robot. Our goal through building Air Learning is to provide researchers with a cross-domain infrastructure that allows them to holistically study and evaluate reinforcement learning algorithms for autonomous aerial machines. We use stable-baselines to train UAV agent with Deep Q-Networks and Proximal Policy Optimization algorithms.
Snake Game AI¶
AI to play the classic snake game. The game was trained using PPO2 available from stable-baselines and then exported to tensorflowjs to run directly on the browser
Pwnagotchi¶
Pwnagotchi is an A2C-based “AI” powered by bettercap and running on a Raspberry Pi Zero W that learns from its surrounding WiFi environment in order to maximize the crackable WPA key material it captures (either through passive sniffing or by performing deauthentication and association attacks). This material is collected on disk as PCAP files containing any form of handshake supported by hashcat, including full and half WPA handshakes as well as PMKIDs.
Quantized Reinforcement Learning (QuaRL)¶
QuaRL is a open-source framework to study the effects of quantization broad spectrum of reinforcement learning algorithms. The RL algorithms we used in this study are from stable-baselines.
PPO_CPP: C++ version of a Deep Reinforcement Learning algorithm PPO¶
Executes PPO at C++ level yielding notable execution performance speedups. Uses Stable Baselines to create a computational graph which is then used for training with custom environments by machine-code-compiled binary.
Plotting Results¶
-
stable_baselines.results_plotter.main()[source]¶ Example usage in jupyter-notebook
from stable_baselines import results_plotter %matplotlib inline results_plotter.plot_results(["./log"], 10e6, results_plotter.X_TIMESTEPS, "Breakout")
Here ./log is a directory containing the monitor.csv files
-
stable_baselines.results_plotter.plot_curves(xy_list, xaxis, title)[source]¶ plot the curves
Parameters: - xy_list – ([(np.ndarray, np.ndarray)]) the x and y coordinates to plot
- xaxis – (str) the axis for the x and y output (can be X_TIMESTEPS=’timesteps’, X_EPISODES=’episodes’ or X_WALLTIME=’walltime_hrs’)
- title – (str) the title of the plot
-
stable_baselines.results_plotter.plot_results(dirs, num_timesteps, xaxis, task_name)[source]¶ plot the results
Parameters: - dirs – ([str]) the save location of the results to plot
- num_timesteps – (int or None) only plot the points below this value
- xaxis – (str) the axis for the x and y output (can be X_TIMESTEPS=’timesteps’, X_EPISODES=’episodes’ or X_WALLTIME=’walltime_hrs’)
- task_name – (str) the title of the task to plot
-
stable_baselines.results_plotter.rolling_window(array, window)[source]¶ apply a rolling window to a np.ndarray
Parameters: - array – (np.ndarray) the input Array
- window – (int) length of the rolling window
Returns: (np.ndarray) rolling window on the input array
-
stable_baselines.results_plotter.ts2xy(timesteps, xaxis)[source]¶ Decompose a timesteps variable to x ans ys
Parameters: - timesteps – (Pandas DataFrame) the input data
- xaxis – (str) the axis for the x and y output (can be X_TIMESTEPS=’timesteps’, X_EPISODES=’episodes’ or X_WALLTIME=’walltime_hrs’)
Returns: (np.ndarray, np.ndarray) the x and y output
-
stable_baselines.results_plotter.window_func(var_1, var_2, window, func)[source]¶ apply a function to the rolling window of 2 arrays
Parameters: - var_1 – (np.ndarray) variable 1
- var_2 – (np.ndarray) variable 2
- window – (int) length of the rolling window
- func – (numpy function) function to apply on the rolling window on variable 2 (such as np.mean)
Returns: (np.ndarray, np.ndarray) the rolling output with applied function
Citing Stable Baselines¶
To cite this project in publications:
@misc{stable-baselines,
author = {Hill, Ashley and Raffin, Antonin and Ernestus, Maximilian and Gleave, Adam and Kanervisto, Anssi and Traore, Rene and Dhariwal, Prafulla and Hesse, Christopher and Klimov, Oleg and Nichol, Alex and Plappert, Matthias and Radford, Alec and Schulman, John and Sidor, Szymon and Wu, Yuhuai},
title = {Stable Baselines},
year = {2018},
publisher = {GitHub},
journal = {GitHub repository},
howpublished = {\url{https://github.com/hill-a/stable-baselines}},
}
Contributing¶
To any interested in making the rl baselines better, there is still some improvements that needs to be done. A full TODO list is available in the roadmap.
If you want to contribute, please read CONTRIBUTING.md first.