https://github.com/marella/train

A library to build and train reinforcement learning agents in OpenAI Gym environments.
https://github.com/marella/train

gym machine-learning python reinforcement-learning tensorflow

Last synced: about 1 month ago
JSON representation

A library to build and train reinforcement learning agents in OpenAI Gym environments.

• Host: GitHub
• URL: https://github.com/marella/train
• Owner: marella
• License: mit
• Created: 2019-12-14T16:53:01.000Z (over 6 years ago)
• Default Branch: master
• Last Pushed: 2020-01-27T12:07:20.000Z (over 6 years ago)
• Last Synced: 2025-11-06T22:29:17.293Z (7 months ago)
• Topics: gym, machine-learning, python, reinforcement-learning, tensorflow
• Language: Python
• Homepage: https://train.readthedocs.io
• Size: 38.1 KB
• Stars: 1
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.rst
  • License: LICENSE

Awesome Lists containing this project

README

          
Train

=====

A library to build and train reinforcement learning agents in OpenAI Gym environments.

.. image:: https://travis-ci.org/marella/train.svg?branch=master

    :target: https://travis-ci.org/marella/train

    :alt: Build Status

.. image:: https://readthedocs.org/projects/train/badge/?version=latest

    :target: https://train.readthedocs.io/en/latest/?badge=latest

    :alt: Documentation Status

Read full documentation `here `_.

Getting Started

***************

An agent has to implement the ``act()`` method which takes the current ``state`` as input and returns an action:

.. code:: python

    from train import Agent

    class RandomAgent(Agent):

        def act(self, state):

            return self.env.action_space.sample()

Create an environment using OpenAI Gym_:

.. code:: python

    import gym

    env = gym.make('CartPole-v0')

Initialize your agent using the environment:

.. code:: python

    agent = RandomAgent(env=env)

Now you can start training your agent (in this example, the agent acts randomly always and doesn't learn anything):

.. code:: python

    scores = agent.train(episodes=100)

You can also visualize how the training progresses but it will slow down the process:

.. code:: python

    scores = agent.train(episodes=100, render=True)

Once you are done with the training, you can test it:

.. code:: python

    scores = agent.test(episodes=10)

Alternatively, visualize how it performs:

.. code:: python

    scores = agent.test(episodes=10, render=True)

To learn more about how to build an agent that learns see agents_ documentation.

See examples_ directory to see implementations of some algorithms (DQN, A3C, PPO etc.) created using TensorFlow, PyTorch and NN_ libraries.

Installation

************

Requirements:

-   Python >= 3.6

Install from PyPI (recommended):

::

    pip install train

Alternatively, install from source:

::

    git clone https://github.com/marella/train.git

    cd train

    pip install -e .

To run examples and tests, install from source.

Other libraries such as Gym_, TensorFlow_, PyTorch_ and NN_ should be installed separately.

Examples

********

To run examples, install TensorFlow_, PyTorch_ and install other dependencies:

::

    pip install -e .[examples]

and run an example in examples_ directory:

::

    cd examples

    python PPO.py

Testing

*******

To run tests, install dependencies:

::

    pip install -e .[tests]

and run:

::

    pytest tests

.. _agents: https://train.readthedocs.io/en/latest/agents.html

.. _examples: https://github.com/marella/train/tree/master/examples

.. _Gym: https://gym.openai.com/docs/

.. _TensorFlow: https://www.tensorflow.org/install

.. _PyTorch: https://pytorch.org/get-started

.. _NN: https://github.com/marella/nn