rl-tools

This is a simple toolset for reinforcement learning (RL).

Installation

from PIP

pip install rl_tools

from source

python setup.py install

Usage

As a alone script:

for example:

python -m rl_tools.plotter --save --show

• enjoy.py enjoy the trained model, and track avg, med, std, min, max reward of episodes.
• plotter.py plot learning curves for training , see rl-plotter for detail.
• scheduler.py set training scheduler for tasks.

In RL agent:

for example:

from rl_tools.evaluator import Evaluator
from rl_tools.logger import Logger

• evaluator.py evaluate model while training.
• logger.py track learning process while training (same format as OpenAI baseline).