gbartyzel / yadrl

Yet another deep reinforcement learning

Yet Another Deep Reinforcement Learning (YADRL)

YADRL is set of implementations of reinforcement learning algorithms. It is written fully in PyTorch.

Implemented algorithms

Installation

Install YADRL package:

pip install .

Execution example

To execute experiments just type run following command:

yadrl_runner --config_path link/to/config/file

TODO

Algorithm implementation:

• PPO
• TRPO
• VPG
• PER
• APE-X
• MPO
• HER
• IQN and M-IQN
• FQF

Software improvements:

• Documentation
• Distributed training
• On-policy base agent