This repository contains the implementation of the Delayed, Agumented, Oblivious, and RNN agents from the paper: "Acting in Delayed Environments with Non-Stationary Markov Policies", Esther Derman^*, Gal Dalal^*, Shie Mannor (^*equal contribution), published in ICLR 2021.

The agent here supports the Atari environments. The simpler agent that supports Cartpole and Acrobot can be found here.

This is a fork of Stable-Baselines (v2.10.1, based on TensorFlow), with the addition of the delayed agent.

To set up the environment please follow the instructions in Stable-Baselines.

Running the code is straightforward using run_experiment_rl_delay.py.

To cite this repository in publications:

@article{derman2021acting,
  title={Acting in delayed environments with non-stationary markov policies},
  author={Derman, Esther and Dalal, Gal and Mannor, Shie},
  journal={International Conference on Learning Representations (ICLR)},
  year={2021}
}

Happy delaying!