/atari-prediction-benchmark

Primary LanguagePython

Arcade prediction benchmark

For benchmark policy evaluation on Atari. The policies are taken from pre-trained rainbow DQN agent from the Chiner RL model zoo [1].

A small test policy for Pong with sample code is available in test.py.

Link to full policies: Google drive.

You can cite the benchmark as:

@article{javed2023scalable,
  title={Scalable real-time recurrent learning using columnar-constructive networks},
  author={Javed, Khurram and Shah, Haseeb and Sutton, Richard S and White, Martha},
  journal={Journal of Machine Learning Research},
  volume={24},
  pages={1--34},
  year={2023}
}

[1] Fujita, Yasuhiro, et al. "Chainerrl: A deep reinforcement learning library." The Journal of Machine Learning Research 22.1 (2021): 3557-3570.