This is an implementation of the method proposed in
NovelD: A Simple yet Effective Exploration Criterion and BeBold: Exploration Beyond the Boundary of Explored Regions
If you use this code in your own work, please cite our paper:
@article{zhang2021noveld,
title={NovelD: A Simple yet Effective Exploration Criterion},
author={Zhang, Tianjun and Xu, Huazhe and Wang, Xiaolong and Wu, Yi and Keutzer, Kurt and Gonzalez, Joseph E and Tian, Yuandong},
journal={Advances in Neural Information Processing Systems},
volume={34},
year={2021}
}
or
@article{zhang2020bebold,
title={BeBold: Exploration Beyond the Boundary of Explored Regions},
author={Zhang, Tianjun and Xu, Huazhe and Wang, Xiaolong and Wu, Yi and Keutzer, Kurt and Gonzalez, Joseph E and Tian, Yuandong},
journal={arXiv preprint arXiv:2012.08621},
year={2020}
}
# Install Instructions
conda create -n ride python=3.7
conda activate noveld
git clone git@github.com:tianjunz/NovelD.git
cd NovelD
pip install -r requirements.txt
OMP_NUM_THREADS=1 python main.py --model bebold --env MiniGrid-ObstructedMaze-2Dlhb-v0 --total_frames 500000000 --intrinsic_reward_coef 0.05 --entropy_cost 0.0005
Our vanilla RL algorithm is based on RIDE.
This code is under the CC-BY-NC 4.0 (Attribution-NonCommercial 4.0 International) license.