Official code for FTB: Flow to Better: Offline Preference-based Reinforcement Learning via Preferred Trajectory Generation.
First, you need to install Python packages listed in requirements.txt using pip install -r requirements.txt.
Then, just run the scripts in the run_scripts dir. You can try as follows:
For mujoco tasks:
sh run_scripts/run_mujoco.shFor Metaworld tasks:
sh run_scripts/run_metaworld.shFor NeoRL tasks:
sh run_scripts/run_neorl.shIf you find this repository useful for your research, please cite:
@inproceedings{
zhang2024flow,
title={Flow to Better: Offline Preference-based Reinforcement Learning via Preferred Trajectory Generation},
author={Zhilong Zhang and Yihao Sun and Junyin Ye and Tian-Shuo Liu and Jiaji Zhang and Yang Yu},
booktitle={The Twelfth International Conference on Learning Representations},
year={2024},
url={https://openreview.net/forum?id=EG68RSznLT}
}