conda create -n dfbt python=3.10
conda activate dfbt
pip install -r requirement.yaml
pip install "cython<3"
pip install git+https://github.com/Farama-Foundation/d4rl@master#egg=d4rl
python3 scripts/train_dfbt.py --dataset_name halfcheetah --delay 128
python3 scripts/train_dfbt_sac.py --env_name HalfCheetah-v2 --delay 128
| Environment | Truth | DATS | D-Dreamer | D-SAC | DBFT-SAC (ours) |
|---|---|---|---|---|---|
| HalfCheetah-v2 | |||||
| Hopper-v2 | |||||
| Walker2d-v2 |
@inproceedings{wu2025directly,
title={Directly Forecasting Belief for Reinforcement Learning with Delays},
author={Wu, Qingyuan and Wang, Yuhui and Zhan, Simon Sinong and Wang, Yixuan and Lin, Chung-Wei and Lv, Chen and Zhu, Qi and Schmidhuber, J{\"u}rgen and Huang, Chao},
booktitle={International Conference on Machine Learning},
year={2025},
organization={PMLR}
}