OPT-AIL

Official code for OPT-AIL: Provably and Practically Efficient Adversarial Imitation Learning with General Function Approximation.

Quickstart

Prerequisites

• Python version compatible with this project (see requires-python in pyproject.toml)
• uv installed

Install uv

curl -LsSf https://astral.sh/uv/install.sh | sh

Clone and install dependencies

git clone https://github.com/LAMDA-RL/OPT-AIL.git
cd OPT-AIL

uv sync

Usage

The expert trajectories used during the experiments can be found here: https://drive.google.com/file/d/1ZBZPjQITGLkiKusdGAwObLdomTTU3bqs/view?usp=drive_link

Set buffer_folder and model_folder in conf/config.yaml using absolute path.

Then, just run the scripts in the scripts dir. You can try as follows:

Run model-free opt-ail:

sh scripts/run_mf.sh

Run model-based opt-ail:

sh scripts/run_mb.sh

Citation

If you find this repository useful for your research, please cite:

@inproceedings{
	xu2024provably,
	title={Provably and Practically Efficient Adversarial Imitation Learning with General Function Approximation},
	author={Tian Xu, Zhilong Zhang, Ruishuo Chen, Yihao Sun, and Yang Yu},
	booktitle={The 38th Conference on Neural Information Processing System},
	year={2024},
	url={https://openreview.net/forum?id=7YdafFbhxL}
}