Code repository for the following paper:
Deterministic and Discriminative Imitation (D2-Imitation): Revisiting Adversarial Imitation for Sample Efficiency
Mingfei Sun, Sam Devlin, Katja Hofmann, Shimon Whiteson
AAAI 2022, [pdf].
Compared with adversarial methods | Compared with off-policy distribution matching |
---|---|
bash docker_build.sh
- Download the demonstration data
cd src
bash download_data-v2.sh # this might take a while
- Run in single thread mode
cd src
bash docker_shell.sh # start a docker container
python3 -m baselines.imitation.d2_imitation \
--seed 0 \
--env_id HalfCheetah-v2 \
--expert_path traj_configs-v2/halfcheetah-sac-20.txt \
--total_timesteps 500000
The experimental results are logged into log
folder.
Please cite our paper if you find this codebase useful:
@article{sun2021deterministic,
title={Deterministic and Discriminative Imitation (D2-Imitation): Revisiting Adversarial Imitation for Sample Efficiency},
author={Sun, Mingfei and Devlin, Sam and Hofmann, Katja and Whiteson, Shimon},
journal={arXiv preprint arXiv:2112.06054},
year={2021}
}