LIFT

This repository contains the code for the experiments conducted in the paper Trajectory-Level Data Augmentation for Offline Reinforcement Learning by Tobias Schmähling, Matthias Burkhardt, and Tobias Windisch (accepted at ICML26).

Run

To run the experiments for PO (Position only) with d=2 run

export WORKING_DIR=/path/to/po_2

ray start --head --dashboard-host 0.0.0.0 --num-cpus=10 --disable-usage-stats
python scripts/collect_augmentor_data.py --env='po' --n_actions=2 --path=experiment_name
python scripts/train.py --path=experiment_name

Citation

If you use this package for your scientific work, please consider citing:

@misc{schmaehling2026lift,
      title={Trajectory-Level Data Augmentation for Offline Reinforcement Learning}, 
      author={Tobias Schmähling and Matthias Burkhardt and Tobias Windisch},
      year={2026},
      eprint={2605.13401},
      archivePrefix={arXiv},
      primaryClass={cs.LG},
      url={https://arxiv.org/abs/2605.13401}, 
}