ALFWorld Raw Game Files (awawa-agi/alfworld-raw)

Summary

This dataset contains the raw ALFWorld abstract game files (TextWorld + PDDL) used to align high-level language planning with embodied ALFRED tasks.

Schema / Contents

Each entry includes:

• id: unique game instance identifier
• task_type: high-level task (e.g., pick_heat_then_place_in_recep)
• game_file_path: original .tw-pddl path
• game_content: full TextWorld + PDDL specification for abstract planning

Splits: train, eval_in_distribution, eval_out_of_distribution.

Quickstart

from datasets import load_dataset
ds = load_dataset("awawa-agi/alfworld-raw")
print(ds["train"].features)
print(ds["train"][0]["game_content"][:500])

Attribution

The abstract/concrete alignment framework is from ALFWorld, which builds on TextWorld and ties to the embodied ALFRED benchmark.

Disclaimer

This dataset is a redistribution of the original ALFWorld data for research convenience. The uploader does not claim any ownership or copyright over this content. All rights belong to the original authors and their affiliated institutions. This redistribution is provided as-is for academic and research purposes only.

Suggested citation (BibTeX)

@inproceedings{shridhar2021alfworld,
  title={{ALFWorld: Aligning Text and Embodied Environments for Interactive Learning}},
  author={Shridhar, Mohit and Yuan, Xingdi and C{\^o}t{\'e}, Marc-Alexandre and Bisk, Yonatan and Trischler, Adam and Hausknecht, Matthew},
  booktitle={International Conference on Learning Representations (ICLR)},
  year={2021},
  url={https://arxiv.org/abs/2010.03768}
}

@inproceedings{shridhar2020alfred,
  title={{ALFRED: A Benchmark for Interpreting Grounded Instructions for Everyday Tasks}},
  author={Shridhar, Mohit and Thomason, Jesse and Gordon, Daniel and Bisk, Yonatan and Han, Winson and Mottaghi, Roozbeh and Zettlemoyer, Luke and Fox, Dieter},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  year={2020},
  url={https://arxiv.org/abs/1912.01734}
}

@inproceedings{cote2018textworld,
  title={TextWorld: A Learning Environment for Text-based Games},
  author={C{\^o}t{\'e}, Marc-Alexandre and K{\'a}d{\'a}r, {\'A}kos and Yuan, Xingdi and Kybartas, Ben and Barnes, Tavian and Fine, Emery and Moore, James and Tao, Ruo Yu and Hausknecht, Matthew J. and El Asri, Layla and Adada, Mahmoud and others},
  booktitle={Computer Games Workshop at IJCAI},
  pages={41--75},
  year={2018}
}

References

• ALFWorld: aligning TextWorld with ALFRED
• ALFRED benchmark
• TextWorld environment