flowm123/blockworld

Block World Dataset

This repository contains the Block World dataset for experiments of FloWM (Flow Equivariant World Models), presented in the paper Flow Equivariant World Models: Memory for Partially Observed Dynamic Environments.

Project Page | GitHub Repository

Dataset Summary

The Block World dataset is a 3D partially observed video world modeling benchmark. It is designed to evaluate how world models handle continuous sensory input and underlying symmetries in environment dynamics. The dataset includes three main configurations:

• dynamic: The primary environment used for results in the paper, featuring moving objects.
• static: A version of the environment with static external objects.
• tex: A textured version of the environment to test visual complexity.

Each configuration contains both train and validation splits.

Usage

To use this dataset with the FloWM framework, the authors provide a download script in the associated GitHub repository to handle the extraction and setup of the data. For more details, please refer to the official code repository.

Citation

@misc{lillemark2026flowequivariantworldmodels,
    title={Flow Equivariant World Models: Memory for Partially Observed Dynamic Environments}, 
    author={Hansen Jin Lillemark and Benhao Huang and Fangneng Zhan and Yilun Du and Thomas Anderson Keller},
    year={2026},
    eprint={2601.01075},
    archivePrefix={arXiv},
    primaryClass={cs.LG},
    url={https://arxiv.org/abs/2601.01075}, 
  }