xiazaishuo/cloning-deterministic-worlds-maze9
Maze9 Dataset This dataset is released with the CVPR 2026 paper Cloning Deterministic Worlds: The Critical Role of Latent Geometry in Long-Horizon World Models. Contents 9x9_n0.8/: Maze9 training set 9x9_val/: Maze9 validation set Each split contains: info.hdf5: trajectory metadata, states, and actions obs/*.npy: observation videos for each trajectory Usage Set the paths in the training code: export TRAIN_DIR=/path/to/9x9_n0.8 export… See the full description on the dataset page: https://huggingface.co/datasets/xiazaishuo/cloning-deterministic-worlds-maze9.
Maze9 Dataset
This dataset is released with the CVPR 2026 paper Cloning Deterministic Worlds: The Critical Role of Latent Geometry in Long-Horizon World Models.
Contents
9x9_n0.8/: Maze9 training set9x9_val/: Maze9 validation set
Each split contains:
info.hdf5: trajectory metadata, states, and actionsobs/*.npy: observation videos for each trajectory
Usage
Set the paths in the training code:
export TRAIN_DIR=/path/to/9x9_n0.8
export VAL_DIR=/path/to/9x9_valCitation
@InProceedings{Xia_2026_CVPR,
author = {Xia, Zaishuo and Lu, Yukuan and Li, Xinyi and Xu, Yifan and Chen, Yubei},
title = {Cloning Deterministic Worlds: The Critical Role of Latent Geometry in Long-Horizon World Models},
booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
month = {June},
year = {2026},
pages = {32602-32612}
}