README.md

metadata

task_categories:
  - text-to-video
  - image-to-video
  - video-to-video
  - keypoint-detection
language:
  - en
license: other
license_name: license
license_link: https://github.com/AlayaLab/WildWorld/raw/refs/heads/main/LICENSE
arxiv: 2603.23497
extra_gated_heading: >-
  Before submitting your request, please make sure you have reviewed and
  acknowledged the WildWorld dataset license:
  https://github.com/AlayaLab/WildWorld/blob/main/LICENSE
extra_gated_description: >-
  Your submitted information will be used only to verify dataset access
  eligibility and will not be shared or used for any other purpose.
extra_gated_fields:
  First Name: text
  Last Name: text
  Country: country
  Affiliation: text
  By clicking Submit below, I accept the terms of the WildWorld dataset license: checkbox
  geo: ip_location
extra_gated_button_content: Submit

WildWorld: A Large-Scale Dataset for Dynamic World Modeling
with Actions and Explicit State toward Generative ARPG

This repo contains the dataset proposed in

WildWorld: A Large-Scale Dataset for Dynamic World Modeling with Actions and Explicit State toward Generative ARPG

Zhen Li, Zian Meng, Shuwei Shi, Wenshuo Peng, Yuwei Wu, Bo Zheng, Chuanhao Li, Kaipeng Zhang

Alaya Studio, Shanda AI Research Tokyo; Beijing Institute of Technology; Shanghai Innovation Institute

🧠Introduction

TL;DR We present WildWorld, a large-scale action-conditioned world modeling dataset with explicit state annotations, automatically collected from a photorealistic AAA action role-playing game. It features:

• 🎬 108M+ frames with per-frame annotations: character skeletons, actions & states (HP, animation, etc.), camera poses, and depth maps
• ⚔️ 450+ semantically meaningful actions including movement, attacks, and skill casting
• 🐉 Diverse content: 29 monster species, 4 player characters, 4 weapon types, 5 distinct stages
• 🕒 Long-horizon sequences: clips spanning up to 30+ minutes of continuous gameplay
• 📝 Hierarchical captions: both action-level and sample-level natural language descriptions

🚀 Quick Start

Please refer to our Github Repo.

📖Citation

If you find this project helpful, please consider citing:

@article{li2026wildworld,
  title={Wildworld: A large-scale dataset for dynamic world modeling with actions and explicit state toward generative arpg},
  author={Li, Zhen and Meng, Zian and Shi, Shuwei and Peng, Wenshuo and Wu, Yuwei and Zheng, Bo and Li, Chuanhao and Zhang, Kaipeng},
  journal={arXiv preprint arXiv:2603.23497},
  year={2026}
}