Metraipolis › archive › dc_a2ceabce70e1
Why Do Conventional World Models Fail to Learn Cellular Automata? — arXiv:2609.39604
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.39604
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T12:42:53.086Z
- SHA-256 of the archived text
0f69d83a2543a041ce47e69f92f8aeaa8f182f50a55404d1ce8b362547a39afa- Stable URL of this page
https://metraipolis.com/documento/dc_a2ceabce70e1
How to cite
Why Do Conventional World Models Fail to Learn Cellular Automata? — arXiv:2609.39604 — arXiv (en), https://arxiv.org/abs/2609.39604 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_a2ceabce70e1 (Metraipolis archive dc_a2ceabce70e1, sha256 0f69d83a2543a041ce47e69f92f8aeaa8f182f50a55404d1ce8b362547a39afa)
BibTeX
@misc{dca2ceabce70e1,
title = {Why Do Conventional World Models Fail to Learn Cellular Automata? — arXiv:2609.39604},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.39604},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_a2ceabce70e1; sha256 0f69d83a2543a041ce47e69f92f8aeaa8f182f50a55404d1ce8b362547a39afa},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.39604. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Although conventional world models - auto-regressive or diffusion models based on transformers or convolutional networks - may learn surface statistics of world dynamics, can they learn the exact world dynamics from its observed history? Leveraging cellular automata as a simple testbed, we find the answer to be no in many cases. Conventional architectures predict most pixels correctly yet rarely complete a rollout: a CNN predicts 96.3% of cells but completes 18.9% of rollouts; a joint diffusion model completes none. We trace the gap to three failure modes of these world models - namely, they fail to exactly capture spatial locality, temporal locality or temporal stability. Simple changes repair each: (1) for spatial locality, two-dimensional rotary positions lift a transformer from 39.1% to 100% on the Game of Life; (2) for temporal locality, handing each token its cell's previous-frame neighbourhood lifts the same transformer from 25.8% to 99.9% on unseen rules; (3) for temporal stability, causal freezing lifts the same diffusion weights from 42.2% to 99.9%. None of the three changes touches the architectural backbone; each only modifies the information flow within it. We also compare joint and ordered sampling on billiards and, in an exploratory study, on a simulated Burgers equation. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-10-01 in cs.AI). Categorie arXiv: Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2609.39604 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-07T12:14:13.779Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-10-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).