Metraipolis › archive › dc_0183aaf41da1
Does LeWorldModel Preserve Local Action Geometry? — arXiv:2610.31716
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.31716
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-26T20:11:13.276Z
- SHA-256 of the archived text
691668033c092e2cee8619b4ff28d26ea37c430e9c54b2f1dce10a552b274eca- Stable URL of this page
https://metraipolis.com/documento/dc_0183aaf41da1
How to cite
Does LeWorldModel Preserve Local Action Geometry? — arXiv:2610.31716 — arXiv (en), https://arxiv.org/abs/2610.31716 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_0183aaf41da1 (Metraipolis archive dc_0183aaf41da1, sha256 691668033c092e2cee8619b4ff28d26ea37c430e9c54b2f1dce10a552b274eca)
BibTeX
@misc{dc0183aaf41da1,
title = {Does LeWorldModel Preserve Local Action Geometry? — arXiv:2610.31716},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.31716},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_0183aaf41da1; sha256 691668033c092e2cee8619b4ff28d26ea37c430e9c54b2f1dce10a552b274eca},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.31716. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
We examine how closely pretrained LeWorldModel (LeWM) reproduces the effects of small action changes. Using matched simulator and model responses, we measure action sensitivity with Gram matrices at three perturbation scales. Across nearly a thousand states from Reacher and OGBench Cube, the most sensitive action directions differ by median angles of about $36^\circ$ and $42^\circ$, respectively, at the smallest scale. Increasing the perturbation tenfold reduces median normalized Gram discrepancy by roughly 70\% on Reacher and 47\% on Cube. Geometry measured in pixels and embeddings varies more across scales than the chosen physical-state geometry. On held-out directions, environment Grams achieve median embedding response errors of 17\% and 28\%, compared with 26\% and 38\% for an isotropic reference with the same trace. LeWM's errors are higher at this scale. These findings support scale-specific evaluation of action sensitivity; its effect on control remains an open question. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno). Categorie arXiv: Machine Learning (cs.LG) ; Robotics (cs.RO). Pagina del lavoro: https://arxiv.org/abs/2610.31716 — 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 2026-10-26T19:49:03.350Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).