Metraipolis › archive › dc_3ca10242221f
DyMD: Preserving Interaction Dynamics through Distribution Matching Distillation in Few-Step Video World Models — arXiv:2609.31349
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.31349
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-08T01:39:53.217Z
- SHA-256 of the archived text
7e905eedb9265490c5a1167bc393ce17229e6ad4660565c031b7e8cc16ce47d0- Stable URL of this page
https://metraipolis.com/documento/dc_3ca10242221f
How to cite
DyMD: Preserving Interaction Dynamics through Distribution Matching Distillation in Few-Step Video World Models — arXiv:2609.31349 — arXiv (en), https://arxiv.org/abs/2609.31349 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_3ca10242221f (Metraipolis archive dc_3ca10242221f, sha256 7e905eedb9265490c5a1167bc393ce17229e6ad4660565c031b7e8cc16ce47d0)
BibTeX
@misc{dc3ca10242221f,
title = {DyMD: Preserving Interaction Dynamics through Distribution Matching Distillation in Few-Step Video World Models — arXiv:2609.31349},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.31349},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_3ca10242221f; sha256 7e905eedb9265490c5a1167bc393ce17229e6ad4660565c031b7e8cc16ce47d0},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.31349. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Large video diffusion models offer expressive priors for embodied prediction and learning, yet their many-step sampling remains costly for interactive downstream use. Distribution Matching Distillation (DMD) enables few-step video generation, but can suppress robot--object motion while preserving visual quality. Examining DMD's teacher and fake-score signals, we find that weak re-noising keeps the teacher posterior concentrated near motion-deficient rollouts, limiting motion-restoring guidance. Meanwhile, stronger-motion rollouts tend to incur larger fake-score fitting errors, which can hinder the generator's learning of interaction dynamics. We propose DyMD, a DMD framework that adapts both teacher supervision and critic fitting to the evolving student. Temporal affinity--conditioned re-noise sampling adapts the timestep distribution to each rollout's current interaction fidelity by mixing the base schedule with a teacher prior motivated by local posterior variation, thereby balancing motion recovery and appearance refinement. To better track stronger-motion rollouts, dynamics-guided fake-score tracking uses a noise-conditioned predictor to estimate noise-relative fitting difficulty from latent temporal dynamics, then upweights predicted-hard rollouts in the critic loss. Using DyMD, we distill a 14B teacher into a four-step 1.3B student with no auxiliary modules at inference. On embodied-video benchmarks, the student improves R-Bench task adherence by $9.6$ percentage points and PAI-Bench-G Domain score by $5.1$ points over Base DMD while maintaining comparable visual quality. As a backbone for downstream action planning, our student achieves 34% mean success across two WorldArena tasks, compared with 16% for Base DMD. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-28 in cs.AI). Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2609.31349 — 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 2026-10-08T01:12:14.579Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-28?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).