arXiv：cs.AI· Shinseong Kim, Jaeseok Jeong, Jianlong Fu, Youngjung Uh·· 16 小时前
World Models 需要物理注册表：为物理合理世界模型学习内部预测状态
World Models Need a Physical Registry: Learning Internal Predictive States for Physically Plausible World Models
AI 导读
研究提出 Physical Registry，在视频 DiT 内部设置与视频 token 共同演化的注册槽，用观测条件预测初始化并接受未来表征监督，使预测动态成为生成器可读的显式状态。
正文
Abstract:Video world models must generate physically plausible futures, not just realistic frames. Existing approaches typically use semantic, physical, or predictive information to guide the visual generation stream, leaving a fundamental question: where inside the generator should predictions of scene evolution live? Standard video tokens jointly encode appearance and dynamics, but provide no distinct state for learning and reading predictive dynamics. We answer this question with the Physical Registry: temporally grounded registry slots inside a video DiT that are initialized from an observation-conditioned forecast, co-evolve with video tokens, and receive future-representation supervision, making predictive dynamics an explicit, readable state of the generator. In a controlled comparison of predictive-state interfaces on Physics-IQ Verified, Physical Registry improves vanilla fine-tuning by 9.9/8.7 points on I2V/V2V, the largest gain among the interfaces we compare. Across all Cosmos and WAN backbones, Physical Registry improves Physics-IQ Verified scores. With only 0.45% additional parameters, it enables Cosmos 3 Nano 16B to surpass the published I2V score of Cosmos 3 Super 64B (43.56 vs. 42.7). On WorldModelBench, it also improves physical adherence and the total score for all four evaluated backbones. Because the registry becomes candidate-specific while remaining readable, the Physical Registry Selector (PRS) ranks emerging futures by how far their intermediate registry readouts depart from the shared forecast before full denoising is complete. Across four backbones, this selection improves I2V scores by 5.6-15.9% without completed-video external scoring. Together, these results support a broader principle: dynamics prediction should live as an explicit model state that co-evolves with the video representation. Project page: this https URL
| Comments: | 36 pages, 26 figures, 14 tables. Project page: this https URL |
| Subjects: | Artificial Intelligence (cs.AI) |
| ACM classes: | I.2.10; I.4.8 |
| Cite as: | arXiv:2610.16365 [cs.AI] |
| (or arXiv:2610.16365v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.16365 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Shin Seong Kim [view email]
[v1]
Mon, 12 Oct 2026 08:18:05 UTC (8,516 KB)
来源：arXiv：cs.AI · arxiv.org