潜在动力学推理实现视频世界模型的外推泛化
原标题：Learning How the World Evolves: Extrapolative Video World Models via Latent Dynamics Reasoning
AI 摘要
Latent Dynamics Reasoning (LDR) 将运动学动力学集成到结构化潜在空间中，使视频世界模型能够远超训练分布地外推物理规律，同时参数减少26倍、推理速度快143倍。在PhyWorld基准的五个任务上，LDR的分布内外误差差距比视频扩散基线小20倍以上，并能泛化到严重分布偏移场景。这是首个能外推学习动力学至训练分布之外的视频世界模型。
正文节选
Learning How the World Evolves: Extrapolative Video World Models via Latent Dynamics Reasoning Abstract Latent Dynamics Reasoning integrates kinematic dynamics in structured latent space to enable video world models that extrapolate physical laws far beyond training distributions with far fewer parameters and faster inference. The world evolves following its dynamics, i.e., its laws of motion. However, leading video diffusion models largely fit the pixels without modeling how the pixels transit