CAST：交互式世界模型的重建耦合加速
CAST: Reconstruction-Coupled Acceleration of Interactive World Models
- Shanghai Jiao Tong University（上海交通大学）
机构由 AI 辅助整理，请以论文原文为准。
AI总结：
CAST通过重建耦合推理加速交互式世界模型，利用交互敏感性和相位感知重建，在保持视觉质量的同时实现2.15倍和3.48倍加速，并取得最优VBench分数。
AI中文摘要：
交互式世界模型必须快速响应控制，同时保持场景一致性。现有的加速方法在恢复跳过特征时可能遗漏异构控制响应和空间传输。我们观察到，交互引起的特征变化与近似误差相关，而低频插值误差对相位敏感，并表现出更可预测的相位进展。这些发现促使我们提出CAST，一种重建耦合的推理框架。CAST通过交互敏感性和跨层覆盖选择锚点，使用频率和置信度感知的相位感知重建（PAR）重建跳过的残差，并根据下游重建责任协调历史KV路由。在Matrix-Game 3.0和HY-World 1.5上，CAST分别实现了2.15倍和3.48倍的加速，同时保持接近Native的视觉质量（图1）。它还在比较方法中获得了最高的VBench分数，并在WorldMark的13个维度中的七个和六个上领先于非原生基线，展示了在实时控制下生成速度、视觉质量和交互响应性的平衡。代码可在https URL获取。
英文摘要：
Interactive world models must respond quickly to controls while preserving scene consistency. Existing acceleration methods can miss heterogeneous control responses and spatial transport when recovering skipped features. We observe that interaction-induced feature changes correlate with approximation error, while low-frequency interpolation errors are phase-sensitive and show more predictable phase progression. These findings motivate CAST, a reconstruction-coupled inference framework. CAST selects anchors by interaction sensitivity and cross-layer coverage, reconstructs skipped residuals with frequency- and confidence-aware Phase-Aware Reconstruction (PAR), and coordinates historical KV routing according to downstream reconstruction responsibility. On Matrix-Game 3.0 and HY-World 1.5, CAST achieves 2.15x and 3.48x speedups, respectively, while maintaining visual quality close to Native (Figure 1). It also attains the highest VBench scores among compared methods and leads non-native baselines on seven and six of thirteen WorldMark dimensions, demonstrating a balance of generation speed, visual quality, and interactive responsiveness under real-time control. Code is available at https://github.com/lokiniuniu/CAST.
