提出噪声耦合的双分支回溯，实现可验证的反事实视频生成。
Twin Rollouts: Noise-Coupled Counterfactual Branching in Interactive Video World Models
- 共享前缀和未来噪声，仅在干预点后改变动作流生成反事实分支。
- 利用模拟器真值作为奖励，实现无需学习判别器的反事实验证。
- 适用于需要精准因果推断的交互式视频建模与行为分析场景。
交互式视频世界模型在动作序列驱动下自回归生成轨迹，但训练与评估几乎仅限于事实预测。本文研究生成过程中的反事实生成：给定模型自身生成的轨迹，若从第 t* 步起动作不同，会发生什么？我们提出噪声耦合的双分支回溯——一个事实分支与一个反事实分支共享已生成前缀和未来的外生噪声序列，仅在干预点之后的动作流上分叉。由于事实分支是自生成的，其外生噪声完全已知，根据Pearl反事实推理的抽象步骤可精确构造，规避了基于编辑方法的近似逆问题。噪声耦合进一步使最小变化原则成为可逐样本验证的性质：我们定义了一个时空局部性度量，惩罚干预因果后代之外的偏差，该度量可基于模拟器真值计算，无需学习判别器。在干预点 t* 处分叉模拟器状态，可获得反事实重渲染的真值，用于后训练阶段的可验证奖励。本文建立形式框架、度量定义与定位；实验将在后续开展。
原文摘要 · Abstract (English)
Interactive video world models generate rollouts autoregressively under an action stream, yet they are trained and evaluated almost exclusively on factual prediction. We study counterfactual generation inside the rollout: given a trajectory the model has itself generated, what would have happened had the actions differed from step t* onward? We formalize noise-coupled twin rollouts --- a factual and a counterfactual branch sharing the generated prefix and the future exogenous noise sequence, diverging only in the action stream at an intervention point. Because the factual branch is self-generated, its exogenous noise is known exactly: the abduction step of Pearl's counterfactual procedure is exact by construction, sidestepping the approximate-inversion problem faced by editing-based pipelines. Noise coupling further turns the minimal-change principle into a per-sample verifiable property: we define a spatiotemporal locality metric that penalizes divergence outside the causal descendants of the intervention, computable against simulator ground truth without a learned judge. Forking the simulator state at t* yields ground-truth counterfactual re-renders, which we use as verifiable rewards for post-training. This note establishes the formal framework, metric definitions, and positioning; experiments are forthcoming.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。