ARC-Bench: Closed-Loop Replanning Masks Broken Action Ranking in Frozen JEPA World Models

ARC-Bench：闭环重规划掩盖了冻结 JEPA 世界模型中动作排序失效的问题

Abstract: Reward-free latent world models plan by scoring candidate actions with distances in a frozen latent space: an action is preferred if its predicted future embedding lands closer to the goal embedding. This silently assumes that latent closeness is action-rankable, i.e., that ordering candidates by latent distance agrees with ordering them by true cost. We audit this assumption directly.

摘要： 无奖励潜在世界模型通过在冻结的潜在空间中计算候选动作的距离来进行规划：如果一个动作预测的未来嵌入（embedding）更接近目标嵌入，则该动作被优先选择。这隐含了一个假设，即潜在空间的接近度是可以进行动作排序的，也就是说，根据潜在距离对候选动作进行排序的结果与根据真实成本进行排序的结果是一致的。我们直接对这一假设进行了审计。

We introduce ARC-Bench, a no-leak, fixed-candidate protocol that measures whether frozen JEPA-style objectives rank candidate actions correctly, and apply it to official released JEPA-WM checkpoints across navigation and manipulation-style control. The assumption fails, severely and structurally: on the official manipulation audits the top-scored candidate is almost always suboptimal, and the same inversion appears in the maze domains.

我们引入了 ARC-Bench，这是一个无泄漏、固定候选的协议，用于衡量冻结的 JEPA 风格目标函数是否能正确地对候选动作进行排序，并将其应用于导航和操作控制任务中官方发布的 JEPA-WM 检查点。结果显示，该假设在严重且结构性的层面上失效了：在官方的操作审计中，得分最高的候选动作几乎总是次优的，同样的倒置现象也出现在迷宫领域中。

A controlled visual-backbone extension shows that the defect persists when DINOv2 is replaced by video-pretrained V-JEPA 1 and V-JEPA 2 encoders at ViT-L/ViT-G scale. Provenance, undertraining, matched-budget backbone controls, and metric-circularity controls rule out trivial explanations.

一项受控的视觉骨干网络扩展实验表明，当将 DINOv2 替换为 ViT-L/ViT-G 规模的视频预训练 V-JEPA 1 和 V-JEPA 2 编码器时，该缺陷依然存在。通过对来源、训练不足、匹配预算的骨干网络控制以及指标循环性控制的分析，排除了琐碎的解释。

We then explain why this defect has stayed invisible: closed-loop replanning masks it. When we reduce the planner’s replanning frequency, success collapses in both a navigation and a manipulation domain, and the episodes rescued by frequent replanning are enriched for severe first-plan ranking failures in the PointMaze first-plan diagnostic.

随后，我们解释了为什么这一缺陷一直未被发现：闭环重规划掩盖了它。当我们降低规划器的重规划频率时，导航和操作领域的成功率都会崩溃；而在 PointMaze 的首次规划诊断中，那些通过频繁重规划得以“挽救”的片段，恰恰包含了严重的首次规划排序失败。

Closed-loop success rates therefore systematically overstate the rankability of frozen latent representations. ARC-Bench supplies the measurement, and the masking mechanism the explanation, for methods that adapt, amortize, or replan around latent-space planners without directly auditing released JEPA-WM action rankability.

因此，闭环成功率系统性地高估了冻结潜在表征的可排序性。ARC-Bench 为那些在不直接审计已发布 JEPA-WM 动作排序能力的情况下，通过适应、摊销或围绕潜在空间规划器进行重规划的方法，提供了衡量标准和掩盖机制的解释。