发表机构
New York University; Tulane University; UIUC; Princeton University; MIT; Columbia University; Stanford University（纽约大学; 杜兰大学; 伊利诺伊大学厄巴纳-香槟分校; 普林斯顿大学; 麻省理工学院; 哥伦比亚大学; 斯坦福大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
DeepJEPA提出权重共享的联合嵌入预测世界模型，将转换深度作为内部扩展轴，在决策关键事件处分配计算，以更少的更新提升视觉控制规划性能。
AI 中文摘要
世界模型规划器通常通过向外扩展来提升性能，例如滚动更远的距离、采样更多轨迹或优化更长的规划，同时为每个想象的转换分配相同的计算量。我们证明，使每个转换均匀地更深会浪费计算，并可能降低规划质量，因为有用的细化集中在少数决策关键事件上。我们引入了DeepJEPA，一种权重共享的联合嵌入预测世界模型，它将转换深度视为内部测试时扩展轴，并学习对于每个候选和滚动步骤，何时值得计算一次额外的循环更新。在五个视觉控制设置中，DeepJEPA在平均每次转换仅进行1.00-1.26次更新的情况下，优于或匹配最强的固定深度规划器。其额外计算集中在接触起始和持续物体交互上，在这些地方，潜在修正可以改变哪些候选进入规划器的精英集以及选择哪个动作。表征探针进一步表明，改进的规划并不需要均匀更好的物体状态可解码性。因此，DeepJEPA将世界模型扩展重新定义为分配内部计算以改变规划器决策的问题：在决策关键转换处更深入地思考，而不是使每次滚动均匀更深或更长。
英文摘要
World-model planners typically scale outward by rolling farther, sampling more trajectories, or optimizing longer, while assigning the same computation to every imagined transition. We show that making every transition uniformly deeper wastes computation and can degrade planning because useful refinement is concentrated at a small set of decision-critical events. We introduce DeepJEPA, a weight-tied joint-embedding predictive world model that treats transition depth as an inner test-time scaling axis and learns when another recurrent update is worth computing for each candidate and rollout step. Across five visual-control settings, DeepJEPA improves or matches the strongest fixed-depth planner while averaging only 1.00-1.26 updates per transition. Its additional computation concentrates at contact onset and sustained object interaction, where latent corrections can change which candidates enter the planner's elite set and which action is selected. Representation probes further show that improved planning does not require uniformly better object-state decodability. DeepJEPA therefore reframes world-model scaling as a problem of allocating internal computation where it can change the planner's decision: think deeper at decision-critical transitions instead of making every rollout uniformly deeper or longer.
CommentsProject page: https://deepjepa.github.io/
