紧凑但移动：循环世界模型中的干预相关几何
Compact but Moving: Intervention-Relevant Geometry in Recurrent World Models
查看机构详情
- College of Intelligent Robotics and Advanced Manufacturing, Fudan University（复旦大学智能机器人与先进制造学院）
机构由 AI 辅助整理，请以论文原文为准。
浏览论文内容
中文总结 AI 辅助
本研究在循环世界模型中证明，紧凑干预结构可随状态移动并嵌入高维动力学，而非固定低维子空间，且能有效保留反事实预测功能。
中文摘要 AI 辅助
学习到的世界模型即使其循环状态是高维的，也可能具有紧凑的干预，但目前尚不清楚这种修正进入模型后会发生什么。我们在一个受控的循环世界模型中研究这个问题，先前的工作在该模型中识别出一个检查点特定的秩4界面，用于一次性反事实速度干预。在自主滚动过程中，该修正迅速离开这个固定的进入子空间。然而，通过将进入方向沿事实循环雅可比链传输而获得的低秩图像，仍然能捕捉到大部分非线性修正。使用切线预测的修正进行重启，能保留大量的反事实未来功能。这种传输/功能模式在独立训练的结构化GRU模型和参数匹配的LSTM（以特权紧凑修正初始化）中反复出现。我们进一步在全循环载体上刻画了一个有限时域未来响应算子。修补将该算子的主要未来敏感方向移向匹配的原生反事实组织，并且局部算子能在修补和原生反事实基点的注册方向面板上准确排序有限扰动效应。一项单独的全振幅测试发现，在三个检查点中的两个，存在显著的事实端点切线残差，并支持响应重构。总之，这些结果表明，紧凑的干预结构可以作为一种移动的、状态相关的局部几何嵌入在高维循环动力学中持续存在，而不意味着存在一个固定或动态封闭的低维状态。
英文摘要
Learned world models may have compact interventions even when their recurrent state is high-dimensional, but it is unclear what happens to such a correction after it enters the model. We study this question in a controlled recurrent world model where prior work identified a checkpoint-specific rank-4 interface for one-shot counterfactual velocity interventions. The correction rapidly leaves this fixed entry subspace during autonomous rollout. Nevertheless, a low-rank image obtained by transporting the entry directions through the factual recurrent Jacobian chain continues to capture most of the nonlinear correction. Restarts using the tangent-predicted correction preserve substantial counterfactual future function. This transport/function pattern recurs across independently trained structured-GRU models and a parameter-matched LSTM initialized with a privileged compact correction. We further characterize a finite-horizon future-response operator over the full recurrent carrier. Patching shifts its leading future-sensitive directions toward the matched native-counterfactual organization, and the local operator accurately ranks finite perturbation effects over the registered direction panels at the patched and native-counterfactual basepoints. A separate full-amplitude assay finds substantial factual-endpoint tangent residuals and supports response reconfiguration in two of three checkpoints. Together, these results show that compact intervention structure can persist as a moving, state-dependent local geometry embedded in high-dimensional recurrent dynamics, without implying a fixed or dynamically closed low-dimensional state.
