发表机构
University College London（伦敦大学学院）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本文提出COSTGRAD，一种无需训练的令牌选择器，基于规划代价梯度范数筛选关键令牌，在50%稀疏度下实现与全令牌规划相当或更优的性能，并带来2.6倍至5倍的加速，同时揭示了选择器与架构兼容性的重要性。
AI 中文摘要
基于令牌的世界模型能够实现细粒度的潜在规划，但反复处理大型空间令牌网格使得动作搜索代价高昂。我们引入了COSTGRAD，一种无需训练、以目标为条件的选择器，它根据每个输入令牌的规划代价梯度范数对空间令牌进行排序。通过从下游控制目标中推导重要性，COSTGRAD针对的是对规划至关重要的令牌，而不仅仅是预测。在AdaLN条件预测器上，在50%稀疏度下，COSTGRAD在四个连续控制基准中的三个上达到或超过全令牌规划，同时每个环境规划步骤实现了实测的2.6倍墙钟加速。将令牌稀疏性与减少的CEM搜索相结合，总加速比提高到约5倍，同时仍超过全令牌基线。我们还识别了一种依赖架构的失败模式：在匹配的AdaLN与拼接对比中，拼接保持了相当的全令牌性能，但纯COSTGRAD失去了相对于随机选择的优势。这种差异与动作路径漂移有关：在AdaLN上，梯度选择的移除比随机移除产生的漂移更小，但在拼接上则更大。这些结果凸显了选择器与架构的兼容性作为稀疏世界模型规划的一个设计轴。项目页面和演示：此https URL。
英文摘要
Token-based world models enable fine-grained latent planning, but repeatedly processing large spatial token grids makes action search expensive. We introduce COSTGRAD, a training-free, goal-conditioned selector that ranks spatial tokens by the gradient norm of the planning cost with respect to each input token. By deriving importance from the downstream control objective, COSTGRAD targets tokens that matter for planning rather than merely for prediction. On AdaLN-conditioned predictors at $50\%$ sparsity, COSTGRAD matches or exceeds full-token planning on three of four continuous-control benchmarks, while giving a measured $2.6\times$ wall-clock speedup per environment planning step. Combining token sparsity with reduced CEM search increases this to a $\sim 5\times$ total speedup while still exceeding the full-token baseline. We also identify an architecture-dependent failure mode: in a matched AdaLN-vs-concat comparison, concat maintains comparable full-token performance but pure COSTGRAD loses its advantage over random selection. This difference tracks action-pathway drift: gradient-selected removal produces less drift than random removal on AdaLN, but more on concat. These results highlight selector-architecture compatibility as a design axis for sparse world-model planning. Project page and demos: https://ycxuyingchen.github.io/costgrad/
CommentsAccepted at NeurIPS 2026. 20 pages, 6 figures, 8 tables. Project page and demos: https://ycxuyingchen.github.io/costgrad/
