SCOPE：用于视频世界模型的分数隔离智能体优化

中文摘要 AI 辅助

视频世界模型越来越多地被用作规划和具身决策的模拟器，但在推理时改进它们会带来一个微妙的评估问题：提示、采样器、验证器和选择器可能会共同演变，使得难以归因于收益或防止保留的反馈塑造最终策略。我们引入SCOPE（Score-Isolated Agentic Optimization，分数隔离智能体优化），这是一种用于冻结视频世界模型的可审计推理时适应的框架。SCOPE将外部控制表示为类型化状态，仅通过开发证据支持的有界变化来更新此状态，并在保留评估之前冻结生成的策略。在Physics-IQ基准上，SCOPE比精确冻结的基线提高了+14.24（95%置信区间[+8.10,+21.23]）。受控的 ablation（ ablation 指 ablation study，即消融实验）进一步确定了场景指定、采样和学习选择带来的收益，而与最强匹配的智能体基线之间的差距仍未解决。跨骨干网络和前瞻性评估揭示了一个互补结果：有用的推理时更新确实存在，但它们的益处不会在模型和设置之间均匀转移。总之，这些发现表明，可靠的推理时适应不仅需要更好的提案，还需要一种原则性机制来决定哪些更新应成为已部署系统的一部分。代码可在此URL获取。

英文摘要

Video world models are increasingly used as simulators for planning and embodied decision making, yet improving them at inference time introduces a subtle evaluation problem: prompts, samplers, verifiers, and selectors may evolve together, making it difficult to attribute gains or prevent held-out feedback from shaping the final policy. We introduce \scope (\emph{\scopefullname}), a framework for auditable inference-time adaptation of frozen video world models. \scope represents external controls as a typed state, updates this state only through bounded changes supported by development evidence, and freezes the resulting policy before held-out evaluation. On Physics-IQ benchmark, \scope improves over the exact frozen base by $+14.24$ (95\% CI $[+8.10,+21.23]$). Controlled ablations further identify gains from scene specification, sampling, and learned selection, while the margin over the strongest matched agentic baseline remains unresolved. Cross-backbone and prospective evaluations reveal a complementary result: useful inference-time updates exist, but their benefits do not transfer uniformly across models and settings. Together, these findings suggest that reliable inference-time adaptation requires not only better proposals, but also a principled mechanism for deciding which updates should become part of the deployed system. Code is available at https://github.com/YuhuaJiang2002/SCOPE.

发表机构

• Tsinghua University（清华大学）
• National University of Singapore（新加坡国立大学）
• Tencent（腾讯）

机构由 AI 辅助整理，请以论文原文为准。