跳到正文
arXiv cs.AI· Haoyang Liu, Kecheng Liang, Xinyu Li, Weixing Chen, Yang Liu·· 23 小时前AI 评分31

PhysScope:用 grounded 视觉记忆做智能体多模态物理推理

PhysScope: Agentic Multimodal Physics Reasoning with Grounded Visual Memory

AI 导读

PhysScope 是一个推理时视觉框架,不更新模型权重即可让 VLM 把物理适用范围与视觉观察绑定,并对智能体提出的修正做来源与范围校验,被接受的修正会使相关依赖失效以维护范围感知工作记忆。

正文 · 原文

View PDF HTML (experimental)

Abstract:Vision-language models (VLMs) can correctly interpret physics diagrams yet apply constraints outside the states or conditions in which they hold, causing premise errors to propagate through subsequent calculations. We introduce PhysScope, an inference-time visual harness that supports self-correction of these scope mismatches without updating model weights. PhysScope enables the model to link physical scopes directly to its visual observations and explicitly constructs derived views to support reasoning. The harness imposes explicit constraints on state distinctions, conditions, and physical scopes during model reasoning, and checks agent-proposed revisions against sources and scopes. Accepted revisions invalidate registered dependents, thereby maintaining scope-aware working memory. On PhysAlign, PhysScope achieves higher solving scores than direct solving across all six evaluated VLMs. With Gemini-3.8-flash, it outperforms Canvas-CoT by 9.34 percentage points and substantially improves solving accuracy on the out-of-distribution (OOD) PhysUniBench benchmark. Ablation studies support explicit scope management and dependency maintenance as design principles for multimodal physics reasoning agents.
Subjects: Artificial Intelligence (cs.AI)
Cite as: arXiv:2611.23010 [cs.AI]
  (or arXiv:2611.23010v1 [cs.AI] for this version)
  https://doi.org/10.48550/arXiv.2611.23010

arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Haoyang Liu [view email]
[v1] Tue, 17 Nov 2026 16:22:21 UTC (1,330 KB)

来源:arXiv cs.AI · arxiv.org