ChronoVision：通过潜在状态重建实现时间推理

AI 摘要

ChronoVision是一个多模态框架，旨在通过潜在状态重建提升多步时间推理能力。它采用重建视觉头和ROI注意力定位模块进行监督微调，并在后训练中应用强化学习。在Vbvr-VQA数据集上达到74.8%域内和71.6%域外准确率，在IntPhys2上达到55.0%。

正文节选

ChronoVision: Temporal Reasoning via Latent State Reconstruction Abstract Multimodal large language models excel at passive perception but struggle with complex visual cognitive tasks requiring multi-step temporal reasoning. This degradation largely stems from the inherent ambiguity of language-based reasoning, which often fails to accurately articulate continuous visual transformations. To address this, we propose ChronoVision, a multimodal framework designed to align visual logic with latent i

发布时间：—

抓取时间：2026-08-07 10:12

来源机构：Hugging Face