VISTA: Mitigating Semantic Inertia in Video-LLMs via Training-Free Dynamic Chain-of-Thought Routing

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Jin, Hongbo, Ding, Jiayu, Xie, Siyi, Luo, Guibo, Li, Ge
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915713498742784
author Jin, Hongbo
Ding, Jiayu
Xie, Siyi
Luo, Guibo
Li, Ge
author_facet Jin, Hongbo
Ding, Jiayu
Xie, Siyi
Luo, Guibo
Li, Ge
contents Recent advancements in Large Language Models have successfully transitioned towards System 2 reasoning, yet applying these paradigms to video understanding remains challenging. While prevailing research attributes failures in Video-LLMs to perceptual limitations, our empirical analysis reveals a cognitive misalignment termed Semantic Inertia, where models suppress valid visual evidence in favor of dominant language priors. To rectify this, we propose VISTA, a training-free framework designed to align perception with logical deduction. By dynamically routing inference paths and materializing implicit visual features into explicit textual anchors, our approach effectively counterbalances the influence of parametric knowledge. Furthermore, we incorporate a Latent Reasoning Consensus mechanism to mitigate stochastic hallucinations. VISTA showed outstanding results on a wide range of benchmarks, and outperforms its base model by 9.3% on Egochema and 5.6% on VideoEspresso, rivalling or even surpassing larger and proprietary models. Our codebase will be publicly available soon.
format Preprint
id arxiv_https___arxiv_org_abs_2505_11830
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle VISTA: Mitigating Semantic Inertia in Video-LLMs via Training-Free Dynamic Chain-of-Thought Routing
Jin, Hongbo
Ding, Jiayu
Xie, Siyi
Luo, Guibo
Li, Ge
Computer Vision and Pattern Recognition
Artificial Intelligence
Recent advancements in Large Language Models have successfully transitioned towards System 2 reasoning, yet applying these paradigms to video understanding remains challenging. While prevailing research attributes failures in Video-LLMs to perceptual limitations, our empirical analysis reveals a cognitive misalignment termed Semantic Inertia, where models suppress valid visual evidence in favor of dominant language priors. To rectify this, we propose VISTA, a training-free framework designed to align perception with logical deduction. By dynamically routing inference paths and materializing implicit visual features into explicit textual anchors, our approach effectively counterbalances the influence of parametric knowledge. Furthermore, we incorporate a Latent Reasoning Consensus mechanism to mitigate stochastic hallucinations. VISTA showed outstanding results on a wide range of benchmarks, and outperforms its base model by 9.3% on Egochema and 5.6% on VideoEspresso, rivalling or even surpassing larger and proprietary models. Our codebase will be publicly available soon.
title VISTA: Mitigating Semantic Inertia in Video-LLMs via Training-Free Dynamic Chain-of-Thought Routing
topic Computer Vision and Pattern Recognition
Artificial Intelligence
url https://arxiv.org/abs/2505.11830