Cover Image

Seeing but Not Believing: Probing the Disconnect Between Visual Attention and Answer Correctness in VLMs

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Liu, Zhining, Chen, Ziyi, Liu, Hui, Luo, Chen, Tang, Xianfeng, Wang, Suhang, Zeng, Joy, Dai, Zhenwei, Shi, Zhan, Wei, Tianxin, Dumoulin, Benoit, Tong, Hanghang
Format:	Preprint
Published:	2025
Subjects:	Artificial Intelligence Computer Vision and Pattern Recognition
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

CATS: Mitigating Correlation Shift for Multivariate Time Series Classification
by: Lin, Xiao, et al.
Published: (2025)

SelfElicit: Your Language Model Secretly Knows Where is the Relevant Evidence
by: Liu, Zhining, et al.
Published: (2025)

Adaptive Test-Time Reasoning via Reward-Guided Dual-Phase Search
by: Cui, Yingqian, et al.
Published: (2025)

WAPITI: A Watermark for Finetuned Open-Source LLMs
by: Chen, Lingjie, et al.
Published: (2024)

Matcha: Mitigating Graph Structure Shifts with Test-Time Adaptation
by: Bao, Wenxuan, et al.
Published: (2024)

Position: Agentic Evolution is the Path to Evolving LLMs
by: Lin, Minhua, et al.
Published: (2026)

Right this way: Can VLMs Guide Us to See More to Answer Questions?
by: Liu, Li, et al.
Published: (2024)

Hierarchical Multi-Marginal Optimal Transport for Network Alignment
by: Zeng, Zhichen, et al.
Published: (2023)

AIM: Attributing, Interpreting, Mitigating Data Unfairness
by: Liu, Zhining, et al.
Published: (2024)

TRAJECT-Bench:A Trajectory-Aware Benchmark for Evaluating Agentic Tool Use
by: He, Pengfei, et al.
Published: (2025)

BACKTIME: Backdoor Attacks on Multivariate Time Series Forecasting
by: Lin, Xiao, et al.
Published: (2024)

To trust or not to trust: Attention-based Trust Management for LLM Multi-Agent Systems
by: He, Pengfei, et al.
Published: (2025)

Believing vs. Achieving -- The Disconnect between Efficacy Beliefs and Collaborative Outcomes
by: Spitzer, Philipp, et al.
Published: (2026)

Breaking Silos: Adaptive Model Fusion Unlocks Better Time Series Forecasting
by: Liu, Zhining, et al.
Published: (2025)

CLIMB: Class-imbalanced Learning Benchmark on Tabular Data
by: Liu, Zhining, et al.
Published: (2025)

How Far are LLMs from Real Search? A Comprehensive Study on Efficiency, Completeness, and Inherent Capabilities
by: Lin, Minhua, et al.
Published: (2025)

Seeing is Believing? Mitigating OCR Hallucinations in Multimodal Large Language Models
by: He, Zhentao, et al.
Published: (2025)

THeGCN: Temporal Heterophilic Graph Convolutional Network
by: Yan, Yuchen, et al.
Published: (2024)

Do VLMs Have a Moral Backbone? A Study on the Fragile Morality of Vision-Language Models
by: Liu, Zhining, et al.
Published: (2026)

AdaFuse: Adaptive Ensemble Decoding with Test-Time Scaling for LLMs
by: Cui, Chengming, et al.
Published: (2026)

Flow Matching Meets Biology and Life Science: A Survey
by: Li, Zihao, et al.
Published: (2025)

Seeing Isn't Knowing: Do VLMs Know When Not to Answer Spatial Questions (and Why)?
by: Zhang, Yue, et al.
Published: (2026)

ResMoE: Space-efficient Compression of Mixture of Experts LLMs via Residual Restoration
by: Ai, Mengting, et al.
Published: (2025)

AgentTTS: Large Language Model Agent for Test-time Compute-optimal Scaling Strategy in Complex Tasks
by: Wang, Fali, et al.
Published: (2025)

Mem-Gallery: Benchmarking Multimodal Long-Term Conversational Memory for MLLM Agents
by: Bei, Yuanchen, et al.
Published: (2026)

Seeing is Believing: Vision-driven Non-crash Functional Bug Detection for Mobile Apps
by: Liu, Zhe, et al.
Published: (2024)

A General Framework to Enhance Fine-tuning-based LLM Unlearning
by: Ren, Jie, et al.
Published: (2025)

How Do Latent Reasoning Methods Perform Under Weak and Strong Supervision?
by: Cui, Yingqian, et al.
Published: (2026)

Seeing Is Believing: The Role of Videoconferencing in Distance Learning
by: Martin, Marie
Published: (2005)

Firefly: Illuminating Large-Scale Verified Tool-Call Data Generation from Real APIs
by: Lu, Yuxuan, et al.
Published: (2026)

Hierarchical LoRA MoE for Efficient CTR Model Scaling
by: Zeng, Zhichen, et al.
Published: (2025)

Saffron-1: Safety Inference Scaling
by: Qiu, Ruizhong, et al.
Published: (2025)

PLANETALIGN: A Comprehensive Python Library for Benchmarking Network Alignment
by: Yu, Qi, et al.
Published: (2025)

See What LLMs Cannot Answer: A Self-Challenge Framework for Uncovering LLM Weaknesses
by: Chen, Yulong, et al.
Published: (2024)

Beyond Seeing Is Believing: On Crowdsourced Detection of Audiovisual Deepfakes
by: Soprano, Michael, et al.
Published: (2026)

Is Seeing Believing? Evaluating Human Sensitivity to Synthetic Video
by: Wegmann, David, et al.
Published: (2026)

Cite Before You Speak: Enhancing Context-Response Grounding in E-commerce Conversational LLM-Agents
by: Zeng, Jingying, et al.
Published: (2025)

Seeing is Believing: Mitigating Hallucination in Large Vision-Language Models via CLIP-Guided Decoding
by: Deng, Ailin, et al.
Published: (2024)

How Far Are LLMs from Professional Poker Players? Revisiting Game-Theoretic Reasoning with Agentic Tool Use
by: Lin, Minhua, et al.
Published: (2026)

Seeing is Believing, but How Much? A Comprehensive Analysis of Verbalized Calibration in Vision-Language Models
by: Xuan, Weihao, et al.
Published: (2025)