CORTEXA
← Browse
arxivcs.CVcs.CLcs.LG2026-07-24

Small Vision-Language Models Know When They Are Wrong But Cannot Say So: A Two-Model Study of Stated versus Internal Confidence Under Realistic Image Degradation

M M Asif Ferdous

Vision-language models (VLMs) are increasingly deployed on consumer hardware where input images are degraded by compression, camera shake, and poor lighting. In such settings, a reliable uncertainty signal matters more than raw accuracy, because it determines when a system should defer rather than answer. We evaluate two small open-weight VLMs -- Qwen2-VL-2B-Instruct and SmolVLM-Instruct -- across six realistic photographic degradations at three severity levels, comparing two confidence signals: the confidence the model states in natural language, and the model's own mean token probability over its generated answer. Across 3,800 predictions, we find a large and consistent gap. Verbalized confidence in Qwen2-VL is almost constant (mean 0.87-0.90 across all conditions) and detects its own errors at chance level (AUROC 0.39-0.75, typically ~0.50), while internal token probability from the same model separates correct from incorrect answers with AUROC 0.92-0.99. In SmolVLM, verbalized confidence proved largely unobtainable: across three prompt templates, only one of five pilot attempts produced a parseable confidence value, while internal probability again yielded above-chance error detection (AUROC 0.54-0.92). Both models fail in the same place: under severe underexposure, accuracy collapses (0.99->0.22 for Qwen2-VL, 0.97->0.42 for SmolVLM) while both confidence signals barely move, and internal error-detection falls to chance. We conclude that small VLMs encode usable self-knowledge that their verbalized output does not express, that internal probability is therefore the better deferral signal in constrained deployment, and that neither signal should be trusted under severe low-light conditions.

View free PDFSource page

Related papers

arxivcs.CVcs.CLcs.LG2026-07-03

Pathways of Visual Information Flow in Vision-Language Models

Israfel Salazar, Stella Frank, Dan Oneata, Desmond Elliott, Constanza Fierro

We study how visual information is routed in vision-language models (VLMs). Using causal patching on controlled synthetic and natural datasets, we find that models rely on two distinct pathways to solve visual tasks: A direct pathway, where visual information is retained in image…

View free PDFSource page
arxivcs.CVcs.AIcs.CLcs.LGcs.MM2026-07-01

ESC: Emotional Self-Correction for Reliable Vision-Language Models

Tien-Huy Nguyen, Minh-Nhat Nguyen, Nguyen Nhat Huy, Hung Viet Nguyen, Huy Nguyen Minh Nhat, Thanh-Huy Nguyen, et al.

Vision-language models (VLMs) have achieved strong performance across diverse multimodal tasks, yet they remain vulnerable to unreliable reasoning. Existing self-correction methods mitigate these issues but typically rely on post-training or carefully engineered feedback, incurri…

View free PDFSource page
arxivcs.CVcs.CLcs.LG2026-06-26

DataComp-VLM: Improved Open Datasets for Vision-Language Models

Matteo Farina, Vishaal Udandarao, Thao Nguyen, Selim Kuzucu, Maximilian Böther, Andreas Hochlehnert, et al.

Building performant Vision-Language Models (VLMs) requires carefully curating large-scale training datasets, yet the community lacks systematic benchmarks for evaluating such curation strategies. We introduce DataComp for VLMs (DCVLM), a benchmark for controlled data-centric expe…

View free PDFSource page
arxivcs.CVcs.AIcs.CLcs.LG2026-07-17

An Exam for Active Observers

Jiarui Zhang, Muzi Tao, Shangshang Wang, Ollie Liu, Xuezhe Ma, Willie Neiswanger

Human vision is a closed loop: gaze is continuously redirected by intermediate hypotheses rather than a single snapshot. Decades of psychophysics and cognitive science have argued that this active observation is essential for a wide range of tasks. Whether today's multimodal larg…

View free PDFSource page
arxivcs.LGcs.CLcs.CV2026-06-25

Just how sure are you? Improving Verbalized Uncertainty Calibration in Medical VQA

Eren Senoglu, Federico Toschi, Nicolo Brunello, Andrea Sassella, Mark James Carman

Multimodal large language models (MLLMs) applied to Medical Visual Question Answering (VQA) tend to produce overconfident outputs regardless of actual correctness, and existing verbalized confidence calibration methods, developed primarily for text only LLMs, do not account for t…

View free PDFSource page