Contextualized Evaluation of Vision Language Models through Dynamic, Multi-turn Interactions

• URL: http://arxiv.org/abs/2607.14499v1
• Date: Thu, 16 Jul 2026 02:25:03 GMT
• Title: Contextualized Evaluation of Vision Language Models through Dynamic, Multi-turn Interactions
• Abstract summary: We propose a framework that recasts evaluation as a three-party interaction between an evaluatee model, an automated examiner, and a grader.<n>By navigating state-space transitions, CEDI deploys diverse strategies, from clarification requests to adversarial probes, to elicit performance evidence.<n>We show that hallucinations often accumulate over long contexts, through self-reinforcing dialogue history, and models are particularly vulnerable to questions requiring premise rejection or refusal.
• Score: 15.505991441542372
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Multi-modal Large Language Models (MLLMs) have made substantial advances on benchmarks, yet their real-world effectiveness remains uncertain. This gap stems from the fundamental misalignment between benchmarks in controlled, static settings and the dynamic, interactive, and contextualized nature of real-world applications. To bridge this gap, we propose CEDI (Contextualized Evaluations of MLLMs through Dynamic, multi-round Interactions), a framework that recasts evaluation as a three-party interaction between an evaluatee model, an automated examiner, and a grader. The examiner conducts multi-turn, semi-structured conversation guided by a graph-based representation of the task. By navigating state-space transitions, CEDI deploys diverse strategies, from clarification requests to adversarial probes, to elicit performance evidence. We apply CEDI to visual hallucinations. Empirical results across multiple models, diverse settings, datasets, and domains show that contextualized, interactive evaluations reveal not only significantly more hallucinations than conventional static evaluation but also ones that more closely resemble those arising in practical use cases. We further show that hallucinations often accumulate over long contexts, through self-reinforcing dialogue history, and models are particularly vulnerable to questions requiring premise rejection or refusal. Together, these findings highlight CEDI as a step toward realistic, systematic, and ecologically valid assessments of MLLMs' capabilities. Code is available at github.com/williamium3000/cedi.

Related papers

• Beyond Static Dialogues: Benchmarking Realistic, Heterogeneous, and Evolving Long-Term Memory [54.947805187562274]
  We introduce realistic dialogues across diverse interaction scenarios that exhibit dynamic temporal evolution and long-term coherence.<n>The resulting benchmark encompasses challenging question-answer pairs spanning seven inquiry types.<n>We identify at least one of 27 critical memory characteristics that we identify as essential yet underexplored in current research.
  arXiv  Detail & Related papers  (2026-05-29T09:51:22Z)
• PENDULUM: A Benchmark for Assessing Sycophancy in Multimodal Large Language Models [43.767942065379366]
  Sycophancy is a tendency of AI models to agree with user input at the expense of factual accuracy or in contradiction of visual evidence.<n>We introduce a comprehensive evaluation benchmark, textitPENDULUM, comprising approximately 2,000 human-curated Visual Question Answering pairs.<n>We observe substantial variability in model robustness and a pronounced susceptibility to sycophantic and hallucinatory behavior.
  arXiv  Detail & Related papers  (2025-12-22T12:49:12Z)
• Explaining multimodal LLMs via intra-modal token interactions [55.27436637894534]
  Multimodal Large Language Models (MLLMs) have achieved remarkable success across diverse vision-language tasks, yet their internal decision-making mechanisms remain insufficiently understood.<n>We propose enhancing interpretability by leveraging intra-modal interaction.
  arXiv  Detail & Related papers  (2025-09-26T14:39:13Z)
• KnowDR-REC: A Benchmark for Referring Expression Comprehension with Real-World Knowledge [1.5833270109954136]
  We propose KnowDR-REC, built upon real-world knowledge, requiring fine-grained multimodal reasoning across text and image.<n>We evaluate 16 state-of-the-art multimodal models on KnowDR-REC, with experimental results showing that existing MLLMs still struggle with knowledge-driven visual grounding tasks.
  arXiv  Detail & Related papers  (2025-08-12T19:43:44Z)
• V-MAGE: A Game Evaluation Framework for Assessing Vision-Centric Capabilities in Multimodal Large Language Models [84.27290155010533]
  We introduce Vision-centric Multiple Abilities Game Evaluation (V-MAGE), a novel game-based evaluation framework.<n>V-MAGE features five distinct video games comprising over 30 carefully constructed evaluation scenarios.<n>We show V-MAGE provides actionable insights for improving the visual and reasoning capabilities of MLLMs in dynamic, interactive settings.
  arXiv  Detail & Related papers  (2025-04-08T15:43:01Z)
• Dynamic benchmarking framework for LLM-based conversational data capture [0.0]
  This paper introduces a benchmarking framework to assess large language models (LLMs)<n>It integrates generative agent simulation to evaluate performance on key dimensions: information extraction, context awareness, and adaptive engagement.<n>Results show that adaptive strategies improve data extraction accuracy, especially when handling ambiguous responses.
  arXiv  Detail & Related papers  (2025-02-04T15:47:47Z)
• MMIE: Massive Multimodal Interleaved Comprehension Benchmark for Large Vision-Language Models [71.36392373876505]
  We introduce MMIE, a large-scale benchmark for evaluating interleaved multimodal comprehension and generation in Large Vision-Language Models (LVLMs)<n>MMIE comprises 20K meticulously curated multimodal queries, spanning 3 categories, 12 fields, and 102 subfields, including mathematics, coding, physics, literature, health, and arts.<n>It supports both interleaved inputs and outputs, offering a mix of multiple-choice and open-ended question formats to evaluate diverse competencies.
  arXiv  Detail & Related papers  (2024-10-14T04:15:00Z)
• VALOR-EVAL: Holistic Coverage and Faithfulness Evaluation of Large Vision-Language Models [57.43276586087863]
  Large Vision-Language Models (LVLMs) suffer from hallucination issues, wherein the models generate plausible-sounding but factually incorrect outputs. Existing benchmarks are often limited in scope, focusing mainly on object hallucinations. We introduce a multi-dimensional benchmark covering objects, attributes, and relations, with challenging images selected based on associative biases.
  arXiv  Detail & Related papers  (2024-04-22T04:49:22Z)
• 'What are you referring to?' Evaluating the Ability of Multi-Modal Dialogue Models to Process Clarificational Exchanges [65.03196674816772]
  Referential ambiguities arise in dialogue when a referring expression does not uniquely identify the intended referent for the addressee. Addressees usually detect such ambiguities immediately and work with the speaker to repair it using meta-communicative, Clarification Exchanges (CE): a Clarification Request (CR) and a response. Here, we argue that the ability to generate and respond to CRs imposes specific constraints on the architecture and objective functions of multi-modal, visually grounded dialogue models.
  arXiv  Detail & Related papers  (2023-07-28T13:44:33Z)
• Representation-Centric Survey of Skeletal Action Recognition and the ANUBIS Benchmark [43.00059447663327]
  3D skeleton-based human action recognition has emerged as a powerful alternative to traditional RGB and depth-based approaches.<n>Despite remarkable progress, current research remains fragmented across diverse input representations.<n>ANUBIS is a large-scale, challenging skeleton action dataset designed to address critical gaps in existing benchmarks.
  arXiv  Detail & Related papers  (2022-05-04T14:03:43Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.