TempCloze: Can Video-LLMs Identify the Missing Middle?

Hugging Face Daily Papers Papers

Summary

TempCloze is a video cloze benchmark for evaluating visual temporal reasoning in Video-LLMs, requiring models to identify missing video segments from distractors to reduce linguistic shortcuts in semantics, alignment, and progression.

Temporal reasoning benchmarks for Video-LLMs are often mediated by language, leaving room for linguistic shortcuts from option wording, answer correlations, or language priors. To reduce such shortcuts, we introduce TempCloze, a video cloze benchmark for evaluating visual temporal reasoning in Video-LLMs. Given the beginning and ending clips of a video, models must identify the true missing middle from four candidates. TempCloze contains 1,521 carefully filtered videos from seven sources, mainly long-take and egocentric videos. We construct same-source distractors along three dimensions: Semantic asks what event should happen, Alignment probes when it should occur, and Progression tests how it should unfold, while shared scenes and objects reduce appearance cues. Our evaluation of 10 proprietary and 21 open-source Video-LLMs reveals Alignment as the primary bottleneck: models often recognize plausible semantic content and local event progression but struggle with temporal alignment. We further conduct error pattern and behavioral sensitivity analyses on TempCloze-Mixed and TempCloze-Hard with four representative models to examine where errors arise and how candidate order, context direction, visible span, frame density, and test-time scaling influence model choices.
Original Article
View Cached Full Text

Cached at: 09/11/26, 10:17 AM

Paper page - TempCloze: Can Video-LLMs Identify the Missing Middle?

Source: https://huggingface.co/papers/2609.01515

Abstract

TempCloze evaluates visual temporal reasoning in Video-LLMs by requiring identification of missing video segments from distractors targeting semantics, alignment, and progression.

Temporal reasoningbenchmarks forVideo-LLMsare often mediated by language, leaving room for linguistic shortcuts from option wording, answer correlations, or language priors. To reduce such shortcuts, we introduce TempCloze, avideo clozebenchmark for evaluating visualtemporal reasoninginVideo-LLMs. Given the beginning and ending clips of a video, models must identify the true missing middle from four candidates. TempCloze contains 1,521 carefully filtered videos from seven sources, mainly long-take and egocentric videos. We constructsame-source distractorsalong three dimensions: Semantic asks what event should happen, Alignment probes when it should occur, andProgressiontests how it should unfold, while shared scenes and objects reduce appearance cues. Our evaluation of 10 proprietary and 21 open-sourceVideo-LLMsreveals Alignment as the primary bottleneck: models often recognize plausible semantic content and local eventprogressionbut struggle with temporal alignment. We further conduct error pattern and behavioral sensitivity analyses on TempCloze-Mixed and TempCloze-Hard with four representative models to examine where errors arise and how candidate order, context direction, visible span, frame density, andtest-time scalinginfluence model choices.

View arXiv pageView PDFGitHub5Add to collection

Get this paper in your agent:

hf papers read 2609\.01515

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2609.01515 in a model README.md to link it from this page.

Datasets citing this paper1

#### CedPei/TempCloze Viewer• Updatedabout 16 hours ago • 1.52k

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2609.01515 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

When No Answer Is Correct: Diagnosing Absent Answer Detection for MLLMs in Video Understanding

arXiv cs.AI

This paper studies the ability of multimodal large language models (MLLMs) to detect when the correct answer is absent in video understanding tasks, finding that models systematically fail by selecting plausible distractors instead of recognizing no valid option exists. The failure worsens in temporal reasoning and dense frame sampling, and chain-of-thought prompting only partially mitigates the issue.