A Very Big Video Reasoning Suite

Papers with Code Trending 02/23/26, 06:59 PM Papers

Summary

This paper introduces the Very Big Video Reasoning (VBVR) dataset and benchmark, a large-scale resource with over one million video clips across 200 reasoning tasks, enabling systematic study of spatiotemporal reasoning and showing early signs of emergent generalization.

Rapid progress in video models has largely focused on visual quality, leaving their reasoning capabilities underexplored. Video reasoning grounds intelligence in spatiotemporally consistent visual environments that go beyond what text can naturally capture, enabling intuitive reasoning over spatiotemporal structure such as continuity, interaction, and causality. However, systematically studying video reasoning and its scaling behavior is hindered by the lack of large-scale training data. To address this gap, we introduce the Very Big Video Reasoning (VBVR) Dataset, an unprecedentedly large-scale resource spanning 200 curated reasoning tasks following a principled taxonomy and over one million video clips, approximately three orders of magnitude larger than existing datasets. We further present VBVR-Bench, a verifiable evaluation framework that moves beyond model-based judging by incorporating rule-based, human-aligned scorers, enabling reproducible and interpretable diagnosis of video reasoning capabilities. Leveraging the VBVR suite, we conduct one of the first large-scale scaling studies of video reasoning and observe early signs of emergent generalization to unseen reasoning tasks. Together, VBVR lays a foundation for the next stage of research in generalizable video reasoning. The data, benchmark toolkit, and models are publicly available at https://video-reason.com/ .

Original Article

View Cached Full Text

Cached at: 05/26/26, 12:36 PM

Paper page - A Very Big Video Reasoning Suite

Source: https://huggingface.co/papers/2602.20159 Published on Feb 23

#1 Paper of the day Authors:

Abstract

A large-scale video reasoning dataset and benchmark are introduced to study video intelligence capabilities beyond visual quality, enabling systematic analysis of spatiotemporal reasoning and generalization across diverse tasks.

Rapid progress in video models has largely focused on visual quality, leaving their reasoning capabilities underexplored.Video reasoninggrounds intelligence in spatiotemporally consistent visual environments that go beyond what text can naturally capture, enabling intuitive reasoning over spatiotemporal structure such as continuity, interaction, and causality. However, systematically studyingvideo reasoningand its scaling behavior is hindered by the lack of large-scale training data. To address this gap, we introduce the Very BigVideo Reasoning(VBVR) Dataset, an unprecedentedly large-scale resource spanning 200 curated reasoning tasks following a principled taxonomy and over one million video clips, approximately three orders of magnitude larger than existing datasets. We further present VBVR-Bench, a verifiable evaluation framework that moves beyond model-based judging by incorporating rule-based, human-aligned scorers, enabling reproducible and interpretable diagnosis ofvideo reasoningcapabilities. Leveraging the VBVR suite, we conduct one of the first large-scale scaling studies ofvideo reasoningand observe early signs ofemergent generalizationto unseen reasoning tasks. Together, VBVR lays a foundation for the next stage of research in generalizablevideo reasoning. The data, benchmark toolkit, and models are publicly available at https://video-reason.com/ .

View arXiv page View PDF Project page GitHub189auto Add to collection

Get this paper in your agent:

hf papers read 2602\.20159

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