Article

Nanjing University, JD.com and others open-source VideoJev: 73.4% on LVBench, about 1.5 seconds per question

Nanjing University, JD.com and other institutions have open-sourced VideoJev, which unifies video temporal grounding and multiple-choice video question answering into a candidate scoring task, enabling multimodal large models to directly output structured judgments.

Part of the event: Nanjing University and others release open-source VideoJev video understanding model

This page provides a summary and source link only.