Video-MMMU

The 'Video-MMMU' dataset/benchmark contains video-based visual question answering tasks designed to evaluate multi-modal reasoning capabilities in models.

Papers using Video-MMMU (4)

Xiaomi Mimo-vl-miloco Technical Report2025ProcessThinker: Enhancing Multi-modal Large Language Models Reasoning via Rollout-based Process Reward2026Beyond Single-Sample: Reliable Multi-Sample Distillation for Video Understanding2026Video-MMLU: A Massive Multi-Discipline Lecture Understanding Benchmark2025