name: Video-MME-v2
description: Video-MME-v2 is a benchmark for evaluating the next stage of video understanding capabilities of multimodal large language models.
evaluation_framework: video-mme-v2

tasks:
  - id: video-mme-v2
    config: default
    split: test