SAGE-Bench/README.md

metadata

dataset_info:
  features:
    - name: video_id
      dtype: string
    - name: gt_answer
      dtype: string
    - name: ques_type
      dtype: string
    - name: difficulty
      dtype: string
    - name: modality
      dtype: string
    - name: video_duration
      dtype: string
    - name: duration_seconds
      dtype: float64
    - name: question
      dtype: string
  splits:
    - name: test
      num_bytes: 598253
      num_examples: 1744
  download_size: 323758
  dataset_size: 598253
configs:
  - config_name: default
    data_files:
      - split: test
        path: data/test-*
license: odc-by
task_categories:
  - video-text-to-text
language:
  - en

License

This dataset is licensed under ODC-BY 1.0. It is intended for research and educational use in accordance with Ai2's Responsible Use Guidelines. This dataset includes synthetic Q&A pairs generated using Gemini 2.5 Flash, which is subject to Google's Terms of Service.