behavior-in-the-wild's picture
Upload dataset
0bd02e7 verified
|
raw
history blame
1.77 kB
metadata
dataset_info:
  features:
    - name: video_id
      dtype: string
    - name: asr_raw
      list:
        - name: start
          dtype: float64
        - name: end
          dtype: float64
        - name: text
          dtype: string
        - name: words
          list:
            - name: confidence
              dtype: float64
            - name: start
              dtype: float64
            - name: end
              dtype: float64
            - name: text
              dtype: string
    - name: asr_grouped
      list:
        list: string
    - name: ocr
      list:
        list: string
    - name: blip2_annotations
      struct:
        - name: actions
          list: string
        - name: captions
          list: string
        - name: objects
          list: string
    - name: replay_graphs
      struct:
        - name: original_marker_duration
          dtype: float64
        - name: processed_marker_duration
          dtype: float64
        - name: multiplier
          dtype: float64
        - name: markers
          list:
            - name: start
              dtype: float64
            - name: end
              dtype: float64
            - name: replay_score
              dtype: float64
    - name: likes
      dtype: float64
    - name: views
      dtype: float64
    - name: metadata
      struct:
        - name: title
          dtype: string
        - name: description
          dtype: string
        - name: length
          dtype: float64
        - name: date
          dtype: string
    - name: channel_data
      struct:
        - name: channel_id
          dtype: string
        - name: company_name
          dtype: string
        - name: subscribers
          dtype: float64
  splits:
    - name: train
      num_bytes: 396758465
      num_examples: 22569
    - name: test
      num_bytes: 35343326
      num_examples: 2026
  download_size: 135245985
  dataset_size: 432101791
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
      - split: test
        path: data/test-*