---
dataset_info:
  features:
  - name: refs
    sequence: string
  - name: cand
    dtype: string
  - name: img
    dtype: image
  - name: human_score
    dtype: float32
  splits:
  - name: train
    num_bytes: 7927109601.66902
    num_examples: 41705
  - name: validation
    num_bytes: 2752838125.1393332
    num_examples: 13944
  - name: test
    num_bytes: 2701536050.1612816
    num_examples: 14001
  download_size: 13139167235
  dataset_size: 13381483776.969635
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: validation
    path: data/validation-*
  - split: test
    path: data/test-*
---