---
dataset_info:
  features:
  - name: file_name
    dtype: string
  - name: image
    dtype: image
  - name: refs
    sequence: string
  - name: mt
    dtype: string
  - name: human_score
    dtype: float64
  splits:
  - name: train
    num_bytes: 4783117925.25
    num_examples: 26382
  download_size: 4770954560
  dataset_size: 4783117925.25
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---