---
dataset_info:
  features:
  - name: prompt
    dtype: string
  - name: image
    dtype: string
  - name: chosen
    sequence: float64
  - name: rejected
    sequence: float64
  - name: pixel_values
    sequence:
      sequence:
        sequence: float32
  - name: chosen_input_ids
    sequence: int64
  - name: chosen_labels
    sequence: int64
  - name: rejected_input_ids
    sequence: int64
  - name: rejected_labels
    sequence: int64
  splits:
  - name: train
    num_bytes: 30627600339
    num_examples: 25283
  download_size: 7774252430
  dataset_size: 30627600339
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---