---
dataset_info:
  features:
  - name: messages
    list:
    - name: role
      dtype: string
    - name: content
      dtype: string
  - name: meta
    struct:
    - name: source
      dtype: string
    - name: source_index
      dtype: int64
  splits:
  - name: stage1_train
    num_bytes: 3379234
    num_examples: 1000
  - name: stage2_train
    num_bytes: 40320432
    num_examples: 4000
  - name: unused_seed
    num_bytes: 75084016
    num_examples: 7500
  download_size: 118785738
  dataset_size: 118783682
configs:
- config_name: default
  data_files:
  - split: stage1_train
    path: data/stage1_train-*
  - split: stage2_train
    path: data/stage2_train-*
  - split: unused_seed
    path: data/unused_seed-*
---