---
dataset_info:
  features:
  - name: transcript
    dtype: string
  - name: prompt
    dtype: string
  - name: compress_prompt
    dtype: string
  - name: conversations
    list:
    - name: content
      dtype: string
    - name: role
      dtype: string
  splits:
  - name: train
    num_bytes: 4763539094.0
    num_examples: 1000098
  download_size: 1363655926
  dataset_size: 4763539094.0
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---