---
dataset_info:
  features:
  - name: data_source
    dtype: string
  - name: prompt
    list:
    - name: content
      dtype: string
    - name: role
      dtype: string
  - name: reward_model
    struct:
    - name: ground_truth
      dtype: string
  - name: extra_info
    struct:
    - name: detector
      dtype: string
    - name: sample_id
      dtype: string
  splits:
  - name: train
    num_bytes: 128106464
    num_examples: 29700
  - name: validation
    num_bytes: 14160556
    num_examples: 3300
  download_size: 32122706
  dataset_size: 142267020
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: validation
    path: data/validation-*
---