---
dataset_info:
  features:
  - name: instruction
    dtype: string
  - name: score_prompt
    dtype: string
  - name: labels
    struct:
    - name: IDENTITY_ATTACK
      dtype: float64
    - name: INSULT
      dtype: float64
    - name: PROFANITY
      dtype: float64
    - name: SEVERE_TOXICITY
      dtype: float64
    - name: THREAT
      dtype: float64
    - name: TOXICITY
      dtype: float64
  splits:
  - name: train
    num_bytes: 34710845
    num_examples: 16246
  download_size: 4745456
  dataset_size: 34710845
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---