---
dataset_info:
  features:
  - name: prompt
    dtype: string
  - name: prompt_source
    dtype: string
  - name: response_0_source
    dtype: string
  - name: response_1_source
    dtype: string
  - name: is_response_0_safe
    dtype: bool
  - name: is_response_1_safe
    dtype: bool
  - name: response_0_harm_category
    struct:
    - name: Animal Abuse
      dtype: bool
    - name: Copyright Issues
      dtype: bool
    - name: Cybercrime
      dtype: bool
    - name: Discriminatory Behavior
      dtype: bool
    - name: Disrupting Public Order
      dtype: bool
    - name: Drugs
      dtype: bool
    - name: Economic Crime
      dtype: bool
    - name: Endangering National Security
      dtype: bool
    - name: Endangering Public Health
      dtype: bool
    - name: Environmental Damage
      dtype: bool
    - name: Human Trafficking
      dtype: bool
    - name: Insulting Behavior
      dtype: bool
    - name: Mental Manipulation
      dtype: bool
    - name: Physical Harm
      dtype: bool
    - name: Privacy Violation
      dtype: bool
    - name: Psychological Harm
      dtype: bool
    - name: Sexual Content
      dtype: bool
    - name: Violence
      dtype: bool
    - name: White-Collar Crime
      dtype: bool
  - name: response_1_harm_category
    struct:
    - name: Animal Abuse
      dtype: bool
    - name: Copyright Issues
      dtype: bool
    - name: Cybercrime
      dtype: bool
    - name: Discriminatory Behavior
      dtype: bool
    - name: Disrupting Public Order
      dtype: bool
    - name: Drugs
      dtype: bool
    - name: Economic Crime
      dtype: bool
    - name: Endangering National Security
      dtype: bool
    - name: Endangering Public Health
      dtype: bool
    - name: Environmental Damage
      dtype: bool
    - name: Human Trafficking
      dtype: bool
    - name: Insulting Behavior
      dtype: bool
    - name: Mental Manipulation
      dtype: bool
    - name: Physical Harm
      dtype: bool
    - name: Privacy Violation
      dtype: bool
    - name: Psychological Harm
      dtype: bool
    - name: Sexual Content
      dtype: bool
    - name: Violence
      dtype: bool
    - name: White-Collar Crime
      dtype: bool
  - name: response_0_severity_level
    dtype: int64
  - name: response_1_severity_level
    dtype: int64
  - name: better_response_id
    dtype: int64
  - name: safer_response_id
    dtype: int64
  - name: response_0_sha256
    dtype: string
  - name: response_1_sha256
    dtype: string
  - name: response_0
    dtype: string
  - name: response_1
    dtype: string
  splits:
  - name: train
    num_bytes: 126460470
    num_examples: 75075
  download_size: 69194104
  dataset_size: 126460470
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---