README.md

---
dataset_info:
  features:
  - name: dataset_source
    dtype: string
  - name: prompt_id
    dtype: string
  - name: prompt
    dtype: string
  - name: response_1
    dtype: string
  - name: response_2
    dtype: string
  - name: temp_1
    dtype: float64
  - name: temp_2
    dtype: float64
  - name: model
    dtype: string
  - name: generation_timestamp
    dtype: string
  - name: scores
    struct:
    - name: model_1_response
      struct:
      - name: principle_scores
        struct:
        - name: Harmlessness
          dtype: float64
        - name: Helpfulness
          dtype: float64
        - name: Honesty
          dtype: float64
      - name: score
        dtype: float64
    - name: model_2_response
      struct:
      - name: principle_scores
        struct:
        - name: Harmlessness
          dtype: float64
        - name: Helpfulness
          dtype: float64
        - name: Honesty
          dtype: float64
      - name: score
        dtype: float64
  - name: comparison
    struct:
    - name: equal_scores
      dtype: bool
    - name: score_diff
      dtype: float64
  - name: model_1_response_score
    dtype: float64
  - name: model_2_response_score
    dtype: float64
  - name: chosen_response
    dtype: string
  splits:
  - name: train
    num_bytes: 705
    num_examples: 1
  download_size: 16161
  dataset_size: 705
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---