---
dataset_info:
  features:
  - name: prompt
    dtype: string
  - name: candidates
    list:
    - name: decoding_method
      dtype: string
    - name: model
      dtype: string
    - name: scores
      struct:
      - name: pairrm
        dtype: float64
      - name: safety
        dtype: int64
    - name: text
      dtype: string
  splits:
  - name: train
    num_bytes: 50284677.0707871
    num_examples: 57103
  - name: test
    num_bytes: 508103.9292128988
    num_examples: 577
  download_size: 23537168
  dataset_size: 50792781.0
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: test
    path: data/test-*
---