---
dataset_info:
  features:
  - name: dataset
    dtype: string
  - name: prompt
    dtype: string
  - name: initial_reason_steps
    dtype: string
  - name: chosen
    dtype: string
  - name: rejected
    dtype: string
  - name: full_chosen
    dtype: string
  - name: full_rejected
    dtype: string
  - name: answer
    dtype: string
  - name: chosen_steps_rejected_answer
    dtype: string
  - name: rejected_steps_chosen_answer
    dtype: string
  - name: chosen_steps_substitute_answer
    dtype: string
  - name: rejected_steps_substitute_answer
    dtype: string
  - name: score_initial_reason_steps_only
    dtype: float64
  - name: score_chosen
    dtype: float64
  - name: score_rejected
    dtype: float64
  - name: score_full_chosen
    dtype: float64
  - name: score_full_rejected
    dtype: float64
  - name: score_rejected_steps_chosen_answer
    dtype: float64
  - name: score_chosen_steps_rejected_answer
    dtype: float64
  - name: score_chosen_steps_substitute_answer
    dtype: float64
  - name: score_rejected_steps_substitute_answer
    dtype: float64
  splits:
  - name: train
    num_bytes: 57549537
    num_examples: 10795
  download_size: 25307572
  dataset_size: 57549537
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---