---
dataset_info:
  features:
  - name: ability
    dtype: string
  - name: prompt
    list:
    - name: content
      dtype: string
    - name: role
      dtype: string
  - name: answer
    dtype: string
  - name: qwen_prompt
    dtype: string
  - name: llama_prompt
    dtype: string
  - name: deepseek_prompt
    dtype: string
  - name: reward_model
    struct:
    - name: ground_truth
      dtype: string
    - name: style
      dtype: string
  - name: extra_info
    struct:
    - name: answer
      dtype: string
    - name: deepseek_prompt
      dtype: string
    - name: index
      dtype: int64
    - name: llama_prompt
      dtype: string
    - name: question
      dtype: string
    - name: qwen_prompt
      dtype: string
    - name: split
      dtype: string
  splits:
  - name: train
    num_bytes: 213903197
    num_examples: 40000
  - name: test
    num_bytes: 26771735
    num_examples: 5000
  download_size: 57154160
  dataset_size: 240674932
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: test
    path: data/test-*
---