---
dataset_info:
  features:
  - name: Prompt
    dtype: string
  - name: Cot_Response
    dtype: string
  - name: Response
    dtype: string
  - name: raw_token
    dtype: string
  - name: conversations
    list:
    - name: content
      dtype: string
    - name: role
      dtype: string
  splits:
  - name: train
    num_bytes: 9417912884.43974
    num_examples: 565151
  - name: test
    num_bytes: 499932560.5602612
    num_examples: 30000
  download_size: 1040418407
  dataset_size: 9917845445.0
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: test
    path: data/test-*
---
arxiv.org/abs/2502.14669