metadata
dataset_info:
features:
- name: messages
list:
- name: role
dtype: string
- name: content
dtype: string
- name: info
struct:
- name: source
dtype: string
- name: reference_answer
dtype: string
- name: test_case
dtype: string
- name: think_content
dtype: string
- name: answer_content
dtype: string
splits:
- name: train
num_bytes: 14042845756
num_examples: 500000
download_size: 6188819724
dataset_size: 14042845756
configs:
- config_name: default
data_files:
- split: train
path: data/train-*