metadata
dataset_info:
features:
- name: data_source
dtype: string
- name: prompt
list:
- name: content
dtype: string
- name: role
dtype: string
- name: reward_model
struct:
- name: ground_truth
dtype: string
- name: style
dtype: string
- name: question
dtype: string
- name: solution
dtype: string
- name: answer
dtype: string
splits:
- name: train
num_bytes: 26406244
num_examples: 42398
- name: valid
num_bytes: 915137
num_examples: 1513
download_size: 14136451
dataset_size: 27321381
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: valid
path: data/valid-*