metadata
dataset_info:
features:
- name: prompt
dtype: string
- name: responses
sequence: string
- name: pairs
sequence:
sequence: int64
- name: sft_target
dtype: string
- name: level
sequence: int64
- name: last_q
dtype: string
splits:
- name: train
num_bytes: 25237539351
num_examples: 440800
download_size: 13421308188
dataset_size: 25237539351
configs:
- config_name: default
data_files:
- split: train
path: data/train-*