metadata
dataset_info:
features:
- name: text
dtype: string
- name: label
dtype: string
- name: reason
dtype: string
splits:
- name: train
num_bytes: 112194229.02965
num_examples: 18251
download_size: 62910625
dataset_size: 112194229.02965
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
generated by qwen 2.5 7b instruct based on the first 20k samples in the raw deduplicated pile