metadata
dataset_info:
features:
- name: user_id
dtype: uint32
- name: impression_time_fixed
sequence: timestamp[us]
- name: scroll_percentage_fixed
sequence: float32
- name: article_id_fixed
sequence: int32
- name: read_time_fixed
sequence: float32
splits:
- name: train
num_bytes: 2367276468
num_examples: 807677
download_size: 1426808983
dataset_size: 2367276468
configs:
- config_name: default
data_files:
- split: train
path: data/train-*