metadata
dataset_info:
features:
- name: source
dtype: string
- name: input_ids
list: uint16
- name: output_ids
list: uint16
- name: num_input_tokens
dtype: int64
- name: num_output_tokens
dtype: int64
splits:
- name: train
num_bytes: 4666136344.447924
num_examples: 3752140
download_size: 3404769055
dataset_size: 4666136344.447924
configs:
- config_name: default
data_files:
- split: train
path: data/train-*