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: 52372625980
num_examples: 43214441
download_size: 49071098566
dataset_size: 52372625980
configs:
- config_name: default
data_files:
- split: train
path: data/train-*