metadata
dataset_info:
features:
- name: output
dtype: string
- name: input
dtype: string
- name: instruction
dtype: string
- name: input_ids
sequence: int32
- name: attention_mask
sequence: int8
splits:
- name: train
num_bytes: 1089133996.6720784
num_examples: 639794
- name: validation
num_bytes: 272284350.32792157
num_examples: 159949
download_size: 625233880
dataset_size: 1361418347
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*