metadata
dataset_info:
- config_name: tokenized_llama31
features:
- name: input_ids
sequence: int32
- name: labels
sequence: int64
- name: attention_mask
sequence: int8
splits:
- name: train
num_examples: 19089792
configs:
- config_name: tokenized_llama31
data_files:
- split: train
path: tokenized_llama31/train-*