metadata
dataset_info:
features:
- name: text
dtype: string
- name: input_ids
sequence: int32
- name: attention_mask
sequence: int8
splits:
- name: train
num_bytes: 7354858875
num_examples: 2119719
- name: validation
num_bytes: 75776630
num_examples: 21990
download_size: 1800727416
dataset_size: 7430635505
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*
this is the tokenized version of roneneldan/TinyStories tokenized using the falcon-7B tokenizer
this dataset has a total of one billion and eighty-five million two hundred and ninety-six thousand one hundred and twenty-eight tokens