metadata
dataset_info:
features:
- name: input_ids
dtype:
array2_d:
shape:
- 1
- 1024
dtype: int64
- name: attention_mask
dtype:
array3_d:
shape:
- 1
- 1024
- 1024
dtype: int64
- name: position_ids
dtype:
array2_d:
shape:
- 1
- 1024
dtype: int64
splits:
- name: train
num_bytes: 22132782784
num_examples: 2632
download_size: 7490698
dataset_size: 22132782784
configs:
- config_name: default
data_files:
- split: train
path: data/train-*