metadata
dataset_info:
features:
- name: input_ids
list: int32
- name: labels
list: int64
- name: t
dtype: float64
- name: num_prompt_tokens
dtype: int64
- name: num_mask_tokens
dtype: int64
- name: original
list: int64
splits:
- name: train
num_bytes: 6076413508
num_examples: 2219948
- name: test
num_bytes: 61237164
num_examples: 22424
download_size: 1001500154
dataset_size: 6137650672
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*