metadata
dataset_info:
features:
- name: text
dtype: string
- name: token_idx
dtype: int64
- name: hidden_states
sequence: float64
splits:
- name: train
num_bytes: 45946239835
num_examples: 2500100
download_size: 32211930570
dataset_size: 45946239835
configs:
- config_name: default
data_files:
- split: train
path:
- 0_99/train-*
- 100_199/train-*
- 200_299/train-*
- 300_399/train-*
- 400_499/train-*
- 500_599/train-*