metadata
dataset_info:
features:
- name: audio_token
sequence:
sequence:
sequence: int64
- name: audio_duration
dtype: float64
- name: nemo
dtype: string
- name: cer
dtype: float64
- name: text
dtype: string
splits:
- name: train
num_bytes: 23162502823
num_examples: 8282987
download_size: 5158446109
dataset_size: 23162502823
configs:
- config_name: default
data_files:
- split: train
path: data/train-*