metadata
dataset_info:
features:
- name: translation
struct:
- name: phonemize
dtype: string
- name: text
dtype: string
- name: input_ids
sequence: int32
- name: attention_mask
sequence: int8
- name: labels
sequence: int64
splits:
- name: train
num_bytes: 6450187335
num_examples: 500000
- name: test
num_bytes: 322104329
num_examples: 25000
download_size: 1864979283
dataset_size: 6772291664
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*