metadata
dataset_info:
features:
- name: speaker_embeddings
sequence: float32
- name: input_ids
sequence: int32
- name: labels
sequence:
sequence: float32
splits:
- name: train
num_bytes: 272926432
num_examples: 1320
download_size: 271453918
dataset_size: 272926432
configs:
- config_name: default
data_files:
- split: train
path: data/train-*