metadata
dataset_info:
features:
- name: speaker_embeddings
sequence: float32
- name: input_ids
sequence: int32
- name: labels
sequence:
sequence: float32
splits:
- name: train
num_bytes: 352177204
num_examples: 3000
download_size: 350656621
dataset_size: 352177204
configs:
- config_name: default
data_files:
- split: train
path: data/train-*