metadata
dataset_info:
features:
- name: id
dtype: string
- name: speaker
dtype: string
- name: input_ids
sequence: int32
- name: labels
sequence: int64
- name: attention_mask
sequence: int8
splits:
- name: train
num_bytes: 8481114
num_examples: 1000
download_size: 2664428
dataset_size: 8481114
configs:
- config_name: default
data_files:
- split: train
path: data/train-*