metadata
dataset_info:
features:
- name: audio
dtype:
audio:
sampling_rate: 16000
- name: label
dtype: string
- name: category
dtype: string
- name: prompt
dtype: string
- name: input_ids
sequence: int32
- name: binary_mask
sequence: int64
- name: response
dtype: string
splits:
- name: train
num_bytes: 12053937030.429
num_examples: 31241
download_size: 11726646613
dataset_size: 12053937030.429
configs:
- config_name: default
data_files:
- split: train
path: data/train-*