metadata
dataset_info:
features:
- name: instruction
dtype: string
- name: input
dtype: string
- name: audio_id
dtype: string
- name: dataset
dtype: string
- name: task
dtype: string
- name: output
dtype: string
- name: audio
dtype:
audio:
sampling_rate: 16000
- name: question
dtype: string
- name: answer
dtype: string
splits:
- name: train
num_bytes: 2982582637.11
num_examples: 20605
download_size: 2927927207
dataset_size: 2982582637.11
configs:
- config_name: default
data_files:
- split: train
path: data/train-*