metadata
dataset_info:
features:
- name: question
dtype: string
- name: generation_audio
dtype:
audio:
sampling_rate: 16000
- name: ground_truth_audio
dtype:
audio:
sampling_rate: 16000
- name: generation_tokens
dtype: string
- name: ground_truth_tokens
dtype: string
splits:
- name: train
num_bytes: 2335100
num_examples: 6
download_size: 2284911
dataset_size: 2335100
configs:
- config_name: default
data_files:
- split: train
path: data/train-*